You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by tq...@apache.org on 2023/03/24 04:14:01 UTC
[tvm-site] branch asf-site updated: deploying docs (apache/tvm@b98755637542b52ad8177955d76e471a190540f1)
This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a commit to branch asf-site
in repository https://gitbox.apache.org/repos/asf/tvm-site.git
The following commit(s) were added to refs/heads/asf-site by this push:
new 555b69fec8 deploying docs (apache/tvm@b98755637542b52ad8177955d76e471a190540f1)
555b69fec8 is described below
commit 555b69fec849a8c6f8e44f241ceecb9365ca5a22
Author: tvm-bot <95...@users.noreply.github.com>
AuthorDate: Fri Mar 24 04:13:53 2023 +0000
deploying docs (apache/tvm@b98755637542b52ad8177955d76e471a190540f1)
---
docs/_images/sphx_glr_micro_train_001.png | Bin 333957 -> 349354 bytes
docs/_images/sphx_glr_micro_train_thumb.png | Bin 23827 -> 24292 bytes
.../how_to/compile_models/from_darknet.rst.txt | 2 +-
.../how_to/compile_models/from_mxnet.rst.txt | 2 +-
.../how_to/compile_models/from_oneflow.rst.txt | 2 +-
.../how_to/compile_models/from_pytorch.rst.txt | 2 +-
.../how_to/compile_models/from_tensorflow.rst.txt | 2 +-
.../compile_models/sg_execution_times.rst.txt | 22 +-
.../deploy_models/deploy_model_on_adreno.rst.txt | 2 +-
.../deploy_model_on_adreno_tvmc.rst.txt | 2 +-
.../deploy_models/deploy_model_on_android.rst.txt | 2 +-
.../deploy_object_detection_pytorch.rst.txt | 4 +-
.../deploy_models/deploy_prequantized.rst.txt | 6 +-
.../deploy_prequantized_tflite.rst.txt | 4 +-
.../how_to/deploy_models/deploy_quantized.rst.txt | 2 +-
.../deploy_models/deploy_ssd_gluoncv.rst.txt | 4 +-
.../deploy_models/sg_execution_times.rst.txt | 22 +-
.../extend_tvm/bring_your_own_datatypes.rst.txt | 2 +-
.../how_to/extend_tvm/sg_execution_times.rst.txt | 10 +-
.../how_to/extend_tvm/use_pass_instrument.rst.txt | 16 +-
.../optimize_operators/opt_conv_cuda.rst.txt | 2 +-
.../optimize_operators/opt_conv_tensorcore.rst.txt | 2 +-
.../how_to/optimize_operators/opt_gemm.rst.txt | 16 +-
.../optimize_operators/sg_execution_times.rst.txt | 8 +-
.../sg_execution_times.rst.txt | 14 +-
.../tune_conv2d_layer_cuda.rst.txt | 1181 ++-
.../tune_network_cuda.rst.txt | 4 +-
.../tune_network_x86.rst.txt | 4 +-
.../tune_sparse_x86.rst.txt | 90 +-
.../tune_with_autotvm/sg_execution_times.rst.txt | 10 +-
.../tune_with_autotvm/tune_conv2d_cuda.rst.txt | 436 +-
.../work_with_microtvm/micro_autotune.rst.txt | 18 +-
.../work_with_microtvm/micro_pytorch.rst.txt | 4 +-
.../how_to/work_with_microtvm/micro_train.rst.txt | 18 +-
.../work_with_microtvm/sg_execution_times.rst.txt | 14 +-
.../work_with_relay/sg_execution_times.rst.txt | 10 +-
.../how_to/work_with_schedules/intrin_math.rst.txt | 2 +-
.../work_with_schedules/sg_execution_times.rst.txt | 18 +-
.../tutorials/autotvm/sg_execution_times.rst.txt | 6 +-
.../frontend/deploy_classification.rst.txt | 2 +-
.../tutorials/frontend/deploy_detection.rst.txt | 2 +-
.../tutorials/frontend/sg_execution_times.rst.txt | 6 +-
.../tutorials/optimize/sg_execution_times.rst.txt | 6 +-
.../topic/vta/tutorials/sg_execution_times.rst.txt | 6 +-
.../tutorial/auto_scheduler_matmul_x86.rst.txt | 4 +-
docs/_sources/tutorial/autotvm_matmul_x86.rst.txt | 20 +-
docs/_sources/tutorial/autotvm_relay_x86.rst.txt | 58 +-
.../tutorial/cross_compilation_and_rpc.rst.txt | 2 +-
docs/_sources/tutorial/intro_topi.rst.txt | 2 +-
docs/_sources/tutorial/sg_execution_times.rst.txt | 18 +-
.../tutorial/tensor_expr_get_started.rst.txt | 44 +-
docs/commit_hash | 2 +-
docs/genindex.html | 8 +-
docs/how_to/compile_models/from_darknet.html | 2 +-
docs/how_to/compile_models/from_mxnet.html | 2 +-
docs/how_to/compile_models/from_oneflow.html | 15 +-
docs/how_to/compile_models/from_pytorch.html | 12 +-
docs/how_to/compile_models/from_tensorflow.html | 2 +-
docs/how_to/compile_models/sg_execution_times.html | 22 +-
.../deploy_models/deploy_model_on_adreno.html | 2 +-
.../deploy_models/deploy_model_on_adreno_tvmc.html | 25 +-
.../deploy_models/deploy_model_on_android.html | 2 +-
.../deploy_object_detection_pytorch.html | 54 +-
docs/how_to/deploy_models/deploy_prequantized.html | 9 +-
.../deploy_models/deploy_prequantized_tflite.html | 4 +-
docs/how_to/deploy_models/deploy_quantized.html | 2 +-
docs/how_to/deploy_models/deploy_ssd_gluoncv.html | 39 +-
docs/how_to/deploy_models/sg_execution_times.html | 22 +-
.../extend_tvm/bring_your_own_datatypes.html | 2 +-
docs/how_to/extend_tvm/sg_execution_times.html | 10 +-
docs/how_to/extend_tvm/use_pass_instrument.html | 16 +-
docs/how_to/optimize_operators/opt_conv_cuda.html | 2 +-
.../optimize_operators/opt_conv_tensorcore.html | 2 +-
docs/how_to/optimize_operators/opt_gemm.html | 16 +-
.../optimize_operators/sg_execution_times.html | 8 +-
.../sg_execution_times.html | 14 +-
.../tune_conv2d_layer_cuda.html | 1181 ++-
.../tune_with_autoscheduler/tune_network_cuda.html | 4 +-
.../tune_with_autoscheduler/tune_network_x86.html | 4 +-
.../tune_with_autoscheduler/tune_sparse_x86.html | 90 +-
.../tune_with_autotvm/sg_execution_times.html | 10 +-
.../how_to/tune_with_autotvm/tune_conv2d_cuda.html | 436 +-
docs/how_to/work_with_microtvm/micro_autotune.html | 18 +-
docs/how_to/work_with_microtvm/micro_pytorch.html | 6 +-
docs/how_to/work_with_microtvm/micro_train.html | 16 +-
.../work_with_microtvm/sg_execution_times.html | 14 +-
.../how_to/work_with_relay/sg_execution_times.html | 10 +-
docs/how_to/work_with_schedules/intrin_math.html | 2 +-
.../work_with_schedules/sg_execution_times.html | 18 +-
docs/objects.inv | Bin 24746 -> 24732 bytes
docs/reference/api/doxygen/annotated.html | 148 +-
docs/reference/api/doxygen/builtin_8h.html | 4 +-
.../api/doxygen/c__backend__api_8h_source.html | 4 +-
.../api/doxygen/c__runtime__api_8h_source.html | 4 +-
docs/reference/api/doxygen/classes.html | 480 +-
.../classtvm_1_1BaseExprNode__coll__graph.svg | 40 +-
.../classtvm_1_1BaseExprNode__inherit__graph.svg | 826 +-
.../doxygen/classtvm_1_1BaseExpr__coll__graph.svg | 20 +-
.../classtvm_1_1BaseExpr__inherit__graph.svg | 690 +-
.../api/doxygen/classtvm_1_1PrimExpr.html | 2 +-
.../api/doxygen/classtvm_1_1PrimExprNode.html | 2 +-
.../classtvm_1_1PrimExprNode__coll__graph.svg | 56 +-
.../classtvm_1_1PrimExprNode__inherit__graph.svg | 672 +-
.../doxygen/classtvm_1_1PrimExpr__coll__graph.svg | 28 +-
.../classtvm_1_1PrimExpr__inherit__graph.svg | 447 +-
...asstvm_1_1runtime_1_1ObjectRef__coll__graph.svg | 12 +-
.../classtvm_1_1runtime_1_1Object__coll__graph.svg | 8 +-
.../doxygen/classtvm_1_1tir_1_1BufferLoadNode.html | 4 +-
.../classtvm_1_1tir_1_1BufferStoreNode.html | 4 +-
...asstvm_1_1tir_1_1DataTypeLegalizer-members.html | 52 +-
.../classtvm_1_1tir_1_1DataTypeLegalizer.html | 4 -
...vm_1_1tir_1_1DataTypeLegalizer__coll__graph.svg | 4 +-
...1_1tir_1_1DataTypeLegalizer__inherit__graph.svg | 4 +-
...rimExpr_01_6n_00_01Args_8_8_8_08_4-members.html | 65 +-
...onst_01PrimExpr_01_6n_00_01Args_8_8_8_08_4.html | 108 +-
...xpr_01_6n_00_01Args_8_8_8_08_4__coll__graph.svg | 2 +-
.../classtvm_1_1tir_1_1ExprMutator-members.html | 65 +-
.../doxygen/classtvm_1_1tir_1_1ExprMutator.html | 96 +-
...classtvm_1_1tir_1_1ExprMutator__coll__graph.svg | 2 +-
...sstvm_1_1tir_1_1ExprMutator__inherit__graph.svg | 2 +-
.../classtvm_1_1tir_1_1ExprVisitor-members.html | 65 +-
.../doxygen/classtvm_1_1tir_1_1ExprVisitor.html | 96 +-
...classtvm_1_1tir_1_1ExprVisitor__coll__graph.svg | 2 +-
...sstvm_1_1tir_1_1ExprVisitor__inherit__graph.svg | 2 +-
..._1_1tir_1_1IndexDataTypeNormalizer-members.html | 54 +-
...classtvm_1_1tir_1_1IndexDataTypeNormalizer.html | 4 -
...tir_1_1IndexDataTypeNormalizer__coll__graph.svg | 4 +-
..._1_1IndexDataTypeNormalizer__inherit__graph.svg | 4 +-
...vm_1_1tir_1_1IndexDataTypeRewriter-members.html | 56 +-
.../classtvm_1_1tir_1_1IndexDataTypeRewriter.html | 4 -
..._1tir_1_1IndexDataTypeRewriter__coll__graph.svg | 4 +-
...ir_1_1IndexDataTypeRewriter__inherit__graph.svg | 4 +-
.../doxygen/classtvm_1_1tir_1_1Load-members.html | 107 -
.../api/doxygen/classtvm_1_1tir_1_1Load.html | 289 -
.../classtvm_1_1tir_1_1LoadNode-members.html | 123 -
.../api/doxygen/classtvm_1_1tir_1_1LoadNode.html | 436 -
.../classtvm_1_1tir_1_1LoadNode__coll__graph.svg | 337 -
...classtvm_1_1tir_1_1LoadNode__inherit__graph.svg | 127 -
.../classtvm_1_1tir_1_1Load__coll__graph.svg | 135 -
.../classtvm_1_1tir_1_1Load__inherit__graph.svg | 105 -
.../classtvm_1_1tir_1_1ProducerLoadNode.html | 4 +-
.../classtvm_1_1tir_1_1ProducerStoreNode.html | 4 +-
.../api/doxygen/classtvm_1_1tir_1_1Stmt.html | 2 +-
...classtvm_1_1tir_1_1StmtExprMutator-members.html | 86 +-
.../classtvm_1_1tir_1_1StmtExprMutator.html | 4 -
...stvm_1_1tir_1_1StmtExprMutator__coll__graph.svg | 4 +-
...m_1_1tir_1_1StmtExprMutator__inherit__graph.svg | 4 +-
...classtvm_1_1tir_1_1StmtExprVisitor-members.html | 82 +-
.../classtvm_1_1tir_1_1StmtExprVisitor.html | 4 -
...stvm_1_1tir_1_1StmtExprVisitor__coll__graph.svg | 4 +-
...m_1_1tir_1_1StmtExprVisitor__inherit__graph.svg | 4 +-
..._01_6n_00_01Args_8_8_8_01args_08_4-members.html | 25 +-
...t_01Stmt_01_6n_00_01Args_8_8_8_01args_08_4.html | 76 +-
...6n_00_01Args_8_8_8_01args_08_4__coll__graph.svg | 2 +-
.../classtvm_1_1tir_1_1StmtMutator-members.html | 21 +-
.../doxygen/classtvm_1_1tir_1_1StmtMutator.html | 64 +-
...classtvm_1_1tir_1_1StmtMutator__coll__graph.svg | 2 +-
...sstvm_1_1tir_1_1StmtMutator__inherit__graph.svg | 2 +-
.../api/doxygen/classtvm_1_1tir_1_1StmtNode.html | 2 +-
.../classtvm_1_1tir_1_1StmtNode__coll__graph.svg | 40 +-
...classtvm_1_1tir_1_1StmtNode__inherit__graph.svg | 238 +-
.../classtvm_1_1tir_1_1StmtVisitor-members.html | 21 +-
.../doxygen/classtvm_1_1tir_1_1StmtVisitor.html | 64 +-
...classtvm_1_1tir_1_1StmtVisitor__coll__graph.svg | 2 +-
...sstvm_1_1tir_1_1StmtVisitor__inherit__graph.svg | 2 +-
.../classtvm_1_1tir_1_1Stmt__coll__graph.svg | 20 +-
.../classtvm_1_1tir_1_1Stmt__inherit__graph.svg | 169 +-
.../doxygen/classtvm_1_1tir_1_1Store-members.html | 103 -
.../api/doxygen/classtvm_1_1tir_1_1Store.html | 278 -
.../classtvm_1_1tir_1_1StoreNode-members.html | 124 -
.../api/doxygen/classtvm_1_1tir_1_1StoreNode.html | 448 --
.../classtvm_1_1tir_1_1StoreNode__coll__graph.svg | 282 -
...lasstvm_1_1tir_1_1StoreNode__inherit__graph.svg | 109 -
.../classtvm_1_1tir_1_1Store__coll__graph.svg | 112 -
.../classtvm_1_1tir_1_1Store__inherit__graph.svg | 82 -
.../api/doxygen/compilation__config_8h_source.html | 2 +-
.../doxygen/data__type__rewriter_8h_source.html | 22 +-
.../api/doxygen/detail_2extern_8h_source.html | 2 +-
.../api/doxygen/device__api_8h_source.html | 2 +-
docs/reference/api/doxygen/functions__.html | 2 -
docs/reference/api/doxygen/functions_b.html | 4 +-
docs/reference/api/doxygen/functions_func_l.html | 5 +-
docs/reference/api/doxygen/functions_func_s.html | 23 +-
docs/reference/api/doxygen/functions_func_t.html | 14 +-
docs/reference/api/doxygen/functions_func_u.html | 2 +-
docs/reference/api/doxygen/functions_func_v.html | 44 +-
docs/reference/api/doxygen/functions_i.html | 6 +-
docs/reference/api/doxygen/functions_l.html | 3 +-
docs/reference/api/doxygen/functions_p.html | 2 -
docs/reference/api/doxygen/functions_s.html | 23 +-
docs/reference/api/doxygen/functions_t.html | 16 +-
docs/reference/api/doxygen/functions_v.html | 47 +-
docs/reference/api/doxygen/functions_vars.html | 2 -
docs/reference/api/doxygen/functions_vars_b.html | 2 -
docs/reference/api/doxygen/functions_vars_i.html | 2 -
docs/reference/api/doxygen/functions_vars_p.html | 2 -
docs/reference/api/doxygen/functions_vars_v.html | 1 -
docs/reference/api/doxygen/hierarchy.html | 84 +-
docs/reference/api/doxygen/inherit_graph_112.svg | 8290 ++++++++++----------
docs/reference/api/doxygen/inherit_graph_12.svg | 16 +-
docs/reference/api/doxygen/inherit_graph_121.svg | 6126 +++++++--------
docs/reference/api/doxygen/inherit_graph_209.svg | 16 +-
docs/reference/api/doxygen/inherit_graph_210.svg | 16 +-
docs/reference/api/doxygen/inherit_graph_41.svg | 16 +-
docs/reference/api/doxygen/inherit_graph_45.svg | 8 +-
docs/reference/api/doxygen/inherit_graph_95.svg | 8 +-
docs/reference/api/doxygen/inherits.html | 4 +-
.../reference/api/doxygen/namespacetvm_1_1tir.html | 20 +-
.../doxygen/namespacetvm_1_1tir_1_1builtin.html | 8 +-
.../reference/api/doxygen/profiling_8h_source.html | 4 +-
.../reference/api/doxygen/reduction_8h_source.html | 4 +-
.../api/doxygen/relay_2type_8h_source.html | 4 +-
docs/reference/api/doxygen/search/all_1.js | 2 +-
docs/reference/api/doxygen/search/all_11.js | 6 +-
docs/reference/api/doxygen/search/all_13.js | 2 +-
docs/reference/api/doxygen/search/all_14.js | 20 +-
docs/reference/api/doxygen/search/all_15.js | 24 +-
docs/reference/api/doxygen/search/all_16.js | 4 +-
docs/reference/api/doxygen/search/all_17.js | 12 +-
docs/reference/api/doxygen/search/all_3.js | 2 +-
docs/reference/api/doxygen/search/all_6.js | 2 +-
docs/reference/api/doxygen/search/all_9.js | 2 +-
docs/reference/api/doxygen/search/all_a.js | 6 +-
docs/reference/api/doxygen/search/all_d.js | 3 +-
docs/reference/api/doxygen/search/all_e.js | 2 +-
docs/reference/api/doxygen/search/classes_10.js | 6 +-
docs/reference/api/doxygen/search/classes_11.js | 8 +-
docs/reference/api/doxygen/search/classes_13.js | 2 +-
docs/reference/api/doxygen/search/classes_4.js | 2 +-
docs/reference/api/doxygen/search/classes_7.js | 2 +-
docs/reference/api/doxygen/search/classes_8.js | 2 +-
docs/reference/api/doxygen/search/classes_9.js | 2 -
docs/reference/api/doxygen/search/functions_13.js | 7 +-
docs/reference/api/doxygen/search/functions_14.js | 10 +-
docs/reference/api/doxygen/search/functions_15.js | 4 +-
docs/reference/api/doxygen/search/functions_16.js | 6 +-
docs/reference/api/doxygen/search/functions_c.js | 2 +-
docs/reference/api/doxygen/search/functions_d.js | 2 +-
docs/reference/api/doxygen/search/variables_0.js | 2 +-
docs/reference/api/doxygen/search/variables_14.js | 2 +-
docs/reference/api/doxygen/search/variables_2.js | 2 +-
docs/reference/api/doxygen/search/variables_9.js | 2 +-
docs/reference/api/doxygen/search/variables_f.js | 2 +-
.../api/doxygen/serializer_8h_source.html | 2 +-
docs/reference/api/doxygen/stmt_8h.html | 10 +-
docs/reference/api/doxygen/stmt_8h_source.html | 485 +-
.../api/doxygen/stmt__functor_8h_source.html | 82 +-
.../api/doxygen/strided__slice_8h_source.html | 2 +-
.../api/doxygen/target__kind_8h_source.html | 2 +-
.../api/doxygen/te_2schedule_8h_source.html | 2 +-
.../api/doxygen/tir_2analysis_8h_source.html | 8 +-
docs/reference/api/doxygen/tir_2expr_8h.html | 10 +-
.../reference/api/doxygen/tir_2expr_8h_source.html | 142 +-
.../api/doxygen/tir_2expr__functor_8h_source.html | 82 +-
.../api/doxygen/tir_2frame_8h_source.html | 2 +-
docs/reference/api/doxygen/tir_2ir_8h_source.html | 4 +-
docs/reference/api/doxygen/tir_2op_8h_source.html | 8 +-
.../doxygen/tir_2schedule_2schedule_8h_source.html | 6 +-
.../api/doxygen/tir_2usmp_2utils_8h_source.html | 4 +-
docs/reference/api/python/auto_scheduler.html | 4 +-
docs/reference/api/python/tir.html | 79 +-
.../api/typedoc/classes/bytestreamreader.html | 12 +-
.../api/typedoc/classes/cachedcallstack.html | 34 +-
docs/reference/api/typedoc/classes/dldatatype.html | 12 +-
docs/reference/api/typedoc/classes/dldevice.html | 10 +-
.../reference/api/typedoc/classes/environment.html | 12 +-
docs/reference/api/typedoc/classes/ffilibrary.html | 20 +-
docs/reference/api/typedoc/classes/instance.html | 58 +-
docs/reference/api/typedoc/classes/memory.html | 34 +-
docs/reference/api/typedoc/classes/module.html | 10 +-
docs/reference/api/typedoc/classes/ndarray.html | 22 +-
.../api/typedoc/classes/packedfunccell.html | 6 +-
docs/reference/api/typedoc/classes/rpcserver.html | 14 +-
.../api/typedoc/classes/runtimecontext.html | 22 +-
docs/reference/api/typedoc/classes/scalar.html | 6 +-
docs/reference/api/typedoc/classes/tvmarray.html | 16 +-
docs/reference/api/typedoc/classes/tvmobject.html | 12 +-
.../api/typedoc/classes/webgpucontext.html | 12 +-
docs/reference/api/typedoc/enums/argtypecode.html | 30 +-
.../api/typedoc/enums/aynccallbackcode.html | 4 +-
.../api/typedoc/enums/dldatatypecode.html | 8 +-
.../api/typedoc/enums/rpcserverstate.html | 12 +-
docs/reference/api/typedoc/enums/sizeof.html | 18 +-
docs/reference/api/typedoc/index.html | 124 +-
.../api/typedoc/interfaces/disposable.html | 2 +-
.../api/typedoc/interfaces/functioninfo.html | 6 +-
.../api/typedoc/interfaces/libraryprovider.html | 4 +-
docs/searchindex.js | 2 +-
.../vta/tutorials/autotvm/sg_execution_times.html | 6 +-
.../tutorials/frontend/deploy_classification.html | 2 +-
.../vta/tutorials/frontend/deploy_detection.html | 2 +-
.../vta/tutorials/frontend/sg_execution_times.html | 6 +-
.../vta/tutorials/optimize/sg_execution_times.html | 6 +-
docs/topic/vta/tutorials/sg_execution_times.html | 6 +-
docs/tutorial/auto_scheduler_matmul_x86.html | 4 +-
docs/tutorial/autotvm_matmul_x86.html | 20 +-
docs/tutorial/autotvm_relay_x86.html | 274 +-
docs/tutorial/cross_compilation_and_rpc.html | 2 +-
docs/tutorial/intro_topi.html | 2 +-
docs/tutorial/sg_execution_times.html | 18 +-
docs/tutorial/tensor_expr_get_started.html | 44 +-
301 files changed, 13700 insertions(+), 15693 deletions(-)
diff --git a/docs/_images/sphx_glr_micro_train_001.png b/docs/_images/sphx_glr_micro_train_001.png
index b0c19c3436..b147d290fc 100644
Binary files a/docs/_images/sphx_glr_micro_train_001.png and b/docs/_images/sphx_glr_micro_train_001.png differ
diff --git a/docs/_images/sphx_glr_micro_train_thumb.png b/docs/_images/sphx_glr_micro_train_thumb.png
index 475ea5ff8a..a474f132c9 100644
Binary files a/docs/_images/sphx_glr_micro_train_thumb.png and b/docs/_images/sphx_glr_micro_train_thumb.png differ
diff --git a/docs/_sources/how_to/compile_models/from_darknet.rst.txt b/docs/_sources/how_to/compile_models/from_darknet.rst.txt
index 132f17aac2..7e359a83e2 100644
--- a/docs/_sources/how_to/compile_models/from_darknet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_darknet.rst.txt
@@ -318,7 +318,7 @@ The process is no different from other examples.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 23.554 seconds)
+ **Total running time of the script:** ( 1 minutes 20.288 seconds)
.. _sphx_glr_download_how_to_compile_models_from_darknet.py:
diff --git a/docs/_sources/how_to/compile_models/from_mxnet.rst.txt b/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
index 1421bba976..652ecd2957 100644
--- a/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_mxnet.rst.txt
@@ -116,7 +116,7 @@ In this section, we download a pretrained imagenet model and classify an image.
.. code-block:: none
- Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zip0345609a-aa53-4983-888b-c0a10eb99acb from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+ Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipb72f8b5f-aacb-4c5e-a189-0ae1bb48fe1d from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
x (1, 3, 224, 224)
diff --git a/docs/_sources/how_to/compile_models/from_oneflow.rst.txt b/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
index a90903e7fd..4ce2e2f46f 100644
--- a/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_oneflow.rst.txt
@@ -121,7 +121,7 @@ Load a pretrained OneFlow model and save model
.. code-block:: none
Downloading: "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/ResNet/resnet18.zip" to /workspace/.oneflow/flowvision_cache/resnet18.zip
-
0%| | 0.00/41.5M [00:00<?, ?B/s]
15%|#5 | 6.33M/41.5M [00:00<00:00, 66.1MB/s]
30%|### | 12.6M/41.5M [00:00<00:00, 51.6MB/s]
43%|####2 | 17.8M/41.5M [00:00<00:00, 33.7MB/s]
58%|#####7 | 24.0M/41.5M [00:00<00:00, 35.9MB/s]
77%|#######7 | 32.0M/41.5M [00:00<00:00, 46.0MB/s]
92%|#########2| 38.4M/41.5M [00:00<00:00, 51.2MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 46.5MB/s]
+
0%| | 0.00/41.5M [00:00<?, ?B/s]
19%|#9 | 7.99M/41.5M [00:00<00:00, 55.3MB/s]
35%|###4 | 14.3M/41.5M [00:00<00:00, 47.3MB/s]
45%|####5 | 18.9M/41.5M [00:00<00:00, 44.5MB/s]
57%|#####6 | 23.5M/41.5M [00:00<00:00, 46.0MB/s]
67%|######7 | 27.9M/41.5M [00:00<00:00, 43.3MB/s]
77%|#######7 | 32.1M/41.5M [00:00<00:00, 43.5MB/s]
96%|#########6| 40.0M/41.5M [00:00<00:00, 48.0MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 47.8MB/s]
diff --git a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
index 61201223eb..ffff4e5d94 100644
--- a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
@@ -101,7 +101,7 @@ Load a pretrained PyTorch model
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.
warnings.warn(msg)
Downloading: "https://download.pytorch.org/models/resnet18-f37072fd.pth" to /workspace/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth
-
0%| | 0.00/44.7M [00:00<?, ?B/s]
18%|#7 | 7.99M/44.7M [00:00<00:00, 76.6MB/s]
40%|#### | 18.1M/44.7M [00:00<00:00, 86.7MB/s]
59%|#####8 | 26.3M/44.7M [00:00<00:00, 79.6MB/s]
76%|#######5 | 33.9M/44.7M [00:00<00:00, 78.7MB/s]
96%|#########5| 42.8M/44.7M [00:00<00:00, 83.6MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 74.4MB/s]
+
0%| | 0.00/44.7M [00:00<?, ?B/s]
18%|#7 | 7.99M/44.7M [00:00<00:00, 67.9MB/s]
32%|###2 | 14.5M/44.7M [00:00<00:00, 45.2MB/s]
46%|####6 | 20.7M/44.7M [00:00<00:00, 51.9MB/s]
72%|#######1 | 32.0M/44.7M [00:00<00:00, 60.6MB/s]
99%|#########8| 44.2M/44.7M [00:00<00:00, 68.0MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 62.8MB/s]
diff --git a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
index 1c611705a0..1910734089 100644
--- a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
@@ -430,7 +430,7 @@ Run the corresponding model on tensorflow
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 35.258 seconds)
+ **Total running time of the script:** ( 1 minutes 29.717 seconds)
.. _sphx_glr_download_how_to_compile_models_from_tensorflow.py:
diff --git a/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt b/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt
index 1c5d283a4b..6f898f9b69 100644
--- a/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/compile_models/sg_execution_times.rst.txt
@@ -5,26 +5,26 @@
Computation times
=================
-**07:04.515** total execution time for **how_to_compile_models** files:
+**06:47.377** total execution time for **how_to_compile_models** files:
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:35.258 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:29.717 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:23.554 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:20.288 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:59.622 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:56.265 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:39.309 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:37.813 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:33.471 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:32.203 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:31.014 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:30.112 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:28.503 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:28.207 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:27.353 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:26.670 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:23.635 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:23.425 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.796 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.677 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/deploy_models/deploy_model_on_adreno.rst.txt b/docs/_sources/how_to/deploy_models/deploy_model_on_adreno.rst.txt
index cef636381e..f71e2ca7a9 100644
--- a/docs/_sources/how_to/deploy_models/deploy_model_on_adreno.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_model_on_adreno.rst.txt
@@ -637,7 +637,7 @@ well as provides information about the model's performance
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 2750.3544 2750.1791 2754.7262 2747.9918 1.7600
+ 2537.8528 2536.9948 2544.7511 2535.1032 2.6745
diff --git a/docs/_sources/how_to/deploy_models/deploy_model_on_adreno_tvmc.rst.txt b/docs/_sources/how_to/deploy_models/deploy_model_on_adreno_tvmc.rst.txt
index 85f3fe88a7..7ab89809d1 100644
--- a/docs/_sources/how_to/deploy_models/deploy_model_on_adreno_tvmc.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_model_on_adreno_tvmc.rst.txt
@@ -127,7 +127,7 @@ Make a Keras Resnet50 Model
.. code-block:: none
Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/resnet/resnet50_weights_tf_dim_ordering_tf_kernels.h5
-
8192/102967424 [..............................] - ETA: 0s
4775936/102967424 [>.............................] - ETA: 1s
8380416/102967424 [=>............................] - ETA: 1s
10575872/102967424 [==>...........................] - ETA: 1s
15024128/102967424 [===>..........................] - ETA: 1s
16769024/102967424 [===>..........................] - ETA: 1s
25157632/102967424 [======>.......................] - ETA: 1s
33546240/102967424 [========>.....................] - ETA: 1s
40189952/102967424 [==========>...................] - ETA: 0s
41934848/102967424 [===========>..................] - ETA: 1s
47046656/102967424 [============>.................] - ETA: 0s
50323456/102967424 [=============>................] - ETA: 0s
56967168/102967424 [===============>..............] - ETA: 0s
58712064/102967424 [================>.............] - ETA: 0s
65323008/102967424 [==================>...........] - ETA: 0s
67305472/102967424 [==================>...........] -
ETA: 0s
69296128/102967424 [===================>..........] - ETA: 0s
75489280/102967424 [====================>.........] - ETA: 0s
82878464/102967424 [=======================>......] - ETA: 0s
83877888/102967424 [=======================>......] - ETA: 0s
90521600/102967424 [=========================>....] - ETA: 0s
92266496/102967424 [=========================>....] - ETA: 0s
98910208/102967424 [===========================>..] - ETA: 0s
102850560/102967424
[============================>.] - ETA: 0s
102967424/102967424 [==============================] - 2s 0us/step
+
8192/102967424 [..............................] - ETA: 0s
8380416/102967424 [=>............................] - ETA: 1s
15024128/102967424 [===>..........................] - ETA: 0s
16769024/102967424 [===>..........................] - ETA: 1s
25157632/102967424 [======>.......................] - ETA: 0s
30605312/102967424 [=======>......................] - ETA: 0s
33546240/102967424 [========>.....................] - ETA: 0s
41934848/102967424 [===========>..................] - ETA: 0s
48578560/102967424 [=============>................] - ETA: 0s
50323456/102967424 [=============>................] - ETA: 0s
58318848/102967424 [===============>..............] - ETA: 0s
58712064/102967424 [================>.............] - ETA: 0s
65355776/102967424 [==================>...........] - ETA: 0s
67100672/102967424 [==================>...........] - ETA: 0s
69296128/102967424 [===================>..........] - ETA: 0s
75489280/102967424 [====================>.........] -
ETA: 0s
81362944/102967424 [======================>.......] - ETA: 0s
83877888/102967424 [=======================>......] - ETA: 0s
91979776/102967424 [=========================>....] - ETA: 0s
92266496/102967424 [=========================>....] - ETA: 0s
98910208/102967424 [===========================>..] - ETA: 0s
102449152/102967424 [============================>.] - ETA: 0s
102850560/102967424 [============================>.] - ETA: 0s
102967424/102967424
[==============================] - 2s 0us/step
diff --git a/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt b/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
index 128928981f..d90127982c 100644
--- a/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_model_on_android.rst.txt
@@ -437,7 +437,7 @@ Execute on TVM
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 16.1671 16.1450 16.5010 15.9790 0.1270
+ 16.2733 16.4747 16.7356 15.5997 0.4025
diff --git a/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt b/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt
index d59b47dc87..656c7ff54d 100644
--- a/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_object_detection_pytorch.rst.txt
@@ -130,7 +130,7 @@ Load pre-trained maskrcnn from torchvision and do tracing
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=MaskRCNN_ResNet50_FPN_Weights.COCO_V1`. You can also use `weights=MaskRCNN_ResNet50_FPN_Weights.DEFAULT` to get the most up-to-date weights.
warnings.warn(msg)
Downloading: "https://download.pytorch.org/models/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth" to /workspace/.cache/torch/hub/checkpoints/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth
-
0%| | 0.00/170M [00:00<?, ?B/s]
4%|3 | 6.30M/170M [00:00<00:02, 61.9MB/s]
7%|7 | 12.2M/170M [00:00<00:02, 57.3MB/s]
12%|#1 | 19.8M/170M [00:00<00:02, 66.9MB/s]
15%|#5 | 26.2M/170M [00:00<00:02, 53.9MB/s]
19%|#8 | 32.0M/170M [00:00<00:02, 50.7MB/s]
24%|##3 | 40.0M/170M [00:00<00:02, 52.2MB/s]
28%|##8 | 48.0M/170M [00:00<00:02, 55.5MB/s]
33%|###2 | 56.0M/170M [00:01<00:02, 57.5MB/s]
37%|###6 | 62.3M/170M [00:01<00:01, 58.4MB/s]
40%|#### | 68.1M/170M [00:01<00:01, 59.0MB/s]
43%|####3 | 73.8M/170M [00:01<00:01, 54.7MB/s]
47%|####7 | 80.0M/170M [00:01<00:01, 53.6MB/s]
51%|##### | 86.3M/170M [00:01<00:01, 55.2MB/s]
54%|#####3 | 91.6M/170M [00:01<00:01, 47.5MB/s]
57%|#####6 | 96.3M/170M [00:02<00:02, 34.5MB/s]
61%|######1 | 104M/170M [00:02<00:01, 40.7MB/s]
66%|######5 | 112M/170M [00:02<00:01, 43.0MB/s
]
71%|####### | 120M/170M [00:02<00:01, 46.7MB/s]
75%|#######5 | 128M/170M [00:02<00:00, 50.3MB/s]
80%|######## | 136M/170M [00:02<00:00, 54.8MB/s]
85%|########4 | 144M/170M [00:02<00:00, 56.6MB/s]
89%|########9 | 152M/170M [00:03<00:00, 54.0MB/s]
93%|#########3| 158M/170M [00:03<00:00, 55.7MB/s]
97%|#########6| 165M/170M [00:03<00:00, 58.2MB/s]
100%|##########| 170M/170M [00:03<00:00, 51.9MB/s]
+
0%| | 0.00/170M [00:00<?, ?B/s]
4%|3 | 6.30M/170M [00:00<00:03, 56.7MB/s]
7%|6 | 11.7M/170M [00:00<00:03, 45.4MB/s]
9%|9 | 16.1M/170M [00:00<00:04, 33.1MB/s]
14%|#4 | 24.0M/170M [00:00<00:04, 36.6MB/s]
19%|#8 | 32.0M/170M [00:00<00:03, 41.4MB/s]
24%|##3 | 40.0M/170M [00:00<00:03, 45.0MB/s]
28%|##8 | 48.0M/170M [00:01<00:02, 50.7MB/s]
34%|###3 | 57.1M/170M [00:01<00:01, 61.1MB/s]
38%|###7 | 64.0M/170M [00:01<00:02, 50.5MB/s]
42%|####2 | 72.0M/170M [00:01<00:01, 51.4MB/s]
46%|####6 | 78.3M/170M [00:01<00:01, 51.4MB/s]
50%|####9 | 84.6M/170M [00:01<00:01, 54.8MB/s]
53%|#####3 | 90.2M/170M [00:01<00:01, 50.4MB/s]
56%|#####6 | 95.2M/170M [00:02<00:01, 48.6MB/s]
59%|#####9 | 101M/170M [00:02<00:01, 51.0MB/s]
62%|######2 | 106M/170M [00:02<00:01, 39.7MB/s]
66%|######5 | 112M/170M [00:02<00:01, 37.4MB/s]
71%|####### | 120M/170M [00:02<00:01, 44.6MB/s]
75%|#######5 | 128M/170M [00:02<00:00, 51.9MB/s]
79%|#######9 | 134M/170M [00:02<00:00, 47.8MB/s]
82%|########1 | 139M/170M [00:03<00:00, 40.7MB/s]
85%|########4 | 144M/170M [00:03<00:00, 37.6MB/s]
89%|########9 | 152M/170M [00:03<00:00, 40.6MB/s]
93%|#########3| 158M/170M [00:03<00:00, 44.8MB/s]
96%|#########5| 163M/170M [00:03<00:00, 38.7MB/s]
99%|#########9| 169M/170M [00:03<00:00, 43.9MB/s]
100%|##########| 170M/170M [00:03<00:00, 45.6MB/s]
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torch/nn/functional.py:3897: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
for i in range(dim)
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/detection/anchor_utils.py:124: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor').
@@ -299,7 +299,7 @@ Get boxes with score larger than 0.9
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 3 minutes 46.558 seconds)
+ **Total running time of the script:** ( 3 minutes 34.020 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_object_detection_pytorch.py:
diff --git a/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt b/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
index 2592831a15..fb067c715d 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
@@ -227,7 +227,7 @@ training. Other models require a full post training calibration.
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=MobileNet_V2_Weights.IMAGENET1K_V1`. You can also use `weights=MobileNet_V2_Weights.DEFAULT` to get the most up-to-date weights.
warnings.warn(msg)
Downloading: "https://download.pytorch.org/models/mobilenet_v2-b0353104.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2-b0353104.pth
-
0%| | 0.00/13.6M [00:00<?, ?B/s]
59%|#####8 | 7.99M/13.6M [00:00<00:00, 38.6MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 58.9MB/s]
+
0%| | 0.00/13.6M [00:00<?, ?B/s]
47%|####6 | 6.30M/13.6M [00:00<00:00, 66.1MB/s]
93%|#########3| 12.6M/13.6M [00:00<00:00, 53.1MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 57.6MB/s]
@@ -409,7 +409,7 @@ Here we give an example of how to measure performance of TVM compiled models.
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 88.9308 88.2800 91.3655 88.0628 0.9692
+ 90.0841 90.0309 91.8659 89.8894 0.2243
@@ -458,7 +458,7 @@ TODO
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 19.728 seconds)
+ **Total running time of the script:** ( 1 minutes 16.487 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_prequantized.py:
diff --git a/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt b/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt
index 5c779b4616..aa467d3795 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized_tflite.rst.txt
@@ -423,7 +423,7 @@ Here we give an example of how to measure performance of TVM compiled models.
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 121.0844 121.2201 122.4309 118.6074 0.7515
+ 119.4306 119.5678 121.8174 118.1030 0.7109
@@ -460,7 +460,7 @@ Here we give an example of how to measure performance of TVM compiled models.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 2 minutes 29.938 seconds)
+ **Total running time of the script:** ( 2 minutes 34.338 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_prequantized_tflite.py:
diff --git a/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt b/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
index 08de27c30d..e71bc25549 100644
--- a/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_quantized.rst.txt
@@ -257,7 +257,7 @@ We create a Relay VM to build and execute the model.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 39.936 seconds)
+ **Total running time of the script:** ( 1 minutes 45.103 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_quantized.py:
diff --git a/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt b/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt
index 0b35540685..d69cb05e39 100644
--- a/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_ssd_gluoncv.rst.txt
@@ -170,7 +170,7 @@ Convert and compile model for CPU.
data: None
input_sym_arg_type = in_param.infer_type()[0]
Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/ssd_512_resnet50_v1_voc-9c8b225a.zip...
-
0%| | 0/132723 [00:00<?, ?KB/s]
1%| | 1308/132723 [00:00<00:10, 13040.62KB/s]
3%|3 | 4237/132723 [00:00<00:05, 22586.65KB/s]
8%|7 | 10121/132723 [00:00<00:03, 39126.63KB/s]
12%|#2 | 16441/132723 [00:00<00:02, 48624.36KB/s]
18%|#7 | 23636/132723 [00:00<00:01, 57032.13KB/s]
23%|##3 | 31133/132723 [00:00<00:01, 63125.25KB/s]
29%|##9 | 39038/132723 [00:00<00:01, 68327.34KB/s]
35%|###5 | 46987/132723 [00:00<00:01, 71876.62KB/s]
41%|####1 | 54929/132723 [00:00<00:01, 74232.14KB/s]
47%|####7 | 62701/132723 [00:01<00:00, 75299.27KB/s]
53%|#####3 | 70589/132723 [00:01<00:00, 76390.87KB/s]
59%|#####9 | 78560/132723 [00:01<00:00, 77398.99KB/s]
65%|######5 | 86458/132723 [00:01<00:00, 77874.69KB/s]
71%|#######1 | 94448/132723 [00:01<00:00, 78484.07KB/s]
77%|#######7 | 102380/132723 [00:01<00:00, 78735.08KB/s]
83%|########3 |
110273/132723 [00:01<00:00, 78790.94KB/s]
89%|########9 | 118255/132723 [00:01<00:00, 79098.68KB/s]
95%|#########5| 126172/132723 [00:01<00:00, 79118.26KB/s]
100%|##########| 132723/132723 [00:01<00:00, 70493.20KB/s]
+
0%| | 0/132723 [00:00<?, ?KB/s]
1%|1 | 1613/132723 [00:00<00:08, 16128.59KB/s]
4%|4 | 5330/132723 [00:00<00:04, 28504.02KB/s]
8%|8 | 10902/132723 [00:00<00:02, 40929.15KB/s]
15%|#4 | 19465/132723 [00:00<00:01, 58574.48KB/s]
21%|##1 | 28045/132723 [00:00<00:01, 68387.65KB/s]
28%|##7 | 36629/132723 [00:00<00:01, 74319.99KB/s]
34%|###4 | 45148/132723 [00:00<00:01, 77869.29KB/s]
41%|#### | 53804/132723 [00:00<00:00, 80630.79KB/s]
47%|####7 | 62459/132723 [00:00<00:00, 82478.00KB/s]
54%|#####3 | 71143/132723 [00:01<00:00, 83822.82KB/s]
60%|###### | 79773/132723 [00:01<00:00, 84575.19KB/s]
67%|######6 | 88453/132723 [00:01<00:00, 85250.04KB/s]
73%|#######3 | 97105/132723 [00:01<00:00, 85631.59KB/s]
80%|#######9 | 105812/132723 [00:01<00:00, 86064.69KB/s]
86%|########6 | 114483/132723 [00:01<00:00, 86256.98KB/s]
93%|#########2
| 123165/132723 [00:01<00:00, 86424.23KB/s]
99%|#########9| 131808/132723 [00:01<00:00, 86417.50KB/s]
100%|##########| 132723/132723 [00:01<00:00, 77487.62KB/s]
@@ -246,7 +246,7 @@ Display result
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 3 minutes 59.099 seconds)
+ **Total running time of the script:** ( 3 minutes 48.257 seconds)
.. _sphx_glr_download_how_to_deploy_models_deploy_ssd_gluoncv.py:
diff --git a/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt b/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt
index 851f61871e..65a695188f 100644
--- a/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/deploy_models/sg_execution_times.rst.txt
@@ -5,28 +5,28 @@
Computation times
=================
-**16:49.430** total execution time for **how_to_deploy_models** files:
+**16:25.272** total execution time for **how_to_deploy_models** files:
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 03:59.099 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 03:48.257 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:46.558 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:34.020 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:29.938 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:34.338 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:39.936 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:45.103 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:19.728 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:16.487 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno.py` (``deploy_model_on_adreno.py``) | 00:58.394 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno.py` (``deploy_model_on_adreno.py``) | 00:54.919 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno_tvmc.py` (``deploy_model_on_adreno_tvmc.py``) | 00:53.202 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno_tvmc.py` (``deploy_model_on_adreno_tvmc.py``) | 00:52.118 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:44.689 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:42.890 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_nano.py` (``deploy_model_on_nano.py``) | 00:29.210 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_nano.py` (``deploy_model_on_nano.py``) | 00:28.694 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:28.670 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:28.440 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_deploy_models_deploy_sparse.py` (``deploy_sparse.py``) | 00:00.006 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt b/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
index 1bf33b1e0b..61ad223229 100644
--- a/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/bring_your_own_datatypes.rst.txt
@@ -463,7 +463,7 @@ First let us define two helper functions to get the mobilenet model and a cat im
.. code-block:: none
- Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip929e5e9d-8d88-4ea5-95fc-5c8e6ad4ee23 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+ Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zipeea5ca4e-641e-4da1-8968-6e0f02dd652e from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
diff --git a/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt b/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
index c5d02f6a30..bd61d5ab2f 100644
--- a/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/sg_execution_times.rst.txt
@@ -5,14 +5,14 @@
Computation times
=================
-**00:58.769** total execution time for **how_to_extend_tvm** files:
+**00:54.099** total execution time for **how_to_extend_tvm** files:
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:54.661 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:50.187 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.952 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.801 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:01.149 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:01.105 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``) | 00:00.008 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_low_level_custom_pass.py` (``low_level_custom_pass.py``) | 00:00.007 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt b/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
index 9f241b2239..b1de514d61 100644
--- a/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
+++ b/docs/_sources/how_to/extend_tvm/use_pass_instrument.rst.txt
@@ -220,10 +220,10 @@ profile the execution time of each passes.
.. code-block:: none
Printing results of timing profile...
- InferType: 23565us [23565us] (49.07%; 49.07%)
- FoldScaleAxis: 24454us [9us] (50.93%; 50.93%)
- FoldConstant: 24444us [1762us] (50.91%; 99.96%)
- InferType: 22683us [22683us] (47.24%; 92.79%)
+ InferType: 22163us [22163us] (48.64%; 48.64%)
+ FoldScaleAxis: 23400us [8us] (51.36%; 51.36%)
+ FoldConstant: 23391us [1749us] (51.34%; 99.96%)
+ InferType: 21643us [21643us] (47.50%; 92.52%)
@@ -262,10 +262,10 @@ Refer to following sections and :py:func:`tvm.instrument.pass_instrument` for th
.. code-block:: none
Printing results of timing profile...
- InferType: 23804us [23804us] (49.22%; 49.22%)
- FoldScaleAxis: 24559us [8us] (50.78%; 50.78%)
- FoldConstant: 24551us [1875us] (50.76%; 99.97%)
- InferType: 22675us [22675us] (46.89%; 92.36%)
+ InferType: 21773us [21773us] (48.20%; 48.20%)
+ FoldScaleAxis: 23399us [6us] (51.80%; 51.80%)
+ FoldConstant: 23393us [1794us] (51.79%; 99.97%)
+ InferType: 21599us [21599us] (47.82%; 92.33%)
diff --git a/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt b/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt
index f59edd18d2..ca8941505e 100644
--- a/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_conv_cuda.rst.txt
@@ -331,7 +331,7 @@ latency of convolution.
.. code-block:: none
- Convolution: 53.512542 ms
+ Convolution: 34.863105 ms
diff --git a/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt b/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt
index 216860cd2d..0e81fc60b0 100644
--- a/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_conv_tensorcore.rst.txt
@@ -598,7 +598,7 @@ be able to run on our build server
.. code-block:: none
- conv2d with tensor core: 12.223469 ms
+ conv2d with tensor core: 12.234599 ms
diff --git a/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt b/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
index da859c2f73..70921ae7f1 100644
--- a/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
@@ -134,8 +134,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
.. code-block:: none
- Numpy running time: 0.019328
- Baseline: 3.498159
+ Numpy running time: 0.017909
+ Baseline: 3.397584
@@ -227,7 +227,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
.. code-block:: none
- Opt1: 0.329814
+ Opt1: 0.303157
@@ -318,7 +318,7 @@ In this tutorial, we chose to vectorize the inner loop row data since it is cach
.. code-block:: none
- Opt2: 0.353135
+ Opt2: 0.331700
@@ -406,7 +406,7 @@ the access pattern for A matrix is more cache friendly.
.. code-block:: none
- Opt3: 0.134273
+ Opt3: 0.124434
@@ -523,7 +523,7 @@ flattening.
.. code-block:: none
- Opt4: 0.110898
+ Opt4: 0.110773
@@ -635,7 +635,7 @@ write to C when all the block results are ready.
.. code-block:: none
- Opt5: 0.112132
+ Opt5: 0.111581
@@ -748,7 +748,7 @@ Furthermore, we can also utilize multi-core processors to do the thread-level pa
.. code-block:: none
- Opt6: 0.148916
+ Opt6: 0.146901
diff --git a/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt b/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt
index 4e59eb4bbb..17558b4808 100644
--- a/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/sg_execution_times.rst.txt
@@ -5,12 +5,12 @@
Computation times
=================
-**00:36.796** total execution time for **how_to_optimize_operators** files:
+**00:35.333** total execution time for **how_to_optimize_operators** files:
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:33.689 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:32.406 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.894 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.830 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.212 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.097 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
index 09ebb8bb55..a668250a38 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/sg_execution_times.rst.txt
@@ -5,18 +5,18 @@
Computation times
=================
-**10:36.831** total execution time for **how_to_tune_with_autoscheduler** files:
+**10:16.843** total execution time for **how_to_tune_with_autoscheduler** files:
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 06:28.831 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 06:10.766 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:47.644 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:42.680 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 01:13.908 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 01:10.583 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:36.900 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:44.969 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:15.119 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:14.425 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:14.429 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:13.420 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
index 987f55e14e..4f1d8f0a37 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.rst.txt
@@ -244,85 +244,550 @@ cooperative fetching, unrolling and operator fusion.
def main(data: T.Buffer((1, 512, 7, 7), "float32"), kernel: T.Buffer((512, 512, 3, 3), "float32"), bias: T.Buffer((1, 512, 1, 1), "float32"), compute: T.Buffer((1, 512, 7, 7), "float32")):
T.func_attr({"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True})
blockIdx_x = T.launch_thread("blockIdx.x", 32)
- conv2d_nchw = T.allocate([7], "float32", "local")
- pad_temp_shared = T.allocate([1568], "float32", "shared")
- kernel_shared = T.allocate([512], "float32", "shared")
- threadIdx_x = T.launch_thread("threadIdx.x", 112)
- conv2d_nchw_1 = T.Buffer((7,), data=conv2d_nchw, scope="local", align=16)
+ conv2d_nchw = T.allocate([8], "float32", "local")
+ pad_temp_shared = T.allocate([4032], "float32", "shared")
+ kernel_shared = T.allocate([3072], "float32", "shared")
+ threadIdx_x = T.launch_thread("threadIdx.x", 98)
+ conv2d_nchw_1 = T.Buffer((4,), data=conv2d_nchw, scope="local", align=8)
conv2d_nchw_1[0] = T.float32(0)
- conv2d_nchw_1[1] = T.float32(0)
conv2d_nchw_1[2] = T.float32(0)
- conv2d_nchw_1[3] = T.float32(0)
conv2d_nchw_1[4] = T.float32(0)
- conv2d_nchw_1[5] = T.float32(0)
conv2d_nchw_1[6] = T.float32(0)
- for rc_outer_outer, ry_outer_outer, rx_outer_outer in T.grid(16, 3, 3):
- cse_var_2: T.int32 = rc_outer_outer * 288
- cse_var_1: T.int32 = ry_outer_outer * 3
+ conv2d_nchw_1[1] = T.float32(0)
+ conv2d_nchw_1[3] = T.float32(0)
+ conv2d_nchw_1[5] = T.float32(0)
+ conv2d_nchw_1[7] = T.float32(0)
+ for rc_outer_outer, rx_outer_outer in T.grid(8, 3):
+ cse_var_2: T.int32 = rc_outer_outer * 3136
+ cse_var_1: T.int32 = rc_outer_outer * 576
threadIdx_x_1 = T.env_thread("threadIdx.x")
- pad_temp_shared_1 = T.Buffer((1568,), data=pad_temp_shared, scope="shared")
+ pad_temp_shared_1 = T.Buffer((4032,), data=pad_temp_shared, scope="shared")
data_1 = T.Buffer((25088,), data=data.data)
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer - 8], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 112] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 104], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 224] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 216], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 336] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 328], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 448] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 440], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 560] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 552], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 672] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 664], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 784] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 776], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 896] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 888], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1008] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1000], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1120] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1112], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1232] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1224], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1344] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1336], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1456] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1448], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2] = T.if_then_else(7 <= threadIdx_x_1 * 2 % 63 and threadIdx_x_1 * 2 % 63 < 56 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + threadIdx_x_1 * 2 // 63 * 49 + rx_outer_outer + threadIdx_x_1 * 2 % 63 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1] = T.if_then_else(7 <= (threadIdx_x_1 * 2 + 1) % 63 and (threadIdx_x_1 * 2 + 1) % 63 < 56 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1) // 63 * 49 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 63 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 196] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 1) % 9 and (threadIdx_x_1 * 2 // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 196) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 1) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 197] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 197) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 392] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 2) % 9 and (threadIdx_x_1 * 2 // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 392) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 2) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 393] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 393) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 588] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 3) % 9 and (threadIdx_x_1 * 2 // 7 + 3) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 588) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 3) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 589] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 589) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 784] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 4) % 9 and (threadIdx_x_1 * 2 // 7 + 4) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 784) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 4) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 785] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 785) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 980] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 5) % 9 and (threadIdx_x_1 * 2 // 7 + 5) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 980) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 5) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 981] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 981) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1176] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 6) % 9 and (threadIdx_x_1 * 2 // 7 + 6) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1176) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 6) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1177] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1177) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1372] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 7) % 9 and (threadIdx_x_1 * 2 // 7 + 7) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1372) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 7) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1373] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1373) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1568] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 8) % 9 and (threadIdx_x_1 * 2 // 7 + 8) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1568) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 8) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1569] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1569) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1764] = T.if_then_else(7 <= threadIdx_x_1 * 2 % 63 and threadIdx_x_1 * 2 % 63 < 56 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + threadIdx_x_1 * 2 // 63 * 49 + rx_outer_outer + threadIdx_x_1 * 2 % 63 + 1364], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1765] = T.if_then_else(7 <= (threadIdx_x_1 * 2 + 1) % 63 and (threadIdx_x_1 * 2 + 1) % 63 < 56 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1) // 63 * 49 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 63 + 1364], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1960] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 1) % 9 and (threadIdx_x_1 * 2 // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1960) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 1) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1961] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1961) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2156] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 2) % 9 and (threadIdx_x_1 * 2 // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2156) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 2) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2157] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2157) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2352] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 3) % 9 and (threadIdx_x_1 * 2 // 7 + 3) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2352) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 3) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2353] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2353) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2548] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 4) % 9 and (threadIdx_x_1 * 2 // 7 + 4) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2548) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 4) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2549] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2549) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2744] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 5) % 9 and (threadIdx_x_1 * 2 // 7 + 5) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2744) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 5) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2745] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2745) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2940] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 6) % 9 and (threadIdx_x_1 * 2 // 7 + 6) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2940) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 6) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2941] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2941) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3136] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 7) % 9 and (threadIdx_x_1 * 2 // 7 + 7) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3136) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 7) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3137] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3137) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3332] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 8) % 9 and (threadIdx_x_1 * 2 // 7 + 8) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3332) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 8) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3333] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3333) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3528] = T.if_then_else(7 <= threadIdx_x_1 * 2 % 63 and threadIdx_x_1 * 2 % 63 < 56 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + threadIdx_x_1 * 2 // 63 * 49 + rx_outer_outer + threadIdx_x_1 * 2 % 63 + 2736], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3529] = T.if_then_else(7 <= (threadIdx_x_1 * 2 + 1) % 63 and (threadIdx_x_1 * 2 + 1) % 63 < 56 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1) // 63 * 49 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 63 + 2736], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3724] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 1) % 9 and (threadIdx_x_1 * 2 // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3724) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 1) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3725] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3725) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ if T.likely(threadIdx_x_1 < 56):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3920] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 2) % 9 and (threadIdx_x_1 * 2 // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3920) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 2) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ if T.likely(threadIdx_x_1 < 56):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3921] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3921) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
threadIdx_x_2 = T.env_thread("threadIdx.x")
- kernel_shared_1 = T.Buffer((512,), data=kernel_shared, scope="shared")
+ kernel_shared_1 = T.Buffer((3072,), data=kernel_shared, scope="shared")
kernel_1 = T.Buffer((2359296,), data=kernel.data)
- with T.launch_thread(threadIdx_x_2, 112):
- kernel_shared_1[threadIdx_x_2 * 3] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 * 3 // 32 * 4608 + cse_var_2 + threadIdx_x_2 * 3 % 32 * 9 + cse_var_1 + rx_outer_outer]
- kernel_shared_1[threadIdx_x_2 * 3 + 1] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 1) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 1) % 32 * 9 + cse_var_1 + rx_outer_outer]
- kernel_shared_1[threadIdx_x_2 * 3 + 2] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 2) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 2) % 32 * 9 + cse_var_1 + rx_outer_outer]
- with T.launch_thread(threadIdx_x_2, 112):
- if T.likely(threadIdx_x_2 < 59):
- kernel_shared_1[threadIdx_x_2 * 3 + 336] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 336) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 16) % 32 * 9 + cse_var_1 + rx_outer_outer]
- if T.likely(threadIdx_x_2 < 59):
- kernel_shared_1[threadIdx_x_2 * 3 + 337] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 337) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 17) % 32 * 9 + cse_var_1 + rx_outer_outer]
- if T.likely(threadIdx_x_2 < 58):
- kernel_shared_1[threadIdx_x_2 * 3 + 338] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 338) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 18) % 32 * 9 + cse_var_1 + rx_outer_outer]
- for rc_outer_inner in range(16):
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 1] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 2] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 3] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 4] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 5] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 6] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 49] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 50] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 51] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 52] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 53] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- for i3_inner in range(7):
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 98] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 98) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 98) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 196] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 196) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 12]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 294] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 294) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 102) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 392] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 392) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 24]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 490] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 490) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 106) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 588] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 588) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 36]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 686] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 686) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 110) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 784] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 784) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 48]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 882] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 882) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 114) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 980] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 980) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 60]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1078] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1078) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 118) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1176] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1176) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 72]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1274] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1274) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 122) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1372] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1372) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 84]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1470] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1470) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 126) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1568] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1568) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 96]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1666] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1666) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 130) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1764] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1764) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 108]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1862] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1862) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 134) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1960] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1960) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 120]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2058] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2058) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 138) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2156] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2156) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 132]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2254] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2254) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 142) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2352] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2352) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 144]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2450] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2450) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 146) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2548] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2548) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 156]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2646] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2646) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 150) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2744] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2744) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 168]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2842] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2842) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 154) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2940] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2940) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 180]
+ with T.launch_thread(threadIdx_x_2, 98):
+ if T.likely(threadIdx_x_2 < 34):
+ kernel_shared_1[threadIdx_x_2 + 3038] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3038) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 474]
+ for rc_outer_inner in range(4):
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 768]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1536]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2304]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 769]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1537]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2305]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 770]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1538]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2306]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 3]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 771]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1539]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2307]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 4]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 772]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1540]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2308]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 5]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 773]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1541]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2309]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 6]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 774]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1542]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2310]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 7]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 775]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1543]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2311]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 8]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 776]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1544]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2312]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 9]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 777]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1545]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2313]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 10]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 778]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1546]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2314]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 11]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 779]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1547]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2315]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 12]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 780]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1548]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2316]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 13]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 781]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1549]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2317]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 14]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 782]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1550]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2318]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 15]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 783]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1551]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2319]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 16]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 784]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1552]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2320]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 17]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 785]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1553]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2321]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 18]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 786]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1554]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2322]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 19]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 787]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1555]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2323]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 20]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 788]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1556]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2324]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 21]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 789]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1557]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2325]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 22]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 790]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1558]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2326]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 23]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 791]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1559]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2327]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 24]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 792]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1560]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2328]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 25]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 793]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1561]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2329]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 26]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 794]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1562]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2330]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 27]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 795]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1563]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2331]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 28]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 796]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1564]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2332]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 29]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 797]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1565]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2333]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 30]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 798]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1566]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2334]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 31]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 799]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1567]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2335]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 32]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 800]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1568]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2336]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 33]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 801]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1569]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2337]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 34]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 802]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1570]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2338]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 35]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 803]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1571]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2339]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 36]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 804]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1572]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2340]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 37]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 805]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1573]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2341]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 38]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 806]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1574]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2342]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 39]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 807]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1575]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2343]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 40]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 808]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1576]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2344]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 41]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 809]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1577]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2345]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 42]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 810]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1578]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2346]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 43]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 811]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1579]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2347]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 44]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 812]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1580]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2348]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 45]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 813]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1581]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2349]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 46]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 814]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1582]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2350]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 47]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 815]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1583]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2351]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 192]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 960]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1728]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2496]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 193]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 961]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1729]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2497]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 194]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 962]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1730]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2498]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 195]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 963]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1731]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2499]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 196]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 964]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1732]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2500]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 197]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 965]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1733]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2501]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 198]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 966]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1734]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2502]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 199]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 967]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1735]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2503]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 200]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 968]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1736]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2504]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 201]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 969]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1737]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2505]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 202]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 970]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1738]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2506]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 203]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 971]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1739]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2507]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 204]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 972]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1740]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2508]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 205]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 973]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1741]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2509]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 206]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 974]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1742]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2510]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 207]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 975]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1743]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2511]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 208]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 976]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1744]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2512]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 209]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 977]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1745]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2513]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 210]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 978]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1746]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2514]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 211]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 979]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1747]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2515]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 212]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 980]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1748]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2516]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 213]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 981]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1749]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2517]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 214]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 982]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1750]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2518]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 215]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 983]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1751]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2519]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 216]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 984]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1752]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2520]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 217]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 985]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1753]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2521]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 218]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 986]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1754]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2522]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 219]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 987]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1755]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2523]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 220]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 988]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1756]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2524]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 221]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 989]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1757]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2525]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 222]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 990]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1758]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2526]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 223]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 991]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1759]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2527]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 224]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 992]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1760]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2528]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 225]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 993]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1761]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2529]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 226]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 994]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1762]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2530]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 227]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 995]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1763]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2531]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 228]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 996]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1764]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2532]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 229]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 997]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1765]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2533]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 230]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 998]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1766]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2534]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 231]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 999]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1767]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2535]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 232]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1000]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1768]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2536]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 233]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1001]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1769]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2537]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 234]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1002]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1770]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2538]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 235]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1003]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1771]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2539]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 236]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1004]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1772]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2540]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 237]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1005]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1773]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2541]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 238]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1006]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1774]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2542]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 239]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1007]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1775]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2543]
+ for i1_inner in range(2):
compute_1 = T.Buffer((25088,), data=compute.data)
bias_1 = T.Buffer((512,), data=bias.data)
- compute_1[blockIdx_x * 784 + threadIdx_x * 7 + i3_inner] = T.max(conv2d_nchw_1[i3_inner] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49] = T.max(conv2d_nchw_1[i1_inner] + bias_1[blockIdx_x * 16 + threadIdx_x // 49 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49 + 196] = T.max(conv2d_nchw_1[i1_inner + 2] + bias_1[blockIdx_x * 16 + threadIdx_x // 49 * 2 + i1_inner + 4], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49 + 392] = T.max(conv2d_nchw_1[i1_inner + 4] + bias_1[blockIdx_x * 16 + threadIdx_x // 49 * 2 + i1_inner + 8], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49 + 588] = T.max(conv2d_nchw_1[i1_inner + 6] + bias_1[blockIdx_x * 16 + threadIdx_x // 49 * 2 + i1_inner + 12], T.float32(0))
@@ -372,7 +837,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 0.387 ms
+ Execution time of this operator: 0.229 ms
@@ -421,20 +886,20 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
conv2d_nchw_nn_o_o_o_i, conv2d_nchw_nn_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_i, factor=1)
conv2d_nchw_nn_o_o_o_o, conv2d_nchw_nn_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_o_i, factor=1)
conv2d_nchw_ff_o_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=1)
- conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=1)
- conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=16)
- conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=1)
+ conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=2)
+ conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=2)
+ conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=4)
conv2d_nchw_yy_o_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=1)
conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=7)
conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=1)
- conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=7)
+ conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=1)
conv2d_nchw_xx_o_o_i, conv2d_nchw_xx_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_i, factor=1)
- conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=1)
+ conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=7)
conv2d_nchw_xx_o_o_o_o, conv2d_nchw_xx_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_o_i, factor=1)
- conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=2)
- conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=16)
- conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=1)
+ conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=16)
+ conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=4)
+ conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=3)
conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=1)
conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=1)
conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
@@ -442,14 +907,14 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
compute_i0_o_i, compute_i0_i = s[compute].split(compute_i0, factor=1)
compute_i0_o_o_i, compute_i0_o_i = s[compute].split(compute_i0_o_i, factor=1)
compute_i0_o_o_o, compute_i0_o_o_i = s[compute].split(compute_i0_o_o_i, factor=1)
- compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=1)
- compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
- compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=1)
+ compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=2)
+ compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=2)
+ compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=4)
compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=1)
compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=7)
compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
- compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=7)
- compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=1)
+ compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=1)
+ compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=7)
compute_i3_o_o_o, compute_i3_o_o_i = s[compute].split(compute_i3_o_o_i, factor=1)
s[compute].reorder(compute_i0_o_o_o, compute_i1_o_o_o, compute_i2_o_o_o, compute_i3_o_o_o, compute_i0_o_o_i, compute_i1_o_o_i, compute_i2_o_o_i, compute_i3_o_o_i, compute_i0_o_i, compute_i1_o_i, compute_i2_o_i, compute_i3_o_i, compute_i0_i, compute_i1_i, compute_i2_i, compute_i3_i)
s[conv2d_nchw].compute_at(s[compute], compute_i3_o_i)
@@ -467,16 +932,16 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused = s[compute].fuse(compute_i0_o_i, compute_i1_o_i, compute_i2_o_i, compute_i3_o_i)
s[compute].bind(compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused, te.thread_axis("threadIdx.x"))
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[kernel_shared].fuse(kernel_shared_ax0, kernel_shared_ax1, kernel_shared_ax2, kernel_shared_ax3)
- kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=3)
+ kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[kernel_shared].vectorize(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
- kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=112)
+ kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=98)
s[kernel_shared].bind(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[pad_temp_shared].fuse(pad_temp_shared_ax0, pad_temp_shared_ax1, pad_temp_shared_ax2, pad_temp_shared_ax3)
- pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
+ pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=2)
s[pad_temp_shared].vectorize(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
- pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=112)
+ pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=98)
s[pad_temp_shared].bind(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
- s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 64)
+ s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 1024)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
CUDA source code:
@@ -494,69 +959,495 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
#define int64_t long long
#define uint64_t unsigned long long
#endif
- extern "C" __global__ void __launch_bounds__(112) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[7];
- __shared__ float pad_temp_shared[1568];
- __shared__ float kernel_shared[512];
+ extern "C" __global__ void __launch_bounds__(98) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
+ float conv2d_nchw[8];
+ __shared__ float pad_temp_shared[4032];
+ __shared__ float kernel_shared[3072];
conv2d_nchw[0] = 0.000000e+00f;
- conv2d_nchw[1] = 0.000000e+00f;
conv2d_nchw[2] = 0.000000e+00f;
- conv2d_nchw[3] = 0.000000e+00f;
conv2d_nchw[4] = 0.000000e+00f;
- conv2d_nchw[5] = 0.000000e+00f;
conv2d_nchw[6] = 0.000000e+00f;
- for (int rc_outer_outer = 0; rc_outer_outer < 16; ++rc_outer_outer) {
- for (int ry_outer_outer = 0; ry_outer_outer < 3; ++ry_outer_outer) {
- for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
- __syncthreads();
- pad_temp_shared[((int)threadIdx.x)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) - 8)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 104)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 216)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 328)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 440)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 552)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 672)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 664)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 776)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 896)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 888)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1008)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1000)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1120)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1112)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1232)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1224)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1344)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1336)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1456)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1448)] : 0.000000e+00f);
- kernel_shared[(((int)threadIdx.x) * 3)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) * 3) >> 5) * 4608)) + (rc_outer_outer * 288)) + (((((int)threadIdx.x) * 3) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- kernel_shared[((((int)threadIdx.x) * 3) + 1)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 1) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 1) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- kernel_shared[((((int)threadIdx.x) * 3) + 2)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 2) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 2) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- if (((int)threadIdx.x) < 59) {
- kernel_shared[((((int)threadIdx.x) * 3) + 336)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 336) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 16) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- if (((int)threadIdx.x) < 59) {
- kernel_shared[((((int)threadIdx.x) * 3) + 337)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 337) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 17) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- if (((int)threadIdx.x) < 58) {
- kernel_shared[((((int)threadIdx.x) * 3) + 338)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 338) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 18) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- __syncthreads();
- for (int rc_outer_inner = 0; rc_outer_inner < 16; ++rc_outer_inner) {
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7))] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 1)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 2)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 3)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 4)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 5)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 6)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 49)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 50)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 51)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 52)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 53)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- }
+ conv2d_nchw[1] = 0.000000e+00f;
+ conv2d_nchw[3] = 0.000000e+00f;
+ conv2d_nchw[5] = 0.000000e+00f;
+ conv2d_nchw[7] = 0.000000e+00f;
+ for (int rc_outer_outer = 0; rc_outer_outer < 8; ++rc_outer_outer) {
+ for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
+ __syncthreads();
+ pad_temp_shared[(((int)threadIdx.x) * 2)] = (((((7 <= ((((int)threadIdx.x) * 2) % 63)) && (((((int)threadIdx.x) * 2) % 63) < 56)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + (((((int)threadIdx.x) * 2) / 63) * 49)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 63)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1)] = (((((7 <= (((((int)threadIdx.x) * 2) + 1) % 63)) && ((((((int)threadIdx.x) * 2) + 1) % 63) < 56)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1) / 63) * 49)) + rx_outer_outer) + (((((int)threadIdx.x) * 2) + 1) % 63)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 196)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 1) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 196) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0.00 [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 197)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 197) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) * 7)) + rx_outer_outer) + (((((int)threa [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 392)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 2) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 392) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0.00 [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 393)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 393) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) * 7)) + rx_outer_outer) + (((((int)threa [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 588)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 588) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 3) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0.00 [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 589)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 589) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9) * 7)) + rx_outer_outer) + (((((int)threa [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 784)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 4) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 784) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 4) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0.00 [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 785)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 785) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9) * 7)) + rx_outer_outer) + (((((int)threa [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 980)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 5) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 980) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 5) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0.00 [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 981)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 981) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9) * 7)) + rx_outer_outer) + (((((int)threa [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1176)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1176) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 6) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1177)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1177) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1372)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 7) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1372) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 7) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1373)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1373) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1568)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 8) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1568) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 8) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1569)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1569) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1764)] = (((((7 <= ((((int)threadIdx.x) * 2) % 63)) && (((((int)threadIdx.x) * 2) % 63) < 56)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + (((((int)threadIdx.x) * 2) / 63) * 49)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 63)) + 1364)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1765)] = (((((7 <= (((((int)threadIdx.x) * 2) + 1) % 63)) && ((((((int)threadIdx.x) * 2) + 1) % 63) < 56)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1) / 63) * 49)) + rx_outer_outer) + (((((int)threadIdx.x) * 2) + 1) % 63)) + 1364)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1960)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 1) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1960) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1961)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1961) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2156)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 2) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2156) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2157)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2157) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2352)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2352) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 3) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2353)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2353) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2548)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 4) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2548) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 4) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2549)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2549) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2744)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 5) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2744) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 5) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2745)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2745) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2940)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2940) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 6) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2941)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2941) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3136)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 7) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3136) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 7) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3137)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3137) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3332)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 8) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3332) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 8) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3333)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3333) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3528)] = (((((7 <= ((((int)threadIdx.x) * 2) % 63)) && (((((int)threadIdx.x) * 2) % 63) < 56)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + (((((int)threadIdx.x) * 2) / 63) * 49)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 63)) + 2736)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3529)] = (((((7 <= (((((int)threadIdx.x) * 2) + 1) % 63)) && ((((((int)threadIdx.x) * 2) + 1) % 63) < 56)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1) / 63) * 49)) + rx_outer_outer) + (((((int)threadIdx.x) * 2) + 1) % 63)) + 2736)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3724)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 1) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3724) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : 0. [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3725)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3725) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) * 7)) + rx_outer_outer) + (((((int)thr [...]
+ if (((int)threadIdx.x) < 56) {
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3920)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 2) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3920) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) * 7)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 7)) - 8)] : [...]
+ }
+ if (((int)threadIdx.x) < 56) {
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3921)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3921) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) * 7)) + rx_outer_outer) + (((((int)t [...]
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 98)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 98) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 98) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 196)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 196) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 12)];
+ kernel_shared[(((int)threadIdx.x) + 294)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 294) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 102) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 392)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 392) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 24)];
+ kernel_shared[(((int)threadIdx.x) + 490)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 490) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 106) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 588)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 588) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 36)];
+ kernel_shared[(((int)threadIdx.x) + 686)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 686) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 110) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 784) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 48)];
+ kernel_shared[(((int)threadIdx.x) + 882)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 882) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 114) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 980)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 980) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 60)];
+ kernel_shared[(((int)threadIdx.x) + 1078)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1078) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 118) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1176)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1176) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 72)];
+ kernel_shared[(((int)threadIdx.x) + 1274)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1274) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 122) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1372)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1372) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 84)];
+ kernel_shared[(((int)threadIdx.x) + 1470)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1470) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 126) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1568)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1568) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 96)];
+ kernel_shared[(((int)threadIdx.x) + 1666)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1666) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 130) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1764)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1764) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 108)];
+ kernel_shared[(((int)threadIdx.x) + 1862)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1862) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 134) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1960)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1960) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 120)];
+ kernel_shared[(((int)threadIdx.x) + 2058)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2058) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 138) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2156)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2156) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 132)];
+ kernel_shared[(((int)threadIdx.x) + 2254)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2254) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 142) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2352)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2352) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 144)];
+ kernel_shared[(((int)threadIdx.x) + 2450)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2450) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 146) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2548)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2548) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 156)];
+ kernel_shared[(((int)threadIdx.x) + 2646)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2646) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 150) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2744)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2744) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 168)];
+ kernel_shared[(((int)threadIdx.x) + 2842)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2842) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 154) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2940)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2940) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 180)];
+ if (((int)threadIdx.x) < 34) {
+ kernel_shared[(((int)threadIdx.x) + 3038)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3038) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 474)];
+ }
+ __syncthreads();
+ for (int rc_outer_inner = 0; rc_outer_inner < 4; ++rc_outer_inner) {
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[(((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48))]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 768)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1536)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2304)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 769)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1537)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2305)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 770)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1538)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2306)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 3)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 771)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1539)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2307)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 4)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 772)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1540)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2308)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 5)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 773)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1541)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2309)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 6)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 774)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1542)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2310)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 7)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 775)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1543)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2311)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 8)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 776)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1544)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2312)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 9)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 777)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1545)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2313)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 10)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 778)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1546)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2314)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 11)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 779)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1547)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2315)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 12)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 780)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1548)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2316)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 13)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 781)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1549)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2317)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 14)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 782)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1550)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2318)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 15)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 783)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1551)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2319)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 16)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 784)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1552)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2320)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 17)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 785)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1553)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2321)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 18)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 786)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1554)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2322)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 19)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 787)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1555)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2323)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 20)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 788)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1556)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2324)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 21)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 789)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1557)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2325)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 22)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 790)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1558)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2326)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 23)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 791)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1559)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2327)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 24)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 792)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1560)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2328)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 25)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 793)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1561)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2329)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 26)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 794)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1562)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2330)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 27)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 795)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1563)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2331)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 28)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 796)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1564)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2332)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 29)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 797)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1565)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2333)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 30)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 798)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1566)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2334)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 31)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 799)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1567)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2335)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 32)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 800)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1568)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2336)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 33)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 801)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1569)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2337)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 34)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 802)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1570)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2338)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 35)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 803)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1571)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2339)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 36)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 804)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1572)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2340)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 37)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 805)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1573)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2341)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 38)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 806)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1574)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2342)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 39)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 807)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1575)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2343)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 40)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 808)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1576)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2344)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 41)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 809)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1577)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2345)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 42)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 810)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1578)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2346)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 43)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 811)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1579)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2347)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 44)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 812)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1580)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2348)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 45)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 813)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1581)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2349)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 46)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 814)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1582)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2350)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 47)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 815)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1583)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2351)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 192)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 960)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1728)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2496)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 193)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 961)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1729)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2497)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 194)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 962)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1730)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2498)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 195)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 963)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1731)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2499)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 196)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 964)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1732)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2500)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 197)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 965)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1733)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2501)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 198)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 966)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1734)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2502)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 199)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 967)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1735)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2503)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 200)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 968)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1736)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2504)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 201)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 969)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1737)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2505)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 202)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 970)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1738)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2506)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 203)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 971)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1739)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2507)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 204)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 972)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1740)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2508)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 205)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 973)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1741)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2509)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 206)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 974)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1742)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2510)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 207)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 975)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1743)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2511)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 208)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 976)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1744)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2512)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 209)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 977)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1745)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2513)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 210)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 978)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1746)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2514)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 211)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 979)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1747)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2515)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 212)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 980)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1748)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2516)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 213)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 981)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1749)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2517)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 214)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 982)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1750)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2518)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 215)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 983)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1751)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2519)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 216)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 984)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1752)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2520)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 217)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 985)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1753)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2521)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 218)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 986)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1754)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2522)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 219)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 987)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1755)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2523)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 220)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 988)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1756)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2524)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 221)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 989)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1757)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2525)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 222)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 990)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1758)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2526)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 223)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 991)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1759)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2527)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 224)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 992)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1760)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2528)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 225)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 993)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1761)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2529)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 226)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 994)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1762)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2530)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 227)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 995)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1763)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2531)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 228)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 996)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1764)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2532)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 229)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 997)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1765)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2533)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 230)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 998)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1766)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2534)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 231)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 999)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1767)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2535)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 232)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1000)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1768)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2536)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 233)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1001)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1769)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2537)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 234)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1002)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1770)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2538)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 235)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1003)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1771)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2539)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 236)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1004)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1772)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2540)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 237)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1005)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1773)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2541)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 238)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1006)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1774)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2542)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 239)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1007)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1775)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2543)]));
}
}
}
- for (int i3_inner = 0; i3_inner < 7; ++i3_inner) {
- compute[(((((int)blockIdx.x) * 784) + (((int)threadIdx.x) * 7)) + i3_inner)] = max((conv2d_nchw[i3_inner] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ for (int i1_inner = 0; i1_inner < 2; ++i1_inner) {
+ compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49))] = max((conv2d_nchw[i1_inner] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49)) + 196)] = max((conv2d_nchw[(i1_inner + 2)] + bias[((((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner) + 4)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49)) + 392)] = max((conv2d_nchw[(i1_inner + 4)] + bias[((((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner) + 8)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49)) + 588)] = max((conv2d_nchw[(i1_inner + 6)] + bias[((((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner) + 12)]), 0.000000e+00f);
}
}
@@ -616,7 +1507,7 @@ In the example below we resume the status and do more 5 trials.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 6 minutes 28.831 seconds)
+ **Total running time of the script:** ( 6 minutes 10.766 seconds)
.. _sphx_glr_download_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py:
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt
index e81f52e392..6570b620b3 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_cuda.rst.txt
@@ -647,7 +647,7 @@ so we can read the log file and load the best schedules.
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 8.1199 8.1161 8.1291 8.1144 0.0065
+ 8.0999 8.0999 8.1066 8.0931 0.0055
@@ -675,7 +675,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 13.908 seconds)
+ **Total running time of the script:** ( 1 minutes 10.583 seconds)
.. _sphx_glr_download_how_to_tune_with_autoscheduler_tune_network_cuda.py:
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt
index 01e1769509..1a63c75c50 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_network_x86.rst.txt
@@ -666,7 +666,7 @@ so we can read the log file and load the best schedules.
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 770.8157 771.0289 771.6153 769.8030 0.7551
+ 762.8765 762.7099 764.5182 761.4014 1.2779
@@ -694,7 +694,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 47.644 seconds)
+ **Total running time of the script:** ( 1 minutes 42.680 seconds)
.. _sphx_glr_download_how_to_tune_with_autoscheduler_tune_network_x86.py:
diff --git a/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt b/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt
index 898fb7f092..32e6573506 100644
--- a/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt
+++ b/docs/_sources/how_to/tune_with_autoscheduler/tune_sparse_x86.rst.txt
@@ -389,25 +389,85 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
@T.prim_func
def main(placeholder: T.Buffer((128, 256), "float32"), placeholder_1: T.Buffer((4916, 16, 1), "float32"), placeholder_2: T.Buffer((4916,), "int32"), placeholder_3: T.Buffer((33,), "int32"), placeholder_4: T.Buffer((128, 512), "float32"), compute: T.Buffer((128, 512), "float32")):
T.func_attr({"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True})
- for i0_outer_i1_outer_fused in T.parallel(32):
- compute_1 = T.allocate([2048], "float32", "global")
- compute_2 = T.Buffer((2048,), data=compute_1)
+ for i0_outer_i1_outer_fused in T.parallel(128):
+ compute_1 = T.allocate([512], "float32", "global")
+ compute_2 = T.Buffer((512,), data=compute_1)
for i_outer_inner in range(32):
- for i_inner_init, j_init in T.grid(4, 16):
- compute_2[i_outer_inner * 64 + i_inner_init * 16 + j_init] = T.float32(0)
- for elem_idx, i_inner, j in T.grid(placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused], 4, 16):
+ cse_var_2: T.int32 = i0_outer_i1_outer_fused // 4
+ cse_var_1: T.int32 = i_outer_inner * 16
+ compute_2[cse_var_1] = T.float32(0)
+ compute_2[cse_var_1 + 1] = T.float32(0)
+ compute_2[cse_var_1 + 2] = T.float32(0)
+ compute_2[cse_var_1 + 3] = T.float32(0)
+ compute_2[cse_var_1 + 4] = T.float32(0)
+ compute_2[cse_var_1 + 5] = T.float32(0)
+ compute_2[cse_var_1 + 6] = T.float32(0)
+ compute_2[cse_var_1 + 7] = T.float32(0)
+ compute_2[cse_var_1 + 8] = T.float32(0)
+ compute_2[cse_var_1 + 9] = T.float32(0)
+ compute_2[cse_var_1 + 10] = T.float32(0)
+ compute_2[cse_var_1 + 11] = T.float32(0)
+ compute_2[cse_var_1 + 12] = T.float32(0)
+ compute_2[cse_var_1 + 13] = T.float32(0)
+ compute_2[cse_var_1 + 14] = T.float32(0)
+ compute_2[cse_var_1 + 15] = T.float32(0)
+ for elem_idx in range(placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
placeholder_5 = T.Buffer((33,), "int32", data=placeholder_3.data)
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- placeholder_6 = T.Buffer((78656,), data=placeholder_1.data)
- placeholder_7 = T.Buffer((32768,), data=placeholder.data)
- placeholder_8 = T.Buffer((4916,), "int32", data=placeholder_2.data)
- cse_var_1: T.int32 = i_outer_inner * 64 + i_inner * 16 + j
- compute_2[cse_var_1] = compute_2[cse_var_1] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + j] * T.max(placeholder_7[i_outer_inner * 1024 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
+ placeholder_6 = T.Buffer((78656,), data=placeholder_1.data)
+ placeholder_7 = T.Buffer((32768,), data=placeholder.data)
+ placeholder_8 = T.Buffer((4916,), "int32", data=placeholder_2.data)
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ compute_2[cse_var_1] = compute_2[cse_var_1] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx]], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_3: T.int32 = cse_var_1 + 1
+ compute_2[cse_var_3] = compute_2[cse_var_3] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 1] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx]], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_4: T.int32 = cse_var_1 + 2
+ compute_2[cse_var_4] = compute_2[cse_var_4] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 2] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx]], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_5: T.int32 = cse_var_1 + 3
+ compute_2[cse_var_5] = compute_2[cse_var_5] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 3] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx]], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_6: T.int32 = cse_var_1 + 4
+ compute_2[cse_var_6] = compute_2[cse_var_6] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 256], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_7: T.int32 = cse_var_1 + 5
+ compute_2[cse_var_7] = compute_2[cse_var_7] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 1] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 256], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_8: T.int32 = cse_var_1 + 6
+ compute_2[cse_var_8] = compute_2[cse_var_8] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 2] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 256], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_9: T.int32 = cse_var_1 + 7
+ compute_2[cse_var_9] = compute_2[cse_var_9] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 3] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 256], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_10: T.int32 = cse_var_1 + 8
+ compute_2[cse_var_10] = compute_2[cse_var_10] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 512], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_11: T.int32 = cse_var_1 + 9
+ compute_2[cse_var_11] = compute_2[cse_var_11] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 1] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 512], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_12: T.int32 = cse_var_1 + 10
+ compute_2[cse_var_12] = compute_2[cse_var_12] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 2] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 512], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_13: T.int32 = cse_var_1 + 11
+ compute_2[cse_var_13] = compute_2[cse_var_13] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 3] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 512], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_14: T.int32 = cse_var_1 + 12
+ compute_2[cse_var_14] = compute_2[cse_var_14] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 768], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_15: T.int32 = cse_var_1 + 13
+ compute_2[cse_var_15] = compute_2[cse_var_15] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 1] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 768], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_16: T.int32 = cse_var_1 + 14
+ compute_2[cse_var_16] = compute_2[cse_var_16] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 2] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 768], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_17: T.int32 = cse_var_1 + 15
+ compute_2[cse_var_17] = compute_2[cse_var_17] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 3] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 768], T.float32(0))
for i0_inner in range(128):
- cse_var_2: T.int32 = i0_inner * 512 + i0_outer_i1_outer_fused * 16
+ cse_var_18: T.int32 = i0_inner * 512 + i0_outer_i1_outer_fused * 4
compute_3 = T.Buffer((65536,), data=compute.data)
placeholder_5 = T.Buffer((65536,), data=placeholder_4.data)
- compute_3[cse_var_2:cse_var_2 + 16] = T.max(compute_2[i0_inner * 16:i0_inner * 16 + 16] + placeholder_5[cse_var_2:cse_var_2 + 16], T.Broadcast(T.float32(0), 16))
+ compute_3[cse_var_18:cse_var_18 + 4] = T.max(compute_2[i0_inner * 4:i0_inner * 4 + 4] + placeholder_5[cse_var_18:cse_var_18 + 4], T.Broadcast(T.float32(0), 4))
@@ -457,7 +517,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 1.485 ms
+ Execution time of this operator: 2.434 ms
diff --git a/docs/_sources/how_to/tune_with_autotvm/sg_execution_times.rst.txt b/docs/_sources/how_to/tune_with_autotvm/sg_execution_times.rst.txt
index e185c2fdd9..0a34c85b07 100644
--- a/docs/_sources/how_to/tune_with_autotvm/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/tune_with_autotvm/sg_execution_times.rst.txt
@@ -5,16 +5,16 @@
Computation times
=================
-**00:40.527** total execution time for **how_to_tune_with_autotvm** files:
+**00:50.535** total execution time for **how_to_tune_with_autotvm** files:
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``) | 00:40.491 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``) | 00:50.501 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``) | 00:00.022 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``) | 00:00.021 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_cuda.py` (``tune_relay_cuda.py``) | 00:00.005 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_arm.py` (``tune_relay_arm.py``) | 00:00.005 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_arm.py` (``tune_relay_arm.py``) | 00:00.004 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_mobile_gpu.py` (``tune_relay_mobile_gpu.py``) | 00:00.005 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_mobile_gpu.py` (``tune_relay_mobile_gpu.py``) | 00:00.004 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt b/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt
index d35869dde7..0e246c4743 100644
--- a/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt
+++ b/docs/_sources/how_to/tune_with_autotvm/tune_conv2d_cuda.rst.txt
@@ -390,9 +390,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 2, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 64, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5478804
- No: 2 GFLOPS: 114.52/114.52 result: MeasureResult(costs=(0.0020214115454545455,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.826885461807251, timestamp=1679626739.1992018) [('tile_f', [-1, 1, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6269615
- No: 3 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 1, 128]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2739872
+ No: 2 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -514,8 +513,9 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 4, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4664196
- No: 4 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 8, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8372791
+ No: 3 GFLOPS: 157.98/157.98 result: MeasureResult(costs=(0.0014653737777777778,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.0277414321899414, timestamp=1679627934.8009439) [('tile_f', [-1, 2, 2, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2394051
+ No: 4 GFLOPS: 0.00/157.98 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -637,8 +637,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 512, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4573854
- No: 5 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 32, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 128]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4435362
+ No: 5 GFLOPS: 0.00/157.98 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -760,8 +760,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 64, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4178335
- No: 6 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 2, 32]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 256]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1540411
+ No: 6 GFLOPS: 0.00/157.98 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -883,8 +883,10 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 8, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 64]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7328010
- No: 7 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 128, 2, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6154791
+ No: 7 GFLOPS: 20.05/157.98 result: MeasureResult(costs=(0.011545547454545457,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.0361216068267822, timestamp=1679627944.4602265) [('tile_f', [-1, 1, 2, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 8, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1463583
+ No: 8 GFLOPS: 210.09/210.09 result: MeasureResult(costs=(0.0011019326483516483,), error_no=MeasureErrorNo.NO_ERROR, all_cost=6.238207817077637, timestamp=1679627945.3220174) [('tile_f', [-1, 2, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10020176
+ No: 9 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1006,9 +1008,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 4, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9260858
- No: 8 GFLOPS: 2.80/114.52 result: MeasureResult(costs=(0.08256726824999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=8.456061601638794, timestamp=1679626752.5319996) [('tile_f', [-1, 64, 2, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5821050
- No: 9 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 2, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3385287
+ No: 10 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1130,9 +1131,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 2, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 64, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2606293
- No: 10 GFLOPS: 155.35/155.35 result: MeasureResult(costs=(0.0014902013766233768,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.0830042362213135, timestamp=1679626755.7926762) [('tile_f', [-1, 8, 4, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4879015
- No: 11 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 128, 4]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7724115
+ No: 11 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1254,8 +1254,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 4, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6578462
- No: 12 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 4, 16]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5932037
+ No: 12 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1377,8 +1377,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 64, 1, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 64, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5055302
- No: 13 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 2, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7725665
+ No: 13 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1500,255 +1500,161 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 8, 16]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,342501
- No: 14 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 2, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3708513
+ No: 14 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 742, in __call__
+ yield remote, remote.load_module(os.path.split(build_result.filename)[1])
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 706, in run_through_rpc
+ costs = time_f(*args).results
+ File "/workspace/python/tvm/runtime/module.py", line 357, in evaluator
+ blob = feval(*args)
File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
+ File "tvm/_ffi/_cython/./packed_func.pxi", line 262, in tvm._ffi._cy3.core.FuncCall
+ File "tvm/_ffi/_cython/./packed_func.pxi", line 251, in tvm._ffi._cy3.core.FuncCall3
File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
+ 4: TVMFuncCall
at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
+ 3: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
+ at ../include/tvm/runtime/packed_func.h:1217
+ 2: tvm::runtime::RPCWrappedFunc::operator()(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
+ at ../src/runtime/rpc/rpc_module.cc:129
+ 1: tvm::runtime::RPCClientSession::CallFunc(void*, TVMValue const*, int const*, int, std::function<void (tvm::runtime::TVMArgs)> const&)
+ at ../src/runtime/rpc/rpc_endpoint.cc:1012
+ 0: tvm::runtime::RPCEndpoint::CallFunc(void*, TVMValue const*, int const*, int, std::function<void (tvm::runtime::TVMArgs)>)
+ at ../src/runtime/rpc/rpc_endpoint.cc:804
+ File "../src/runtime/rpc/rpc_endpoint.cc", line 804
+ TVMError:
+ ---------------------------------------------------------------
+ An error occurred during the execution of TVM.
+ For more information, please see: https://tvm.apache.org/docs/errors.html
+ ---------------------------------------------------------------
+ Check failed: (code == RPCCode::kReturn) is false: code=kShutdown
+
+ During handling of the above exception, another exception occurred:
Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 1, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 256, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5487563
- No: 15 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
- File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
- File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 706, in run_through_rpc
+ costs = time_f(*args).results
+ File "/usr/lib/python3.7/contextlib.py", line 130, in __exit__
+ self.gen.throw(type, value, traceback)
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 746, in __call__
+ remote.remove(build_result.filename)
+ File "/workspace/python/tvm/rpc/client.py", line 144, in remove
+ self._remote_funcs["remove"] = self.get_function("tvm.rpc.server.remove")
+ File "/workspace/python/tvm/rpc/client.py", line 72, in get_function
+ return self._sess.get_function(name)
+ File "/workspace/python/tvm/runtime/module.py", line 171, in get_function
+ self.handle, c_str(name), ctypes.c_int(query_imports), ctypes.byref(ret_handle)
+ File "/workspace/python/tvm/_ffi/base.py", line 348, in check_call
+ raise get_last_ffi_error()
tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
+ 52: 0xffffffffffffffff
+ 51: _start
+ 50: __libc_start_main
+ 49: _Py_UnixMain
+ 48: 0x0000000000650da0
+ 47: 0x0000000000650afa
+ 46: _PyFunction_FastCallDict
+ 45: _PyEval_EvalCodeWithName
+ 44: _PyEval_EvalFrameDefault
+ 43: _PyFunction_FastCallKeywords
+ 42: _PyEval_EvalCodeWithName
+ 41: _PyEval_EvalFrameDefault
+ 40: _PyMethodDef_RawFastCallKeywords
+ 39: 0x0000000000546369
+ 38: _PyEval_EvalCodeWithName
+ 37: _PyEval_EvalFrameDefault
+ 36: _PyFunction_FastCallKeywords
+ 35: _PyEval_EvalCodeWithName
+ 34: _PyEval_EvalFrameDefault
+ 33: _PyFunction_FastCallDict
+ 32: _PyEval_EvalCodeWithName
+ 31: _PyEval_EvalFrameDefault
+ 30: _PyObject_FastCallDict
+ 29: 0x00000000004c06e1
+ 28: _PyFunction_FastCallDict
+ 27: _PyEval_EvalFrameDefault
+ 26: _PyMethodDescr_FastCallKeywords
+ 25: 0x00000000005dcb58
+ 24: 0x00000000005dc83f
+ 23: 0x00000000004ba127
+ 22: _PyEval_EvalFrameDefault
+ 21: _PyFunction_FastCallKeywords
+ 20: _PyEval_EvalFrameDefault
+ 19: _PyFunction_FastCallKeywords
+ 18: _PyEval_EvalFrameDefault
+ 17: _PyFunction_FastCallKeywords
+ 16: _PyEval_EvalCodeWithName
+ 15: _PyEval_EvalFrameDefault
+ 14: 0x0000000000537c30
+ 13: _PyObject_FastCallKeywords
+ 12: 0x00007f8eb7416fa2
+ 11: _ctypes_callproc
+ 10: ffi_call
+ 9: ffi_call_unix64
+ 8: TVMModGetFunction
+ at ../src/runtime/c_runtime_api.cc:408
+ 7: tvm::runtime::ModuleNode::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool)
+ at ../src/runtime/module.cc:66
+ 6: tvm::runtime::RPCModuleNode::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, tvm::runtime::ObjectPtr<tvm::runtime::Object> const&)
+ at ../src/runtime/rpc/rpc_module.cc:185
+ 5: tvm::runtime::RPCClientSession::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
+ at ../src/runtime/rpc/rpc_endpoint.cc:1007
+ 4: tvm::runtime::TVMRetValue tvm::runtime::RPCEndpoint::SysCallRemote<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&>(tvm::runtime::RPCCode, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
+ at ../src/runtime/rpc/rpc_endpoint.h:223
+ 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&>(int&&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) const
at ../include/tvm/runtime/packed_func.h:1621
2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
at ../include/tvm/runtime/packed_func.h:1217
1: Call
at ../include/tvm/runtime/packed_func.h:1213
0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
+ at ../src/runtime/rpc/rpc_endpoint.cc:684
+ File "../src/runtime/rpc/rpc_endpoint.cc", line 684
+ TVMError:
+ ---------------------------------------------------------------
+ An error occurred during the execution of TVM.
+ For more information, please see: https://tvm.apache.org/docs/errors.html
+ ---------------------------------------------------------------
+ Check failed: (code == RPCCode::kReturn) is false: code=1
Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 2, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1293394
- No: 16 GFLOPS: 259.48/259.48 result: MeasureResult(costs=(0.0008921655081967214,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.353220224380493, timestamp=1679626758.422771) [('tile_f', [-1, 1, 32, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9729540
- No: 17 GFLOPS: 0.00/259.48 result: Traceback (most recent call last):
+ 52: 0xffffffffffffffff
+ 51: _start
+ 50: __libc_start_main
+ 49: _Py_UnixMain
+ 48: 0x0000000000650da0
+ 47: 0x0000000000650afa
+ 46: _PyFunction_FastCallDict
+ 45: _PyEval_EvalCodeWithName
+ 44: _PyEval_EvalFrameDefault
+ 43: _PyFunction_FastCallKeywords
+ 42: _PyEval_EvalCodeWithName
+ 41: _PyEval_EvalFrameDefault
+ 40: _PyMethodDef_RawFastCallKeywords
+ 39: 0x0000000000546369
+ 38: _PyEval_EvalCodeWithName
+ 37: _PyEval_EvalFrameDefault
+ 36: _PyFunction_FastCallKeywords
+ 35: _PyEval_EvalCodeWithName
+ 34: _PyEval_EvalFrameDefault
+ 33: _PyFunction_FastCallDict
+ 32: _PyEval_EvalCodeWithName
+ 31: _PyEval_EvalFrameDefault
+ 30: _PyObject_FastCallDict
+ 29: 0x00000000004c06e1
+ 28: _PyFunction_FastCallDict
+ 27: _PyEval_EvalFrameDefault
+ 26: _PyMethodDescr_FastCallKeywords
+ 25: 0x00000000005dcb58
+ 24: 0x00000000005dc83f
+ 23: 0x00000000004ba127
+ 22: _PyEval_EvalFrameDefault
+ 21: _PyFunction_FastCallKeywords
+ 20: _PyEval_EvalFrameDefault
+ 19: _PyFunction_FastCall [('tile_f', [-1, 4, 1, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5909586
+ No: 15 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1870,8 +1776,9 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 2, 16]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 16, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9016650
- No: 18 GFLOPS: 0.00/259.48 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 16, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,660600
+ No: 16 GFLOPS: 1.07/210.09 result: MeasureResult(costs=(0.21648690775,), error_no=MeasureErrorNo.NO_ERROR, all_cost=5.180893421173096, timestamp=1679627955.7687864) [('tile_f', [-1, 32, 1, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,201001
+ No: 17 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1993,8 +1900,8 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 128, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,53302
- No: 19 GFLOPS: 0.00/259.48 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 16, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 64, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2052759
+ No: 18 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2116,8 +2023,9 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 1, 128]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 16, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,82932
- No: 20 GFLOPS: 0.00/259.48 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 1, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 128, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9705660
+ No: 19 GFLOPS: 10.49/210.09 result: MeasureResult(costs=(0.022066975833333335,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.2342774868011475, timestamp=1679627959.2058794) [('tile_f', [-1, 2, 1, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,8908206
+ No: 20 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2239,7 +2147,7 @@ for this template
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
- tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 64]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9460496
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 64, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 128]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5019615
@@ -2294,9 +2202,9 @@ and measure running time.
Finish loading 20 records
Best config:
- [('tile_f', [-1, 1, 32, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9729540
+ [('tile_f', [-1, 2, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10020176
Finish loading 20 records
- Time cost of this operator: 0.001312
+ Time cost of this operator: 0.001490
diff --git a/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt b/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt
index 5aec556c6f..def9a73054 100644
--- a/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/micro_autotune.rst.txt
@@ -360,10 +360,10 @@ Timing the untuned program
########## Build without Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
- tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 314.2 98.697 (1, 2, 10, 10, 3) 2 1 [314.2]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.151 0.99 (1, 6, 10, 10) 1 1 [3.151]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.996 0.313 (1, 1, 10, 10, 3) 1 1 [0.996]
- Total_time - 318.347 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 319.3 98.755 (1, 2, 10, 10, 3) 2 1 [319.3]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.057 0.945 (1, 6, 10, 10) 1 1 [3.057]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.968 0.299 (1, 1, 10, 10, 3) 1 1 [0.968]
+ Total_time - 323.325 - - - - -
@@ -428,10 +428,10 @@ Timing the tuned program
########## Build with Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
- tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 136.8 98.078 (1, 6, 10, 10, 1) 2 1 [136.8]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.821 1.305 (1, 6, 10, 10) 1 1 [1.821]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.86 0.617 (1, 3, 10, 10, 1) 1 1 [0.86]
- Total_time - 139.481 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 173.8 97.599 (1, 6, 10, 10, 1) 2 1 [173.8]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 2.708 1.521 (1, 6, 10, 10) 1 1 [2.708]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 1.568 0.88 (1, 1, 10, 10, 3) 1 1 [1.568]
+ Total_time - 178.076 - - - - -
@@ -439,7 +439,7 @@ Timing the tuned program
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 28.427 seconds)
+ **Total running time of the script:** ( 1 minutes 21.031 seconds)
.. _sphx_glr_download_how_to_work_with_microtvm_micro_autotune.py:
diff --git a/docs/_sources/how_to/work_with_microtvm/micro_pytorch.rst.txt b/docs/_sources/how_to/work_with_microtvm/micro_pytorch.rst.txt
index b0837484da..c60e1d9129 100644
--- a/docs/_sources/how_to/work_with_microtvm/micro_pytorch.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/micro_pytorch.rst.txt
@@ -118,7 +118,7 @@ download a cat image and preprocess it to use as the model input.
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torch/ao/quantization/utils.py:281: UserWarning: must run observer before calling calculate_qparams. Returning default values.
"must run observer before calling calculate_qparams. " +
Downloading: "https://download.pytorch.org/models/quantized/mobilenet_v2_qnnpack_37f702c5.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2_qnnpack_37f702c5.pth
-
0%| | 0.00/3.42M [00:00<?, ?B/s]
61%|###### | 2.09M/3.42M [00:00<00:00, 16.8MB/s]
100%|##########| 3.42M/3.42M [00:00<00:00, 26.4MB/s]
+
0%| | 0.00/3.42M [00:00<?, ?B/s]
61%|###### | 2.09M/3.42M [00:00<00:00, 9.77MB/s]
100%|##########| 3.42M/3.42M [00:00<00:00, 15.7MB/s]
/workspace/python/tvm/relay/frontend/pytorch_utils.py:47: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
return LooseVersion(torch_ver) > ver
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/setuptools/_distutils/version.py:346: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
@@ -324,7 +324,7 @@ Look up prediction top 1 index in 1000 class synset.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 22.970 seconds)
+ **Total running time of the script:** ( 1 minutes 17.499 seconds)
.. _sphx_glr_download_how_to_work_with_microtvm_micro_pytorch.py:
diff --git a/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt b/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
index 69cb8c3463..e0bde23a48 100644
--- a/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/micro_train.rst.txt
@@ -217,7 +217,7 @@ take about **2 minutes** to download the Stanford Cars, while COCO 2017 validati
.. code-block:: none
- '/tmp/tmpr8om6t_s/images/random'
+ '/tmp/tmp9g460_h3/images/random'
@@ -308,7 +308,7 @@ objects to other stuff? We can display some examples from our datasets using ``m
.. image-sg:: /how_to/work_with_microtvm/images/sphx_glr_micro_train_001.png
- :alt: [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0]
+ :alt: [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0]
:srcset: /how_to/work_with_microtvm/images/sphx_glr_micro_train_001.png
:class: sphx-glr-single-img
@@ -317,8 +317,8 @@ objects to other stuff? We can display some examples from our datasets using ``m
.. code-block:: none
- /tmp/tmpr8om6t_s/images/target contains 8144 images
- /tmp/tmpr8om6t_s/images/random contains 5000 images
+ /tmp/tmp9g460_h3/images/target contains 8144 images
+ /tmp/tmp9g460_h3/images/random contains 5000 images
@@ -493,13 +493,13 @@ the time on our validation set).
.. code-block:: none
Epoch 1/3
- 328/328 - 42s - loss: 0.2768 - accuracy: 0.9157 - val_loss: 0.1518 - val_accuracy: 0.9517 - 42s/epoch - 128ms/step
+ 328/328 - 40s - loss: 0.2110 - accuracy: 0.9230 - val_loss: 0.1209 - val_accuracy: 0.9577 - 40s/epoch - 122ms/step
Epoch 2/3
- 328/328 - 35s - loss: 0.1149 - accuracy: 0.9561 - val_loss: 0.1394 - val_accuracy: 0.9513 - 35s/epoch - 108ms/step
+ 328/328 - 35s - loss: 0.0927 - accuracy: 0.9640 - val_loss: 0.1028 - val_accuracy: 0.9630 - 35s/epoch - 106ms/step
Epoch 3/3
- 328/328 - 35s - loss: 0.0679 - accuracy: 0.9748 - val_loss: 0.1042 - val_accuracy: 0.9653 - 35s/epoch - 107ms/step
+ 328/328 - 35s - loss: 0.0615 - accuracy: 0.9788 - val_loss: 0.1040 - val_accuracy: 0.9675 - 35s/epoch - 105ms/step
- <keras.callbacks.History object at 0x7f228f798990>
+ <keras.callbacks.History object at 0x7fd033351990>
@@ -860,7 +860,7 @@ Arduino tutorial for how to do that `on GitHub <https://github.com/guberti/tvm-a
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 4 minutes 31.483 seconds)
+ **Total running time of the script:** ( 4 minutes 25.665 seconds)
.. _sphx_glr_download_how_to_work_with_microtvm_micro_train.py:
diff --git a/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt b/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
index 0fcd9313a0..188d64395a 100644
--- a/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/work_with_microtvm/sg_execution_times.rst.txt
@@ -5,20 +5,20 @@
Computation times
=================
-**07:50.419** total execution time for **how_to_work_with_microtvm** files:
+**07:30.160** total execution time for **how_to_work_with_microtvm** files:
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:31.483 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:25.665 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 01:28.427 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 01:21.031 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_pytorch.py` (``micro_pytorch.py``) | 01:22.970 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_pytorch.py` (``micro_pytorch.py``) | 01:17.499 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_aot.py` (``micro_aot.py``) | 00:10.957 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_aot.py` (``micro_aot.py``) | 00:10.356 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_custom_ide.py` (``micro_custom_ide.py``) | 00:08.651 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_custom_ide.py` (``micro_custom_ide.py``) | 00:08.011 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:07.931 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:07.598 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_microtvm_micro_ethosu.py` (``micro_ethosu.py``) | 00:00.000 | 0.0 MB |
+-----------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt b/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
index 30d0e8430c..415a59375e 100644
--- a/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/work_with_relay/sg_execution_times.rst.txt
@@ -5,14 +5,14 @@
Computation times
=================
-**00:38.525** total execution time for **how_to_work_with_relay** files:
+**00:36.777** total execution time for **how_to_work_with_relay** files:
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_pipeline_executor.py` (``using_pipeline_executor.py``) | 00:33.774 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_pipeline_executor.py` (``using_pipeline_executor.py``) | 00:32.040 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:03.004 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:02.928 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.740 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.802 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``) | 00:00.007 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_relay_viz.py` (``using_relay_viz.py``) | 00:00.006 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt b/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
index bd50c0fa5b..3907ad5284 100644
--- a/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/intrin_math.rst.txt
@@ -264,7 +264,7 @@ The following example customizes CUDA lowering rule for :code:`exp`.
.. code-block:: none
- <function my_cuda_math_rule at 0x7f1edd6a7710>
+ <function my_cuda_math_rule at 0x7fced0208ef0>
diff --git a/docs/_sources/how_to/work_with_schedules/sg_execution_times.rst.txt b/docs/_sources/how_to/work_with_schedules/sg_execution_times.rst.txt
index 8c77395aaa..8b3edeaaab 100644
--- a/docs/_sources/how_to/work_with_schedules/sg_execution_times.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/sg_execution_times.rst.txt
@@ -5,22 +5,22 @@
Computation times
=================
-**00:08.721** total execution time for **how_to_work_with_schedules** files:
+**00:08.728** total execution time for **how_to_work_with_schedules** files:
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:05.745 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:05.962 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:01.410 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:01.266 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.642 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.619 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.631 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.603 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.138 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.130 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``) | 00:00.066 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tedd.py` (``tedd.py``) | 00:00.063 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.058 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_schedule_primitives.py` (``schedule_primitives.py``) | 00:00.055 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tuple_inputs.py` (``tuple_inputs.py``) | 00:00.032 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tuple_inputs.py` (``tuple_inputs.py``) | 00:00.029 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
index fc2cb8517c..51a7e19061 100644
--- a/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/autotvm/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
Computation times
=================
-**00:33.229** total execution time for **topic_vta_tutorials_autotvm** files:
+**00:31.580** total execution time for **topic_vta_tutorials_autotvm** files:
+---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:33.222 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:31.574 | 0.0 MB |
+---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``) | 00:00.007 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_alu_vta.py` (``tune_alu_vta.py``) | 00:00.006 | 0.0 MB |
+---------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
index b5ea67b875..c41b7a6e80 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_classification.rst.txt
@@ -293,7 +293,7 @@ The compilation steps are:
DeprecationWarning,
/workspace/vta/tutorials/frontend/deploy_classification.py:213: DeprecationWarning: legacy graph executor behavior of producing json / lib / params will be removed in the next release. Please see documents of tvm.contrib.graph_executor.GraphModule for the new recommended usage.
relay_prog, target=tvm.target.Target(target, host=env.target_host), params=params
- resnet18_v1 inference graph built in 36.54s!
+ resnet18_v1 inference graph built in 33.36s!
diff --git a/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt b/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
index 0e5be90dc4..a73c6c91f7 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt
@@ -337,7 +337,7 @@ The compilation steps are:
/workspace/python/tvm/relay/build_module.py:348: DeprecationWarning: Please use input parameter mod (tvm.IRModule) instead of deprecated parameter mod (tvm.relay.function.Function)
DeprecationWarning,
- yolov3-tiny inference graph built in 24.42s!
+ yolov3-tiny inference graph built in 22.47s!
diff --git a/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt
index 036defa449..6340c3bbf7 100644
--- a/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/frontend/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
Computation times
=================
-**01:45.411** total execution time for **topic_vta_tutorials_frontend** files:
+**01:39.526** total execution time for **topic_vta_tutorials_frontend** files:
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:53.678 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:50.187 | 0.0 MB |
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:51.733 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:49.340 | 0.0 MB |
+------------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
index 32220e4956..1da6e639c5 100644
--- a/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/optimize/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
Computation times
=================
-**00:03.242** total execution time for **topic_vta_tutorials_optimize** files:
+**00:03.157** total execution time for **topic_vta_tutorials_optimize** files:
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.729 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.687 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.513 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.470 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt b/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
index 3121de42d5..8eaace8809 100644
--- a/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
+++ b/docs/_sources/topic/vta/tutorials/sg_execution_times.rst.txt
@@ -5,10 +5,10 @@
Computation times
=================
-**00:00.884** total execution time for **topic_vta_tutorials** files:
+**00:00.804** total execution time for **topic_vta_tutorials** files:
+---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.457 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.414 | 0.0 MB |
+---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.427 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.389 | 0.0 MB |
+---------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
index 2213e38b76..56d2855f4a 100644
--- a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
@@ -318,7 +318,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 100.398 ms
+ Execution time of this operator: 95.445 ms
@@ -434,7 +434,7 @@ operations.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 24.282 seconds)
+ **Total running time of the script:** ( 1 minutes 27.041 seconds)
.. _sphx_glr_download_tutorial_auto_scheduler_matmul_x86.py:
diff --git a/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt b/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
index 2598c8fed7..2e5007448c 100644
--- a/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_matmul_x86.rst.txt
@@ -454,16 +454,16 @@ reduce variance, we take 5 measurements and average them.
waiting for device...
device available
Get devices for measurement successfully!
- No: 1 GFLOPS: 10.03/10.03 result: MeasureResult(costs=(0.026752570600000004,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.7407488822937012, timestamp=1679625052.3152878) [('tile_y', [-1, 16]), ('tile_x', [-1, 32])],None,54
- No: 2 GFLOPS: 11.13/11.13 result: MeasureResult(costs=(0.0241221214,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.649803638458252, timestamp=1679625052.971663) [('tile_y', [-1, 128]), ('tile_x', [-1, 512])],None,97
- No: 3 GFLOPS: 11.54/11.54 result: MeasureResult(costs=(0.0232565404,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6532163619995117, timestamp=1679625054.9199133) [('tile_y', [-1, 8]), ('tile_x', [-1, 256])],None,83
- No: 4 GFLOPS: 8.93/11.54 result: MeasureResult(costs=(0.0300584328,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.7522788047790527, timestamp=1679625056.9847775) [('tile_y', [-1, 256]), ('tile_x', [-1, 256])],None,88
- No: 5 GFLOPS: 12.50/12.50 result: MeasureResult(costs=(0.02148152,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6188480854034424, timestamp=1679625057.795037) [('tile_y', [-1, 32]), ('tile_x', [-1, 128])],None,75
- No: 6 GFLOPS: 7.39/12.50 result: MeasureResult(costs=(0.0363478626,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.9304215908050537, timestamp=1679625059.9495907) [('tile_y', [-1, 1]), ('tile_x', [-1, 64])],None,60
- No: 7 GFLOPS: 0.51/12.50 result: MeasureResult(costs=(0.5292683636,), error_no=MeasureErrorNo.NO_ERROR, all_cost=8.71769094467163, timestamp=1679625068.6794431) [('tile_y', [-1, 128]), ('tile_x', [-1, 1])],None,7
- No: 8 GFLOPS: 2.10/12.50 result: MeasureResult(costs=(0.1277711432,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.311182975769043, timestamp=1679625070.9895515) [('tile_y', [-1, 128]), ('tile_x', [-1, 4])],None,27
- No: 9 GFLOPS: 10.56/12.50 result: MeasureResult(costs=(0.025417857,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6243619918823242, timestamp=1679625071.7493625) [('tile_y', [-1, 2]), ('tile_x', [-1, 512])],None,91
- No: 10 GFLOPS: 11.04/12.50 result: MeasureResult(costs=(0.024324382999999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6199319362640381, timestamp=1679625072.4000835) [('tile_y', [-1, 128]), ('tile_x', [-1, 32])],None,57
+ No: 1 GFLOPS: 11.03/11.03 result: MeasureResult(costs=(0.024329650799999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6992037296295166, timestamp=1679626290.7543154) [('tile_y', [-1, 16]), ('tile_x', [-1, 256])],None,84
+ No: 2 GFLOPS: 0.50/11.03 result: MeasureResult(costs=(0.5394531057999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=8.896685361862183, timestamp=1679626299.6634266) [('tile_y', [-1, 32]), ('tile_x', [-1, 1])],None,5
+ No: 3 GFLOPS: 10.30/11.03 result: MeasureResult(costs=(0.0260593904,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6705293655395508, timestamp=1679626301.5776997) [('tile_y', [-1, 2]), ('tile_x', [-1, 64])],None,61
+ No: 4 GFLOPS: 11.62/11.62 result: MeasureResult(costs=(0.023098842799999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6169741153717041, timestamp=1679626302.2112644) [('tile_y', [-1, 4]), ('tile_x', [-1, 512])],None,92
+ No: 5 GFLOPS: 11.49/11.62 result: MeasureResult(costs=(0.0233601186,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6366961002349854, timestamp=1679626302.9618816) [('tile_y', [-1, 128]), ('tile_x', [-1, 32])],None,57
+ No: 6 GFLOPS: 9.85/11.62 result: MeasureResult(costs=(0.0272440226,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6859402656555176, timestamp=1679626304.9166765) [('tile_y', [-1, 512]), ('tile_x', [-1, 256])],None,89
+ No: 7 GFLOPS: 3.10/11.62 result: MeasureResult(costs=(0.08649252040000001,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.6283063888549805, timestamp=1679626307.7995849) [('tile_y', [-1, 128]), ('tile_x', [-1, 8])],None,37
+ No: 8 GFLOPS: 11.66/11.66 result: MeasureResult(costs=(0.0230121304,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6313371658325195, timestamp=1679626308.4286308) [('tile_y', [-1, 4]), ('tile_x', [-1, 256])],None,82
+ No: 9 GFLOPS: 2.37/11.66 result: MeasureResult(costs=(0.1133947778,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.0219972133636475, timestamp=1679626310.6312978) [('tile_y', [-1, 2]), ('tile_x', [-1, 2])],None,11
+ No: 10 GFLOPS: 13.16/13.16 result: MeasureResult(costs=(0.020398058,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6012001037597656, timestamp=1679626311.219927) [('tile_y', [-1, 128]), ('tile_x', [-1, 128])],None,77
diff --git a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
index b8c7086e77..5135bdb278 100644
--- a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
@@ -311,7 +311,7 @@ standard deviation.
.. code-block:: none
- {'mean': 512.8374819899979, 'median': 513.379893900003, 'std': 2.048731521479577}
+ {'mean': 515.2812953900003, 'median': 514.8744986499992, 'std': 2.452215899798013}
@@ -545,29 +545,31 @@ the tuning data to.
.. code-block:: none
-
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 1/25] Current/Best: 12.64/ 17.36 GFLOPS | Progress: (4/20) | 12.33 s
[Task 1/25] Current/Best: 15.05/ 17.36 GFLOPS | Progress: (8/20) | 16.51 s
[Task 1/25] Current/Best: 6.72/ 17.36 GFLOPS | Progress: (12/20) | 21.17 s
[Task 1/25] Current/Best: 8.38/ 17.36 GFLOPS | Progress: (16/20) | 23.68 s
[Task 1/25] Current/Best: 15.33/ 17.36 GFLOPS | Progress: (20/20) | 26.12 s Done.
-
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 12.04/ 12.04 GFLOPS | Progress: (4/20) | 4.42 s
[Task 2/25] Current/Best: 13.62/ 18.02 GFLOPS | Progress: (8/20) | 6.62 s
[Task 2/25] Current/Best: 9.94/ 18.02 GFLOPS | Progress: (12/20) | 8.40 s
[Task 2/25] Current/Best: 17.13/ 18.02 GFLOPS | Progress: (16/20) | 10.38 s
[Task 2/25] Current/Best: 12.44/ 19.88 GFLOPS | Progress: (20/20) | 12.05 s Done.
-
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 11.38/ 13.39 GFLOPS | Progress: (4/20) | 5.51 s
[Task 3/25] Current/Best: 20.78/ 20.78 GFLOPS | Progress: (8/20) | 7.93 s
[Task 3/25] Current/Best: 3.13/ 24.29 GFLOPS | Progress: (12/20) | 10.67 s
[Task 3/25] Current/Best: 10.26/ 24.29 GFLOPS | Progress: (16/20) | 12.98 s
[Task 3/25] Current/Best: 10.09/ 24.29 GFLOPS | Progress: (20/20) | 15.10 s Done.
-
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 11.76/ 18.70 GFLOPS | Progress: (4/20) | 5.93 s
[Task 4/25] Current/Best: 8.78/ 18.80 GFLOPS | Progress: (8/20) | 9.05 s
[Task 4/25] Current/Best: 15.65/ 18.80 GFLOPS | Progress: (12/20) | 11.83 s
[Task 4/25] Current/Best: 11.77/ 18.80 GFLOPS | Progress: (16/20) | 14.86 s
[Task 4/25] Current/Best: 7.81/ 18.80 GFLOPS | Progress: (20/20) | 17.00 s Done.
-
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 3.18/ 16.10 GFLOPS | Progress: (4/20) | 5.51 s
[Task 5/25] Current/Best: 10.22/ 16.10 GFLOPS | Progress: (8/20) | 8.05 s
[Task 5/25] Current/Best: 10.38/ 17.98 GFLOPS | Progress: (12/20) | 11.51 s
[Task 5/25] Current/Best: 12.33/ 19.56 GFLOPS | Progress: (16/20) | 13.61 s
[Task 5/25] Current/Best: 14.14/ 19.56 GFLOPS | Progress: (20/20) | 15.37 s Done.
-
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 11.86/ 11.86 GFLOPS | Progress: (4/20) | 5.98 s
[Task 6/25] Current/Best: 17.89/ 17.89 GFLOPS | Progress: (8/20) | 9.24 s
[Task 6/25] Current/Best: 11.39/ 20.41 GFLOPS | Progress: (12/20) | 11.54 s
[Task 6/25] Current/Best: 15.14/ 20.41 GFLOPS | Progress: (16/20) | 15.87 s
[Task 6/25] Current/Best: 4.72/ 20.41 GFLOPS | Progress: (20/20) | 18.98 s Done.
-
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 9.62/ 20.97 GFLOPS | Progress: (4/20) | 5.07 s
[Task 7/25] Current/Best: 15.02/ 20.97 GFLOPS | Progress: (8/20) | 7.08 s
[Task 7/25] Current/Best: 11.07/ 20.97 GFLOPS | Progress: (12/20) | 9.93 s
[Task 7/25] Current/Best: 17.47/ 20.97 GFLOPS | Progress: (16/20) | 12.62 s
[Task 7/25] Current/Best: 15.51/ 20.97 GFLOPS | Progress: (20/20) | 15.02 s Done.
-
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 8.60/ 17.59 GFLOPS | Progress: (4/20) | 7.00 s
[Task 8/25] Current/Best: 8.21/ 17.59 GFLOPS | Progress: (8/20) | 11.96 s
[Task 8/25] Current/Best: 19.86/ 19.86 GFLOPS | Progress: (12/20) | 14.00 s
[Task 8/25] Current/Best: 12.89/ 19.86 GFLOPS | Progress: (16/20) | 18.13 s
[Task 8/25] Current/Best: 16.33/ 19.86 GFLOPS | Progress: (20/20) | 22.00 s Done.
-
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 8.56/ 20.82 GFLOPS | Progress: (4/20) | 14.22 s
[Task 9/25] Current/Best: 9.08/ 20.82 GFLOPS | Progress: (8/20) | 17.92 s
[Task 9/25] Current/Best: 6.01/ 20.82 GFLOPS | Progress: (12/20) | 25.78 s
[Task 9/25] Current/Best: 21.81/ 21.81 GFLOPS | Progress: (16/20) | 29.01 s
[Task 9/25] Current/Best: 13.27/ 21.81 GFLOPS | Progress: (20/20) | 30.83 s Done.
-
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 6.10/ 20.34 GFLOPS | Progress: (4/20) | 4.79 s
[Task 10/25] Current/Best: 5.26/ 20.34 GFLOPS | Progress: (8/20) | 7.02 s
[Task 10/25] Current/Best: 12.36/ 20.34 GFLOPS | Progress: (12/20) | 9.50 s
[Task 10/25] Current/Best: 13.67/ 20.34 GFLOPS | Progress: (16/20) | 11.77 s
[Task 10/25] Current/Best: 9.19/ 20.34 GFLOPS | Progress: (20/20) | 14.96 s Done.
-
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 17.38/ 20.84 GFLOPS | Progress: (4/20) | 5.40 s
[Task 11/25] Current/Best: 13.11/ 20.84 GFLOPS | Progress: (8/20) | 8.07 s
[Task 11/25] Current/Best: 10.83/ 20.84 GFLOPS | Progress: (12/20) | 10.75 s
[Task 11/25] Current/Best: 9.73/ 20.84 GFLOPS | Progress: (16/20) | 13.39 s
[Task 11/25] Current/Best: 7.13/ 20.84 GFLOPS | Progress: (20/20) | 15.96 s Done.
-
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 11.21/ 13.36 GFLOPS | Progress: (4/20) | 6.10 s
[Task 12/25] Current/Best: 10.82/ 13.36 GFLOPS | Progress: (8/20) | 9.41 s
[Task 12/25] Current/Best: 14.81/ 18.80 GFLOPS | Progress: (12/20) | 12.50 s
[Task 12/25] Current/Best: 13.27/ 18.80 GFLOPS | Progress: (16/20) | 15.82 s
[Task 12/25] Current/Best: 12.89/ 18.80 GFLOPS | Progress: (20/20) | 20.23 s Done.
-
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 5.17/ 17.55 GFLOPS | Progress: (4/20) | 5.72 s
[Task 13/25] Current/Best: 16.75/ 17.55 GFLOPS | Progress: (8/20) | 8.35 s
[Task 13/25] Current/Best: 12.11/ 17.55 GFLOPS | Progress: (12/20) | 12.48 s
[Task 13/25] Current/Best: 1.56/ 18.01 GFLOPS | Progress: (16/20) | 16.13 s
[Task 13/25] Current/Best: 16.27/ 20.63 GFLOPS | Progress: (20/20) | 18.72 s Done.
-
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 6.03/ 18.11 GFLOPS | Progress: (4/20) | 5.48 s
[Task 14/25] Current/Best: 7.72/ 20.36 GFLOPS | Progress: (8/20) | 8.21 s
[Task 14/25] Current/Best: 17.75/ 20.36 GFLOPS | Progress: (12/20) | 10.66 s
[Task 14/25] Current/Best: 9.24/ 20.36 GFLOPS | Progress: (16/20) | 14.82 s
[Task 14/25] Current/Best: 13.29/ 20.36 GFLOPS | Progress: (20/20) | 20.06 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
-
[Task 15/25] Current/Best: 11.95/ 11.95 GFLOPS | Progress: (4/20) | 6.28 s
[Task 15/25] Current/Best: 20.45/ 20.45 GFLOPS | Progress: (8/20) | 9.17 s
[Task 15/25] Current/Best: 8.68/ 20.45 GFLOPS | Progress: (12/20) | 12.10 s
[Task 15/25] Current/Best: 9.93/ 20.45 GFLOPS | Progress: (16/20) | 14.95 s
[Task 15/25] Current/Best: 20.97/ 20.97 GFLOPS | Progress: (20/20) | 17.87 s
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 5.95/ 18.21 GFLOPS | Progress: (4/20) | 5.08 s
[Task 16/25] Current/Best: 16.04/ 21.43 GFLOPS | Progress: (8/20) | 7.12 s
[Task 16/25] Current/Best: 10.50/ 21.43 GFLOPS | Progress: (12/20) | 9.17 s
[Task 16/25] Current/Best: 20.66/ 21.43 GFLOPS | Progress: (16/20) | 12.79 s
[Task 16/25] Current/Best: 5.71/ 21.43 GFLOPS | Progress: (20/20) | 15.04 s Done.
-
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 22.18/ 22.18 GFLOPS | Progress: (4/20) | 5.22 s
[Task 17/25] Current/Best: 10.24/ 22.18 GFLOPS | Progress: (8/20) | 7.53 s
[Task 17/25] Current/Best: 9.84/ 22.18 GFLOPS | Progress: (12/20) | 10.45 s
[Task 17/25] Current/Best: 6.20/ 22.67 GFLOPS | Progress: (16/20) | 12.77 s
[Task 17/25] Current/Best: 9.48/ 22.67 GFLOPS | Progress: (20/20) | 15.75 s Done.
-
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 9.36/ 13.54 GFLOPS | Progress: (4/20) | 5.25 s
[Task 18/25] Current/Best: 15.16/ 17.87 GFLOPS | Progress: (8/20) | 7.93 s
[Task 18/25] Current/Best: 6.30/ 18.81 GFLOPS | Progress: (12/20) | 10.06 s
[Task 18/25] Current/Best: 6.68/ 19.80 GFLOPS | Progress: (16/20) | 12.11 s
[Task 18/25] Current/Best: 12.52/ 19.80 GFLOPS | Progress: (20/20) | 15.23 s Done.
-
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 11.68/ 14.31 GFLOPS | Progress: (4/20) | 6.02 s
[Task 19/25] Current/Best: 3.08/ 21.42 GFLOPS | Progress: (8/20) | 9.81 s
[Task 19/25] Current/Best: 9.99/ 22.10 GFLOPS | Progress: (12/20) | 12.62 s
[Task 19/25] Current/Best: 3.08/ 22.10 GFLOPS | Progress: (16/20) | 16.86 s
[Task 19/25] Current/Best: 9.96/ 22.10 GFLOPS | Progress: (20/20) | 22.38 s Done.
-
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 9.02/ 11.31 GFLOPS | Progress: (4/20) | 8.38 s
[Task 20/25] Current/Best: 17.26/ 17.26 GFLOPS | Progress: (8/20) | 11.16 s
[Task 20/25] Current/Best: 10.68/ 17.26 GFLOPS | Progress: (12/20) | 13.17 s
[Task 20/25] Current/Best: 10.73/ 17.26 GFLOPS | Progress: (16/20) | 16.87 s
[Task 20/25] Current/Best: 11.54/ 19.17 GFLOPS | Progress: (20/20) | 20.02 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 8.83/ 17.39 GFLOPS | Progress: (4/20) | 10.61 s
[Task 21/25] Current/Best: 15.93/ 22.25 GFLOPS | Progress: (8/20) | 12.39 s
[Task 21/25] Current/Best: 19.25/ 22.25 GFLOPS | Progress: (12/20) | 14.75 s
[Task 21/25] Current/Best: 5.21/ 22.25 GFLOPS | Progress: (16/20) | 20.06 s
[Task 21/25] Current/Best: 10.65/ 22.25 GFLOPS | Progress: (20
/20) | 23.05 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 12.40/ 17.15 GFLOPS | Progress: (4/20) | 4.95 s
[Task 22/25] Current/Best: 18.65/ 22.78 GFLOPS | Progress: (8/20) | 7.13 s
[Task 22/25] Current/Best: 4.72/ 22.78 GFLOPS | Progress: (12/20) | 9.31 s
[Task 22/25] Current/Best: 10.51/ 22.78 GFLOPS | Progress: (16/20) | 12.18 s
[Task 22/25] Current/Best: 10.55/ 22.78 GFLOPS | Progress: (20/20) | 15.28 s Done.
-
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 2.39/ 20.55 GFLOPS | Progress: (4/20) | 6.99 s
[Task 23/25] Current/Best: 1.55/ 23.89 GFLOPS | Progress: (8/20) | 11.92 s
[Task 23/25] Current/Best: 10.25/ 23.89 GFLOPS | Progress: (12/20) | 17.32 s
[Task 23/25] Current/Best: 22.59/ 23.89 GFLOPS | Progress: (16/20) | 22.92 s
[Task 23/25] Current/Best: 3.08/ 23.89 GFLOPS | Progress: (20/20) | 26.33 s Done.
-
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 1.55/ 6.03 GFLOPS | Progress: (4/20) | 13.87 s
[Task 24/25] Current/Best: 3.61/ 6.03 GFLOPS | Progress: (8/20) | 16.97 s
[Task 24/25] Current/Best: 3.82/ 7.11 GFLOPS | Progress: (12/20) | 18.60 s
[Task 24/25] Current/Best: 6.77/ 10.01 GFLOPS | Progress: (16/20) | 25.62 s
[Task 24/25] Current/Best: 3.64/ 10.01 GFLOPS | Progress: (20/20) | 36.61 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
+
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 1/25] Current/Best: 22.62/ 22.88 GFLOPS | Progress: (4/20) | 11.23 s
[Task 1/25] Current/Best: 11.36/ 22.88 GFLOPS | Progress: (8/20) | 15.63 s
[Task 1/25] Current/Best: 22.91/ 22.91 GFLOPS | Progress: (12/20) | 18.05 s
[Task 1/25] Current/Best: 17.58/ 22.91 GFLOPS | Progress: (16/20) | 20.01 s
[Task 1/25] Current/Best: 14.22/ 22.91 GFLOPS | Progress: (20/20) | 22.34 s Done.
+
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 3.51/ 14.80 GFLOPS | Progress: (4/20) | 4.71 s
[Task 2/25] Current/Best: 2.10/ 18.23 GFLOPS | Progress: (8/20) | 6.24 s
[Task 2/25] Current/Best: 15.88/ 18.23 GFLOPS | Progress: (12/20) | 7.83 s
[Task 2/25] Current/Best: 15.42/ 18.23 GFLOPS | Progress: (16/20) | 9.59 s
[Task 2/25] Current/Best: 12.65/ 18.23 GFLOPS | Progress: (20/20) | 12.02 s Done.
+
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 21.09/ 21.09 GFLOPS | Progress: (4/20) | 4.93 s
[Task 3/25] Current/Best: 9.19/ 21.09 GFLOPS | Progress: (8/20) | 8.31 s
[Task 3/25] Current/Best: 19.37/ 21.09 GFLOPS | Progress: (12/20) | 10.34 s
[Task 3/25] Current/Best: 12.23/ 21.09 GFLOPS | Progress: (16/20) | 12.87 s
[Task 3/25] Current/Best: 6.39/ 23.13 GFLOPS | Progress: (20/20) | 16.36 s Done.
+
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 7.68/ 19.95 GFLOPS | Progress: (4/20) | 4.97 s
[Task 4/25] Current/Best: 11.49/ 19.95 GFLOPS | Progress: (8/20) | 8.51 s
[Task 4/25] Current/Best: 6.03/ 19.95 GFLOPS | Progress: (12/20) | 11.48 s
[Task 4/25] Current/Best: 6.10/ 20.55 GFLOPS | Progress: (16/20) | 17.14 s
[Task 4/25] Current/Best: 11.91/ 20.55 GFLOPS | Progress: (20/20) | 19.29 s Done.
+
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 3.26/ 14.32 GFLOPS | Progress: (4/20) | 5.49 s
[Task 5/25] Current/Best: 18.57/ 19.95 GFLOPS | Progress: (8/20) | 7.22 s
[Task 5/25] Current/Best: 6.33/ 19.95 GFLOPS | Progress: (12/20) | 9.01 s
[Task 5/25] Current/Best: 14.62/ 19.95 GFLOPS | Progress: (16/20) | 11.46 s
[Task 5/25] Current/Best: 13.01/ 19.95 GFLOPS | Progress: (20/20) | 13.59 s Done.
+
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 13.06/ 15.34 GFLOPS | Progress: (4/20) | 4.79 s
[Task 6/25] Current/Best: 2.81/ 15.34 GFLOPS | Progress: (8/20) | 8.07 s
[Task 6/25] Current/Best: 14.08/ 15.45 GFLOPS | Progress: (12/20) | 10.59 s
[Task 6/25] Current/Best: 5.76/ 15.45 GFLOPS | Progress: (16/20) | 13.45 s
[Task 6/25] Current/Best: 13.95/ 15.45 GFLOPS | Progress: (20/20) | 16.85 s Done.
+
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 11.67/ 14.88 GFLOPS | Progress: (4/20) | 5.64 s
[Task 7/25] Current/Best: 12.16/ 14.88 GFLOPS | Progress: (8/20) | 8.41 s
[Task 7/25] Current/Best: 17.72/ 17.72 GFLOPS | Progress: (12/20) | 10.90 s
[Task 7/25] Current/Best: 22.92/ 22.92 GFLOPS | Progress: (16/20) | 12.72 s
[Task 7/25] Current/Best: 11.82/ 22.92 GFLOPS | Progress: (20/20) | 14.99 s Done.
+
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 4.55/ 19.25 GFLOPS | Progress: (4/20) | 5.03 s
[Task 8/25] Current/Best: 13.15/ 19.25 GFLOPS | Progress: (8/20) | 8.55 s
[Task 8/25] Current/Best: 17.78/ 19.25 GFLOPS | Progress: (12/20) | 19.90 s
[Task 8/25] Current/Best: 9.06/ 19.25 GFLOPS | Progress: (16/20) | 24.03 s
[Task 8/25] Current/Best: 7.83/ 19.25 GFLOPS | Progress: (20/20) | 28.48 s
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
+
[Task 9/25] Current/Best: 9.97/ 17.45 GFLOPS | Progress: (4/20) | 13.93 s
[Task 9/25] Current/Best: 13.22/ 17.45 GFLOPS | Progress: (8/20) | 19.40 s
[Task 9/25] Current/Best: 7.00/ 23.06 GFLOPS | Progress: (12/20) | 23.03 s
[Task 9/25] Current/Best: 17.93/ 23.06 GFLOPS | Progress: (16/20) | 24.69 s
[Task 9/25] Current/Best: 7.36/ 23.06 GFLOPS | Progress: (20/20) | 35.89 s
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 14.69/ 16.30 GFLOPS | Progress: (4/20) | 6.09 s
[Task 10/25] Current/Best: 2.91/ 16.30 GFLOPS | Progress: (8/20) | 8.95 s
[Task 10/25] Current/Best: 4.22/ 16.30 GFLOPS | Progress: (12/20) | 11.94 s
[Task 10/25] Current/Best: 7.99/ 18.38 GFLOPS | Progress: (16/20) | 13.56 s
[Task 10/25] Current/Best: 12.29/ 18.38 GFLOPS | Progress: (20/20) | 15.63 s Done.
+
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 11.12/ 16.00 GFLOPS | Progress: (4/20) | 5.55 s
[Task 11/25] Current/Best: 19.41/ 22.76 GFLOPS | Progress: (8/20) | 7.58 s
[Task 11/25] Current/Best: 14.02/ 22.76 GFLOPS | Progress: (12/20) | 9.86 s
[Task 11/25] Current/Best: 15.84/ 22.76 GFLOPS | Progress: (16/20) | 12.02 s
[Task 11/25] Current/Best: 18.86/ 22.76 GFLOPS | Progress: (20/20) | 15.71 s Done.
+
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 12.93/ 18.75 GFLOPS | Progress: (4/20) | 5.19 s
[Task 12/25] Current/Best: 1.61/ 18.75 GFLOPS | Progress: (8/20) | 8.27 s
[Task 12/25] Current/Best: 9.48/ 18.75 GFLOPS | Progress: (12/20) | 18.22 s
[Task 12/25] Current/Best: 10.61/ 18.75 GFLOPS | Progress: (16/20) | 25.47 s
[Task 12/25] Current/Best: 9.96/ 18.75 GFLOPS | Progress: (20/20) | 28.65 s Done.
+
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 8.59/ 18.61 GFLOPS | Progress: (4/20) | 6.37 s
[Task 13/25] Current/Best: 17.42/ 18.61 GFLOPS | Progress: (8/20) | 8.76 s
[Task 13/25] Current/Best: 15.68/ 22.73 GFLOPS | Progress: (12/20) | 12.41 s
[Task 13/25] Current/Best: 14.82/ 22.73 GFLOPS | Progress: (16/20) | 14.87 s
[Task 13/25] Current/Best: 20.82/ 22.73 GFLOPS | Progress: (20/20) | 18.17 s Done.
+
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 16.10/ 16.10 GFLOPS | Progress: (4/20) | 4.88 s
[Task 14/25] Current/Best: 18.05/ 18.05 GFLOPS | Progress: (8/20) | 7.40 s
[Task 14/25] Current/Best: 8.93/ 18.05 GFLOPS | Progress: (12/20) | 11.81 s
[Task 14/25] Current/Best: 14.50/ 18.05 GFLOPS | Progress: (16/20) | 14.14 s
[Task 14/25] Current/Best: 9.12/ 18.05 GFLOPS | Progress: (20/20) | 18.17 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 21.89/ 21.89 GFLOPS | Progress: (4/20) | 7.28 s
[Task 15/25] Current/Best: 18.28/ 21.89 GFLOPS | Progress: (8/20) | 9.19 s
[Task 15/25] Current/Best: 16.20/ 21.89 GFLOPS | Progress: (12/20) | 14.05 s Done.
+
[Task 15/25] Current/Best: 13.76/ 21.89 GFLOPS | Progress: (16/20) | 17.48 s
[Task 15/25] Current/Best: 13.13/ 21.89 GFLOPS | Progress: (20/20) | 19.76 s Done.
+
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 9.80/ 13.18 GFLOPS | Progress: (4/20) | 6.86 s
[Task 16/25] Current/Best: 7.50/ 16.80 GFLOPS | Progress: (8/20) | 10.82 s
[Task 16/25] Current/Best: 17.99/ 18.83 GFLOPS | Progress: (12/20) | 12.62 s
[Task 16/25] Current/Best: 5.40/ 18.83 GFLOPS | Progress: (16/20) | 16.04 s
[Task 16/25] Current/Best: 13.59/ 18.83 GFLOPS | Progress: (20/20) | 18.08 s Done.
+
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 16.58/ 22.86 GFLOPS | Progress: (4/20) | 5.56 s
[Task 17/25] Current/Best: 9.99/ 22.86 GFLOPS | Progress: (8/20) | 8.04 s
[Task 17/25] Current/Best: 12.29/ 22.86 GFLOPS | Progress: (12/20) | 10.99 s
[Task 17/25] Current/Best: 12.28/ 22.88 GFLOPS | Progress: (16/20) | 13.10 s
[Task 17/25] Current/Best: 11.18/ 22.88 GFLOPS | Progress: (20/20) | 16.12 s Done.
+
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 10.01/ 20.28 GFLOPS | Progress: (4/20) | 5.09 s
[Task 18/25] Current/Best: 3.00/ 20.28 GFLOPS | Progress: (8/20) | 8.62 s
[Task 18/25] Current/Best: 11.85/ 20.28 GFLOPS | Progress: (12/20) | 11.85 s
[Task 18/25] Current/Best: 19.38/ 20.28 GFLOPS | Progress: (16/20) | 15.93 s
[Task 18/25] Current/Best: 14.32/ 20.28 GFLOPS | Progress: (20/20) | 23.90 s Done.
+
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 8.00/ 12.17 GFLOPS | Progress: (4/20) | 8.40 s
[Task 19/25] Current/Best: 17.82/ 17.82 GFLOPS | Progress: (8/20) | 11.22 s
[Task 19/25] Current/Best: 12.63/ 18.01 GFLOPS | Progress: (12/20) | 14.59 s
[Task 19/25] Current/Best: 14.76/ 18.01 GFLOPS | Progress: (16/20) | 17.23 s
[Task 19/25] Current/Best: 10.37/ 22.12 GFLOPS | Progress: (20/20) | 20.34 s Done.
+
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 1.58/ 13.86 GFLOPS | Progress: (4/20) | 6.33 s
[Task 20/25] Current/Best: 15.84/ 15.84 GFLOPS | Progress: (8/20) | 8.40 s
[Task 20/25] Current/Best: 4.04/ 20.63 GFLOPS | Progress: (12/20) | 12.10 s
[Task 20/25] Current/Best: 17.41/ 20.63 GFLOPS | Progress: (16/20) | 14.38 s
[Task 20/25] Current/Best: 13.79/ 20.63 GFLOPS | Progress: (20/20) | 18.43 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 2.48/ 17.47 GFLOPS | Progress: (4/20) | 5.81 s
[Task 21/25] Current/Best: 5.41/ 17.77 GFLOPS | Progress: (8/20) | 9.34 s
[Task 21/25] Current/Best: 13.61/ 17.77 GFLOPS | Progress: (12/20) | 11.04 s
[Task 21/25] Current/Best: 21.19/ 21.19 GFLOPS | Progress: (16/20) | 14.01 s
[Task 21/25] Current/Best: 17.64/ 21.19 GFLOPS | Progress: (20/20
) | 15.63 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 10.29/ 20.49 GFLOPS | Progress: (4/20) | 4.59 s Done.
Done.
-
[Task 25/25] Current/Best: 3.00/ 9.72 GFLOPS | Progress: (4/20) | 8.53 s
[Task 25/25] Current/Best: 8.85/ 9.72 GFLOPS | Progress: (8/20) | 19.52 s
[Task 25/25] Current/Best: 7.36/ 9.72 GFLOPS | Progress: (12/20) | 22.86 s
[Task 25/25] Current/Best: 2.97/ 9.72 GFLOPS | Progress: (16/20) | 28.51 s
[Task 25/25] Current/Best: 5.56/ 9.72 GFLOPS | Progress: (20/20) | 35.66 s
+
[Task 22/25] Current/Best: 6.77/ 20.49 GFLOPS | Progress: (8/20) | 7.45 s
[Task 22/25] Current/Best: 19.89/ 20.49 GFLOPS | Progress: (12/20) | 11.61 s
[Task 22/25] Current/Best: 4.74/ 20.49 GFLOPS | Progress: (16/20) | 14.43 s
[Task 22/25] Current/Best: 18.12/ 20.49 GFLOPS | Progress: (20/20) | 16.87 s Done.
+
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 9.22/ 20.66 GFLOPS | Progress: (4/20) | 6.74 s
[Task 23/25] Current/Best: 18.31/ 20.66 GFLOPS | Progress: (8/20) | 10.74 s
[Task 23/25] Current/Best: 10.87/ 20.66 GFLOPS | Progress: (12/20) | 14.14 s
[Task 23/25] Current/Best: 12.47/ 20.66 GFLOPS | Progress: (16/20) | 17.60 s
[Task 23/25] Current/Best: 7.45/ 21.96 GFLOPS | Progress: (20/20) | 20.12 s Done.
+
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 3.39/ 5.58 GFLOPS | Progress: (4/20) | 13.49 s
[Task 24/25] Current/Best: 2.79/ 5.58 GFLOPS | Progress: (8/20) | 24.55 s
[Task 24/25] Current/Best: 3.66/ 5.58 GFLOPS | Progress: (12/20) | 30.43 s
[Task 24/25] Current/Best: 8.43/ 8.43 GFLOPS | Progress: (16/20) | 41.08 s
[Task 24/25] Current/Best: 8.05/ 8.43 GFLOPS | Progress: (20/20) | 52.03 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
+ Done.
+
[Task 25/25] Current/Best: 1.54/ 7.67 GFLOPS | Progress: (4/20) | 5.55 s
[Task 25/25] Current/Best: 5.14/ 7.67 GFLOPS | Progress: (8/20) | 16.49 s
[Task 25/25] Current/Best: 3.00/ 7.67 GFLOPS | Progress: (12/20) | 28.76 s
[Task 25/25] Current/Best: 5.62/ 9.28 GFLOPS | Progress: (16/20) | 37.04 s
[Task 25/25] Current/Best: 6.80/ 9.28 GFLOPS | Progress: (20/20) | 38.52 s
@@ -663,8 +665,8 @@ Verify that the optimized model runs and produces the same results:
.. code-block:: none
- class='n02123045 tabby, tabby cat' with probability=0.621103
- class='n02123159 tiger cat' with probability=0.356379
+ class='n02123045 tabby, tabby cat' with probability=0.621104
+ class='n02123159 tiger cat' with probability=0.356378
class='n02124075 Egyptian cat' with probability=0.019712
class='n02129604 tiger, Panthera tigris' with probability=0.001215
class='n04040759 radiator' with probability=0.000262
@@ -721,8 +723,8 @@ improvement in comparing the optimized model to the unoptimized model.
.. code-block:: none
- optimized: {'mean': 412.0973418800054, 'median': 412.5166654000054, 'std': 1.5141097041492024}
- unoptimized: {'mean': 512.8374819899979, 'median': 513.379893900003, 'std': 2.048731521479577}
+ optimized: {'mean': 409.33492624999644, 'median': 409.87332964998586, 'std': 2.0441469975383506}
+ unoptimized: {'mean': 515.2812953900003, 'median': 514.8744986499992, 'std': 2.452215899798013}
@@ -745,7 +747,7 @@ profiling/benchmarking.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 12 minutes 21.049 seconds)
+ **Total running time of the script:** ( 12 minutes 47.991 seconds)
.. _sphx_glr_download_tutorial_autotvm_relay_x86.py:
diff --git a/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt b/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
index 6c9783930a..054df7b66d 100644
--- a/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
+++ b/docs/_sources/tutorial/cross_compilation_and_rpc.rst.txt
@@ -274,7 +274,7 @@ device and returns the measured cost. Network overhead is excluded.
.. code-block:: none
- 1.377e-07 secs/op
+ 1.288e-07 secs/op
diff --git a/docs/_sources/tutorial/intro_topi.rst.txt b/docs/_sources/tutorial/intro_topi.rst.txt
index 9d1afe3a67..9cee5d493a 100644
--- a/docs/_sources/tutorial/intro_topi.rst.txt
+++ b/docs/_sources/tutorial/intro_topi.rst.txt
@@ -270,7 +270,7 @@ As you can see, scheduled stages of computation have been accumulated and we can
.. code-block:: none
- [stage(a, placeholder(a, 0xdfead00)), stage(b, placeholder(b, 0x10038f90)), stage(T_add, compute(T_add, body=[a[ax0, ax1, ax2] + b[ax1, ax2]], axis=[T.iter_var(ax0, T.Range(0, 100), "DataPar", ""), T.iter_var(ax1, T.Range(0, 10), "DataPar", ""), T.iter_var(ax2, T.Range(0, 10), "DataPar", "")], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[a[ax0, ax1, ax2] * b[ax1, ax2]], axis=[T.iter_var(ax0, T.Range(0, 100), "DataPar", ""), T.iter_var(ax1, T. [...]
+ [stage(a, placeholder(a, 0x2227b040)), stage(b, placeholder(b, 0x2a11bcf0)), stage(T_add, compute(T_add, body=[a[ax0, ax1, ax2] + b[ax1, ax2]], axis=[T.iter_var(ax0, T.Range(0, 100), "DataPar", ""), T.iter_var(ax1, T.Range(0, 10), "DataPar", ""), T.iter_var(ax2, T.Range(0, 10), "DataPar", "")], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[a[ax0, ax1, ax2] * b[ax1, ax2]], axis=[T.iter_var(ax0, T.Range(0, 100), "DataPar", ""), T.iter_var(ax1, T [...]
diff --git a/docs/_sources/tutorial/sg_execution_times.rst.txt b/docs/_sources/tutorial/sg_execution_times.rst.txt
index f774a768f1..03fad0ad68 100644
--- a/docs/_sources/tutorial/sg_execution_times.rst.txt
+++ b/docs/_sources/tutorial/sg_execution_times.rst.txt
@@ -5,24 +5,24 @@
Computation times
=================
-**15:56.194** total execution time for **tutorial** files:
+**16:23.387** total execution time for **tutorial** files:
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 12:21.049 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 12:47.991 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 01:24.282 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 01:27.041 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 01:02.551 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 01:01.413 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:37.875 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:36.483 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:27.550 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:27.823 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:01.825 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:01.609 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.864 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.857 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.198 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.170 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_uma.py` (``uma.py``) | 00:00.000 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
index 2c7b0d983c..553d1e644b 100644
--- a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
+++ b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
@@ -285,7 +285,7 @@ helper function to run a profile of the TVM generated code.
.. code-block:: none
- Numpy running time: 0.000008
+ Numpy running time: 0.000007
naive: 0.000007
@@ -389,7 +389,7 @@ compile and run this new schedule with the parallel operation applied:
.. code-block:: none
- parallel: 0.000007
+ parallel: 0.000006
@@ -498,10 +498,10 @@ We can now compare the different schedules
.. code-block:: none
Operator Timing Performance
- numpy 7.911820000572334e-06 1.0
- naive 6.7071000000000005e-06 0.8477316217399807
- parallel 7.0035e-06 0.8851945569405488
- vector 2.48371e-05 3.1392397701418013
+ numpy 7.001609999406355e-06 1.0
+ naive 6.6951000000000005e-06 0.9562229259509823
+ parallel 6.084e-06 0.868943000326474
+ vector 2.4646800000000003e-05 3.52016179165788
@@ -922,7 +922,7 @@ matrix multiplication.
.. code-block:: none
- Numpy running time: 0.019915
+ Numpy running time: 0.018042
@@ -980,7 +980,7 @@ optimizations.
.. code-block:: none
- none: 3.473186
+ none: 3.453903
@@ -1080,7 +1080,7 @@ schedule.
.. code-block:: none
- blocking: 0.323740
+ blocking: 0.297601
@@ -1164,7 +1164,7 @@ already cache friendly from our previous optimizations.
.. code-block:: none
- vectorization: 0.353062
+ vectorization: 0.333232
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1230,7 +1230,7 @@ more cache friendly.
.. code-block:: none
- loop permutation: 0.121582
+ loop permutation: 0.114607
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1321,7 +1321,7 @@ optimized schedule.
.. code-block:: none
- array packing: 0.108879
+ array packing: 0.108105
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1404,7 +1404,7 @@ to `C` when all the block results are ready.
.. code-block:: none
- block caching: 0.111291
+ block caching: 0.111731
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1478,7 +1478,7 @@ of thread-level parallelization.
.. code-block:: none
- parallelization: 0.147603
+ parallelization: 0.146658
# from tvm.script import ir as I
# from tvm.script import tir as T
@@ -1548,13 +1548,13 @@ working, we can compare the results.
.. code-block:: none
Operator Timing Performance
- none 3.4731861198 1.0
- blocking 0.3237404139 0.09321136349544157
- vectorization 0.3530619727 0.10165362883585713
- loop permutation 0.12158248149999999 0.035006036908555076
- array packing 0.10887909850000002 0.03134847795207408
- block caching 0.11129050930000002 0.032042771524840884
- parallelization 0.1476034903 0.04249800765312848
+ none 3.4539027075999997 1.0
+ blocking 0.2976011993 0.08616374706941095
+ vectorization 0.3332319551 0.09647983261565338
+ loop permutation 0.11460730920000002 0.03318197381409066
+ array packing 0.10810513790000001 0.03129941606696809
+ block caching 0.1117314031 0.03234931975765999
+ parallelization 0.1466577059 0.04246144675045218
@@ -1596,7 +1596,7 @@ the computation for specific platforms.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 2.551 seconds)
+ **Total running time of the script:** ( 1 minutes 1.413 seconds)
.. _sphx_glr_download_tutorial_tensor_expr_get_started.py:
diff --git a/docs/commit_hash b/docs/commit_hash
index a995f521ab..4ed23c773c 100644
--- a/docs/commit_hash
+++ b/docs/commit_hash
@@ -1 +1 @@
-e5ae4347ddf492a9e017dd1dba47df364a27a197
+b98755637542b52ad8177955d76e471a190540f1
diff --git a/docs/genindex.html b/docs/genindex.html
index 4249f39e9a..c7da667ec1 100644
--- a/docs/genindex.html
+++ b/docs/genindex.html
@@ -2451,8 +2451,6 @@
<li><a href="reference/api/python/target.html#tvm.target.list_tags">list_tags() (in module tvm.target)</a>
</li>
<li><a href="reference/api/python/contrib.html#tvm.contrib.utils.TempDirectory.listdir">listdir() (tvm.contrib.utils.TempDirectory method)</a>
-</li>
- <li><a href="reference/api/python/tir.html#tvm.tir.Load">Load (class in tvm.tir)</a>
</li>
<li><a href="reference/api/python/auto_scheduler.html#tvm.auto_scheduler.ApplyHistoryBest.load">load() (tvm.auto_scheduler.ApplyHistoryBest method)</a>
@@ -2490,10 +2488,10 @@
<li><a href="reference/api/python/rpc.html#tvm.rpc.RPCSession.load_module">(tvm.rpc.RPCSession method)</a>
</li>
</ul></li>
- </ul></td>
- <td style="width: 33%; vertical-align: top;"><ul>
<li><a href="reference/api/python/runtime.html#tvm.runtime.load_param_dict">load_param_dict() (in module tvm.runtime)</a>
</li>
+ </ul></td>
+ <td style="width: 33%; vertical-align: top;"><ul>
<li><a href="reference/api/python/runtime.html#tvm.runtime.load_param_dict_from_file">load_param_dict_from_file() (in module tvm.runtime)</a>
</li>
<li><a href="reference/api/python/graph_executor.html#tvm.contrib.graph_executor.GraphModule.load_params">load_params() (tvm.contrib.graph_executor.GraphModule method)</a>
@@ -3916,8 +3914,6 @@
<li><a href="reference/api/python/tir.html#tvm.tir.transform.StorageFlatten">StorageFlatten() (in module tvm.tir.transform)</a>
</li>
<li><a href="reference/api/python/tir.html#tvm.tir.transform.StorageRewrite">StorageRewrite() (in module tvm.tir.transform)</a>
-</li>
- <li><a href="reference/api/python/tir.html#tvm.tir.Store">Store (class in tvm.tir)</a>
</li>
<li><a href="reference/api/python/topi.html#tvm.topi.nn.Workload.stride_h">stride_h (tvm.topi.nn.Workload property)</a>
</li>
diff --git a/docs/how_to/compile_models/from_darknet.html b/docs/how_to/compile_models/from_darknet.html
index f4cc9a8431..9a8c4f9b85 100644
--- a/docs/how_to/compile_models/from_darknet.html
+++ b/docs/how_to/compile_models/from_darknet.html
@@ -590,7 +590,7 @@ class:['truck 0.9266'] left:471 top:83 right:689 bottom:169
class:['bicycle 0.9984'] left:111 top:113 right:577 bottom:447
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 23.554 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 20.288 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-compile-models-from-darknet-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/7716f96385bd5abb6e822041e285be54/from_darknet.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">from_darknet.py</span></code></a></p>
diff --git a/docs/how_to/compile_models/from_mxnet.html b/docs/how_to/compile_models/from_mxnet.html
index 024bfb88a5..32a5ce512b 100644
--- a/docs/how_to/compile_models/from_mxnet.html
+++ b/docs/how_to/compile_models/from_mxnet.html
@@ -444,7 +444,7 @@
<span class="nb">print</span><span class="p">(</span><span class="s2">"x"</span><span class="p">,</span> <a href="https://docs.python.org/3/library/stdtypes.html#tuple" title="builtins.tuple" class="sphx-glr-backref-module-builtins sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">x</span><span class="o">.</span><span class="n">shape</span></a><span class="p">)</span>
</pre></div>
</div>
-<img src="../../_images/sphx_glr_from_mxnet_001.png" srcset="../../_images/sphx_glr_from_mxnet_001.png" alt="from mxnet" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zip0345609a-aa53-4983-888b-c0a10eb99acb from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+<img src="../../_images/sphx_glr_from_mxnet_001.png" srcset="../../_images/sphx_glr_from_mxnet_001.png" alt="from mxnet" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipb72f8b5f-aacb-4c5e-a189-0ae1bb48fe1d from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
x (1, 3, 224, 224)
</pre></div>
</div>
diff --git a/docs/how_to/compile_models/from_oneflow.html b/docs/how_to/compile_models/from_oneflow.html
index a7a07147df..5caf00e897 100644
--- a/docs/how_to/compile_models/from_oneflow.html
+++ b/docs/how_to/compile_models/from_oneflow.html
@@ -454,13 +454,14 @@ Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdo
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading: "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/ResNet/resnet18.zip" to /workspace/.oneflow/flowvision_cache/resnet18.zip
0%| | 0.00/41.5M [00:00<?, ?B/s]
- 15%|#5 | 6.33M/41.5M [00:00<00:00, 66.1MB/s]
- 30%|### | 12.6M/41.5M [00:00<00:00, 51.6MB/s]
- 43%|####2 | 17.8M/41.5M [00:00<00:00, 33.7MB/s]
- 58%|#####7 | 24.0M/41.5M [00:00<00:00, 35.9MB/s]
- 77%|#######7 | 32.0M/41.5M [00:00<00:00, 46.0MB/s]
- 92%|#########2| 38.4M/41.5M [00:00<00:00, 51.2MB/s]
-100%|##########| 41.5M/41.5M [00:00<00:00, 46.5MB/s]
+ 19%|#9 | 7.99M/41.5M [00:00<00:00, 55.3MB/s]
+ 35%|###4 | 14.3M/41.5M [00:00<00:00, 47.3MB/s]
+ 45%|####5 | 18.9M/41.5M [00:00<00:00, 44.5MB/s]
+ 57%|#####6 | 23.5M/41.5M [00:00<00:00, 46.0MB/s]
+ 67%|######7 | 27.9M/41.5M [00:00<00:00, 43.3MB/s]
+ 77%|#######7 | 32.1M/41.5M [00:00<00:00, 43.5MB/s]
+ 96%|#########6| 40.0M/41.5M [00:00<00:00, 48.0MB/s]
+100%|##########| 41.5M/41.5M [00:00<00:00, 47.8MB/s]
</pre></div>
</div>
</div>
diff --git a/docs/how_to/compile_models/from_pytorch.html b/docs/how_to/compile_models/from_pytorch.html
index 3968a87a39..f66ab8bde1 100644
--- a/docs/how_to/compile_models/from_pytorch.html
+++ b/docs/how_to/compile_models/from_pytorch.html
@@ -437,12 +437,12 @@ be unstable.</p>
Downloading: "https://download.pytorch.org/models/resnet18-f37072fd.pth" to /workspace/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth
0%| | 0.00/44.7M [00:00<?, ?B/s]
- 18%|#7 | 7.99M/44.7M [00:00<00:00, 76.6MB/s]
- 40%|#### | 18.1M/44.7M [00:00<00:00, 86.7MB/s]
- 59%|#####8 | 26.3M/44.7M [00:00<00:00, 79.6MB/s]
- 76%|#######5 | 33.9M/44.7M [00:00<00:00, 78.7MB/s]
- 96%|#########5| 42.8M/44.7M [00:00<00:00, 83.6MB/s]
-100%|##########| 44.7M/44.7M [00:00<00:00, 74.4MB/s]
+ 18%|#7 | 7.99M/44.7M [00:00<00:00, 67.9MB/s]
+ 32%|###2 | 14.5M/44.7M [00:00<00:00, 45.2MB/s]
+ 46%|####6 | 20.7M/44.7M [00:00<00:00, 51.9MB/s]
+ 72%|#######1 | 32.0M/44.7M [00:00<00:00, 60.6MB/s]
+ 99%|#########8| 44.2M/44.7M [00:00<00:00, 68.0MB/s]
+100%|##########| 44.7M/44.7M [00:00<00:00, 62.8MB/s]
</pre></div>
</div>
</div>
diff --git a/docs/how_to/compile_models/from_tensorflow.html b/docs/how_to/compile_models/from_tensorflow.html
index 85c03f65bf..a258fedc97 100644
--- a/docs/how_to/compile_models/from_tensorflow.html
+++ b/docs/how_to/compile_models/from_tensorflow.html
@@ -657,7 +657,7 @@ banana (score = 0.00022)
desk (score = 0.00019)
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 35.258 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 29.717 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-compile-models-from-tensorflow-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/7f1d3d1b878694c201c614c807cdebc8/from_tensorflow.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">from_tensorflow.py</span></code></a></p>
diff --git a/docs/how_to/compile_models/sg_execution_times.html b/docs/how_to/compile_models/sg_execution_times.html
index cc532e22e2..cb8915be60 100644
--- a/docs/how_to/compile_models/sg_execution_times.html
+++ b/docs/how_to/compile_models/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-compile-models-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>07:04.515</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
+<p><strong>06:47.377</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 81%" />
@@ -354,43 +354,43 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="from_tensorflow.html#sphx-glr-how-to-compile-models-from-tensorflow-py"><span class="std std-ref">Compile Tensorflow Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_tensorflow.py</span></code>)</p></td>
-<td><p>01:35.258</p></td>
+<td><p>01:29.717</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_darknet.html#sphx-glr-how-to-compile-models-from-darknet-py"><span class="std std-ref">Compile YOLO-V2 and YOLO-V3 in DarkNet Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_darknet.py</span></code>)</p></td>
-<td><p>01:23.554</p></td>
+<td><p>01:20.288</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="from_paddle.html#sphx-glr-how-to-compile-models-from-paddle-py"><span class="std std-ref">Compile PaddlePaddle Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_paddle.py</span></code>)</p></td>
-<td><p>00:59.622</p></td>
+<td><p>00:56.265</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_oneflow.html#sphx-glr-how-to-compile-models-from-oneflow-py"><span class="std std-ref">Compile OneFlow Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_oneflow.py</span></code>)</p></td>
-<td><p>00:39.309</p></td>
+<td><p>00:37.813</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="from_coreml.html#sphx-glr-how-to-compile-models-from-coreml-py"><span class="std std-ref">Compile CoreML Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_coreml.py</span></code>)</p></td>
-<td><p>00:33.471</p></td>
+<td><p>00:32.203</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_mxnet.html#sphx-glr-how-to-compile-models-from-mxnet-py"><span class="std std-ref">Compile MXNet Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_mxnet.py</span></code>)</p></td>
-<td><p>00:31.014</p></td>
+<td><p>00:30.112</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="from_tflite.html#sphx-glr-how-to-compile-models-from-tflite-py"><span class="std std-ref">Compile TFLite Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_tflite.py</span></code>)</p></td>
-<td><p>00:28.503</p></td>
+<td><p>00:28.207</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_pytorch.html#sphx-glr-how-to-compile-models-from-pytorch-py"><span class="std std-ref">Compile PyTorch Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_pytorch.py</span></code>)</p></td>
-<td><p>00:27.353</p></td>
+<td><p>00:26.670</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="from_keras.html#sphx-glr-how-to-compile-models-from-keras-py"><span class="std std-ref">Compile Keras Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_keras.py</span></code>)</p></td>
-<td><p>00:23.635</p></td>
+<td><p>00:23.425</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="from_onnx.html#sphx-glr-how-to-compile-models-from-onnx-py"><span class="std std-ref">Compile ONNX Models</span></a> (<code class="docutils literal notranslate"><span class="pre">from_onnx.py</span></code>)</p></td>
-<td><p>00:02.796</p></td>
+<td><p>00:02.677</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/deploy_models/deploy_model_on_adreno.html b/docs/how_to/deploy_models/deploy_model_on_adreno.html
index 4676140e85..88c409a59c 100644
--- a/docs/how_to/deploy_models/deploy_model_on_adreno.html
+++ b/docs/how_to/deploy_models/deploy_model_on_adreno.html
@@ -799,7 +799,7 @@ Top5 predictions:
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 2750.3544 2750.1791 2754.7262 2747.9918 1.7600
+ 2537.8528 2536.9948 2544.7511 2535.1032 2.6745
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-model-on-adreno-py">
diff --git a/docs/how_to/deploy_models/deploy_model_on_adreno_tvmc.html b/docs/how_to/deploy_models/deploy_model_on_adreno_tvmc.html
index 121556d85c..686a1ced43 100644
--- a/docs/how_to/deploy_models/deploy_model_on_adreno_tvmc.html
+++ b/docs/how_to/deploy_models/deploy_model_on_adreno_tvmc.html
@@ -443,28 +443,27 @@ to run this tutorial with a real device over rpc.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/resnet/resnet50_weights_tf_dim_ordering_tf_kernels.h5
8192/102967424 [..............................] - ETA: 0s
- 4775936/102967424 [>.............................] - ETA: 1s
8380416/102967424 [=>............................] - ETA: 1s
- 10575872/102967424 [==>...........................] - ETA: 1s
- 15024128/102967424 [===>..........................] - ETA: 1s
+ 15024128/102967424 [===>..........................] - ETA: 0s
16769024/102967424 [===>..........................] - ETA: 1s
- 25157632/102967424 [======>.......................] - ETA: 1s
- 33546240/102967424 [========>.....................] - ETA: 1s
- 40189952/102967424 [==========>...................] - ETA: 0s
- 41934848/102967424 [===========>..................] - ETA: 1s
- 47046656/102967424 [============>.................] - ETA: 0s
+ 25157632/102967424 [======>.......................] - ETA: 0s
+ 30605312/102967424 [=======>......................] - ETA: 0s
+ 33546240/102967424 [========>.....................] - ETA: 0s
+ 41934848/102967424 [===========>..................] - ETA: 0s
+ 48578560/102967424 [=============>................] - ETA: 0s
50323456/102967424 [=============>................] - ETA: 0s
- 56967168/102967424 [===============>..............] - ETA: 0s
+ 58318848/102967424 [===============>..............] - ETA: 0s
58712064/102967424 [================>.............] - ETA: 0s
- 65323008/102967424 [==================>...........] - ETA: 0s
- 67305472/102967424 [==================>...........] - ETA: 0s
+ 65355776/102967424 [==================>...........] - ETA: 0s
+ 67100672/102967424 [==================>...........] - ETA: 0s
69296128/102967424 [===================>..........] - ETA: 0s
75489280/102967424 [====================>.........] - ETA: 0s
- 82878464/102967424 [=======================>......] - ETA: 0s
+ 81362944/102967424 [======================>.......] - ETA: 0s
83877888/102967424 [=======================>......] - ETA: 0s
- 90521600/102967424 [=========================>....] - ETA: 0s
+ 91979776/102967424 [=========================>....] - ETA: 0s
92266496/102967424 [=========================>....] - ETA: 0s
98910208/102967424 [===========================>..] - ETA: 0s
+102449152/102967424 [============================>.] - ETA: 0s
102850560/102967424 [============================>.] - ETA: 0s
102967424/102967424 [==============================] - 2s 0us/step
</pre></div>
diff --git a/docs/how_to/deploy_models/deploy_model_on_android.html b/docs/how_to/deploy_models/deploy_model_on_android.html
index 7cbd333a1e..4982298ad2 100644
--- a/docs/how_to/deploy_models/deploy_model_on_android.html
+++ b/docs/how_to/deploy_models/deploy_model_on_android.html
@@ -667,7 +667,7 @@ to the remote android device.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 16.1671 16.1450 16.5010 15.9790 0.1270
+ 16.2733 16.4747 16.7356 15.5997 0.4025
</pre></div>
</div>
</div>
diff --git a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
index 92f6630a77..92e66bf754 100644
--- a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
+++ b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
@@ -459,31 +459,33 @@ be unstable.</p>
Downloading: "https://download.pytorch.org/models/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth" to /workspace/.cache/torch/hub/checkpoints/maskrcnn_resnet50_fpn_coco-bf2d0c1e.pth
0%| | 0.00/170M [00:00<?, ?B/s]
- 4%|3 | 6.30M/170M [00:00<00:02, 61.9MB/s]
- 7%|7 | 12.2M/170M [00:00<00:02, 57.3MB/s]
- 12%|#1 | 19.8M/170M [00:00<00:02, 66.9MB/s]
- 15%|#5 | 26.2M/170M [00:00<00:02, 53.9MB/s]
- 19%|#8 | 32.0M/170M [00:00<00:02, 50.7MB/s]
- 24%|##3 | 40.0M/170M [00:00<00:02, 52.2MB/s]
- 28%|##8 | 48.0M/170M [00:00<00:02, 55.5MB/s]
- 33%|###2 | 56.0M/170M [00:01<00:02, 57.5MB/s]
- 37%|###6 | 62.3M/170M [00:01<00:01, 58.4MB/s]
- 40%|#### | 68.1M/170M [00:01<00:01, 59.0MB/s]
- 43%|####3 | 73.8M/170M [00:01<00:01, 54.7MB/s]
- 47%|####7 | 80.0M/170M [00:01<00:01, 53.6MB/s]
- 51%|##### | 86.3M/170M [00:01<00:01, 55.2MB/s]
- 54%|#####3 | 91.6M/170M [00:01<00:01, 47.5MB/s]
- 57%|#####6 | 96.3M/170M [00:02<00:02, 34.5MB/s]
- 61%|######1 | 104M/170M [00:02<00:01, 40.7MB/s]
- 66%|######5 | 112M/170M [00:02<00:01, 43.0MB/s]
- 71%|####### | 120M/170M [00:02<00:01, 46.7MB/s]
- 75%|#######5 | 128M/170M [00:02<00:00, 50.3MB/s]
- 80%|######## | 136M/170M [00:02<00:00, 54.8MB/s]
- 85%|########4 | 144M/170M [00:02<00:00, 56.6MB/s]
- 89%|########9 | 152M/170M [00:03<00:00, 54.0MB/s]
- 93%|#########3| 158M/170M [00:03<00:00, 55.7MB/s]
- 97%|#########6| 165M/170M [00:03<00:00, 58.2MB/s]
-100%|##########| 170M/170M [00:03<00:00, 51.9MB/s]
+ 4%|3 | 6.30M/170M [00:00<00:03, 56.7MB/s]
+ 7%|6 | 11.7M/170M [00:00<00:03, 45.4MB/s]
+ 9%|9 | 16.1M/170M [00:00<00:04, 33.1MB/s]
+ 14%|#4 | 24.0M/170M [00:00<00:04, 36.6MB/s]
+ 19%|#8 | 32.0M/170M [00:00<00:03, 41.4MB/s]
+ 24%|##3 | 40.0M/170M [00:00<00:03, 45.0MB/s]
+ 28%|##8 | 48.0M/170M [00:01<00:02, 50.7MB/s]
+ 34%|###3 | 57.1M/170M [00:01<00:01, 61.1MB/s]
+ 38%|###7 | 64.0M/170M [00:01<00:02, 50.5MB/s]
+ 42%|####2 | 72.0M/170M [00:01<00:01, 51.4MB/s]
+ 46%|####6 | 78.3M/170M [00:01<00:01, 51.4MB/s]
+ 50%|####9 | 84.6M/170M [00:01<00:01, 54.8MB/s]
+ 53%|#####3 | 90.2M/170M [00:01<00:01, 50.4MB/s]
+ 56%|#####6 | 95.2M/170M [00:02<00:01, 48.6MB/s]
+ 59%|#####9 | 101M/170M [00:02<00:01, 51.0MB/s]
+ 62%|######2 | 106M/170M [00:02<00:01, 39.7MB/s]
+ 66%|######5 | 112M/170M [00:02<00:01, 37.4MB/s]
+ 71%|####### | 120M/170M [00:02<00:01, 44.6MB/s]
+ 75%|#######5 | 128M/170M [00:02<00:00, 51.9MB/s]
+ 79%|#######9 | 134M/170M [00:02<00:00, 47.8MB/s]
+ 82%|########1 | 139M/170M [00:03<00:00, 40.7MB/s]
+ 85%|########4 | 144M/170M [00:03<00:00, 37.6MB/s]
+ 89%|########9 | 152M/170M [00:03<00:00, 40.6MB/s]
+ 93%|#########3| 158M/170M [00:03<00:00, 44.8MB/s]
+ 96%|#########5| 163M/170M [00:03<00:00, 38.7MB/s]
+ 99%|#########9| 169M/170M [00:03<00:00, 43.9MB/s]
+100%|##########| 170M/170M [00:03<00:00, 45.6MB/s]
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torch/nn/functional.py:3897: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
for i in range(dim)
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/torchvision/models/detection/anchor_utils.py:124: UserWarning: __floordiv__ is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode=& [...]
@@ -581,7 +583,7 @@ torchvision rcnn models.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Get 9 valid boxes
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 46.558 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 34.020 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-object-detection-pytorch-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/7795da4b258c8feff986668b95ef57ad/deploy_object_detection_pytorch.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_object_detection_pytorch.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_prequantized.html b/docs/how_to/deploy_models/deploy_prequantized.html
index 9e78af6c0e..4d9fe5ca6f 100644
--- a/docs/how_to/deploy_models/deploy_prequantized.html
+++ b/docs/how_to/deploy_models/deploy_prequantized.html
@@ -500,8 +500,9 @@ training. Other models require a full post training calibration.</p>
Downloading: "https://download.pytorch.org/models/mobilenet_v2-b0353104.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2-b0353104.pth
0%| | 0.00/13.6M [00:00<?, ?B/s]
- 59%|#####8 | 7.99M/13.6M [00:00<00:00, 38.6MB/s]
-100%|##########| 13.6M/13.6M [00:00<00:00, 58.9MB/s]
+ 47%|####6 | 6.30M/13.6M [00:00<00:00, 66.1MB/s]
+ 93%|#########3| 12.6M/13.6M [00:00<00:00, 53.1MB/s]
+100%|##########| 13.6M/13.6M [00:00<00:00, 57.6MB/s]
</pre></div>
</div>
</div>
@@ -592,7 +593,7 @@ output values are identical out of 1000 outputs from mobilenet v2.</p>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 88.9308 88.2800 91.3655 88.0628 0.9692
+ 90.0841 90.0309 91.8659 89.8894 0.2243
</pre></div>
</div>
<div class="admonition note">
@@ -631,7 +632,7 @@ This includes support for the VNNI 8 bit dot product instruction (CascadeLake or
<div class="section" id="deploy-a-quantized-tflite-model">
<h2>Deploy a quantized TFLite Model<a class="headerlink" href="#deploy-a-quantized-tflite-model" title="Permalink to this headline">¶</a></h2>
<p>TODO</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 19.728 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 16.487 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-prequantized-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/fb8217c13f4351224c6cf3aacf1a87fc/deploy_prequantized.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_prequantized.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_prequantized_tflite.html b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
index 0c6cf17cf8..ad367acfa3 100644
--- a/docs/how_to/deploy_models/deploy_prequantized_tflite.html
+++ b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
@@ -585,7 +585,7 @@ TFLite Top-5 labels: [387 102 386 341 349]
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 121.0844 121.2201 122.4309 118.6074 0.7515
+ 119.4306 119.5678 121.8174 118.1030 0.7109
</pre></div>
</div>
<div class="admonition note">
@@ -613,7 +613,7 @@ network for ARM CPU</span></a>.</p></li>
</ul>
</div></blockquote>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 29.938 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 34.338 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-prequantized-tflite-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/56691c7a27d45da61d112276334640d3/deploy_prequantized_tflite.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_prequantized_tflite.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_quantized.html b/docs/how_to/deploy_models/deploy_quantized.html
index 1ecfb7df10..ed8e4ad1fa 100644
--- a/docs/how_to/deploy_models/deploy_quantized.html
+++ b/docs/how_to/deploy_models/deploy_quantized.html
@@ -526,7 +526,7 @@ for calibration. But the accuracy might be impacted.</p>
DeprecationWarning,
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 39.936 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 45.103 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-quantized-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/7810ecf51bfc05f7d5e8a400ac3e815d/deploy_quantized.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_quantized.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
index 7cef05f46f..d9da073194 100644
--- a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
+++ b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
@@ -468,25 +468,24 @@ to your device.</p>
Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/ssd_512_resnet50_v1_voc-9c8b225a.zip...
0%| | 0/132723 [00:00<?, ?KB/s]
- 1%| | 1308/132723 [00:00<00:10, 13040.62KB/s]
- 3%|3 | 4237/132723 [00:00<00:05, 22586.65KB/s]
- 8%|7 | 10121/132723 [00:00<00:03, 39126.63KB/s]
- 12%|#2 | 16441/132723 [00:00<00:02, 48624.36KB/s]
- 18%|#7 | 23636/132723 [00:00<00:01, 57032.13KB/s]
- 23%|##3 | 31133/132723 [00:00<00:01, 63125.25KB/s]
- 29%|##9 | 39038/132723 [00:00<00:01, 68327.34KB/s]
- 35%|###5 | 46987/132723 [00:00<00:01, 71876.62KB/s]
- 41%|####1 | 54929/132723 [00:00<00:01, 74232.14KB/s]
- 47%|####7 | 62701/132723 [00:01<00:00, 75299.27KB/s]
- 53%|#####3 | 70589/132723 [00:01<00:00, 76390.87KB/s]
- 59%|#####9 | 78560/132723 [00:01<00:00, 77398.99KB/s]
- 65%|######5 | 86458/132723 [00:01<00:00, 77874.69KB/s]
- 71%|#######1 | 94448/132723 [00:01<00:00, 78484.07KB/s]
- 77%|#######7 | 102380/132723 [00:01<00:00, 78735.08KB/s]
- 83%|########3 | 110273/132723 [00:01<00:00, 78790.94KB/s]
- 89%|########9 | 118255/132723 [00:01<00:00, 79098.68KB/s]
- 95%|#########5| 126172/132723 [00:01<00:00, 79118.26KB/s]
-100%|##########| 132723/132723 [00:01<00:00, 70493.20KB/s]
+ 1%|1 | 1613/132723 [00:00<00:08, 16128.59KB/s]
+ 4%|4 | 5330/132723 [00:00<00:04, 28504.02KB/s]
+ 8%|8 | 10902/132723 [00:00<00:02, 40929.15KB/s]
+ 15%|#4 | 19465/132723 [00:00<00:01, 58574.48KB/s]
+ 21%|##1 | 28045/132723 [00:00<00:01, 68387.65KB/s]
+ 28%|##7 | 36629/132723 [00:00<00:01, 74319.99KB/s]
+ 34%|###4 | 45148/132723 [00:00<00:01, 77869.29KB/s]
+ 41%|#### | 53804/132723 [00:00<00:00, 80630.79KB/s]
+ 47%|####7 | 62459/132723 [00:00<00:00, 82478.00KB/s]
+ 54%|#####3 | 71143/132723 [00:01<00:00, 83822.82KB/s]
+ 60%|###### | 79773/132723 [00:01<00:00, 84575.19KB/s]
+ 67%|######6 | 88453/132723 [00:01<00:00, 85250.04KB/s]
+ 73%|#######3 | 97105/132723 [00:01<00:00, 85631.59KB/s]
+ 80%|#######9 | 105812/132723 [00:01<00:00, 86064.69KB/s]
+ 86%|########6 | 114483/132723 [00:01<00:00, 86256.98KB/s]
+ 93%|#########2| 123165/132723 [00:01<00:00, 86424.23KB/s]
+ 99%|#########9| 131808/132723 [00:01<00:00, 86417.50KB/s]
+100%|##########| 132723/132723 [00:01<00:00, 77487.62KB/s]
</pre></div>
</div>
<p>Create TVM runtime and do inference
@@ -525,7 +524,7 @@ Downloading /workspace/.mxnet/models/ssd_512_resnet50_v1_voc-9c8b225a.zip from h
<span class="n">plt</span><span class="o">.</span><span class="n">show</span><span class="p">()</span>
</pre></div>
</div>
-<img src="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" srcset="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" alt="deploy ssd gluoncv" class = "sphx-glr-single-img"/><p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 59.099 seconds)</p>
+<img src="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" srcset="../../_images/sphx_glr_deploy_ssd_gluoncv_001.png" alt="deploy ssd gluoncv" class = "sphx-glr-single-img"/><p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 48.257 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-ssd-gluoncv-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/cccb17d28e5e8b2e94ea8cd5ec59f6ed/deploy_ssd_gluoncv.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_ssd_gluoncv.py</span></code></a></p>
diff --git a/docs/how_to/deploy_models/sg_execution_times.html b/docs/how_to/deploy_models/sg_execution_times.html
index 356aa10380..fafde17d4d 100644
--- a/docs/how_to/deploy_models/sg_execution_times.html
+++ b/docs/how_to/deploy_models/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-deploy-models-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>16:49.430</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
+<p><strong>16:25.272</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 86%" />
@@ -354,43 +354,43 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_ssd_gluoncv.html#sphx-glr-how-to-deploy-models-deploy-ssd-gluoncv-py"><span class="std std-ref">Deploy Single Shot Multibox Detector(SSD) model</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_ssd_gluoncv.py</span></code>)</p></td>
-<td><p>03:59.099</p></td>
+<td><p>03:48.257</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_object_detection_pytorch.html#sphx-glr-how-to-deploy-models-deploy-object-detection-pytorch-py"><span class="std std-ref">Compile PyTorch Object Detection Models</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_object_detection_pytorch.py</span></code>)</p></td>
-<td><p>03:46.558</p></td>
+<td><p>03:34.020</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_prequantized_tflite.html#sphx-glr-how-to-deploy-models-deploy-prequantized-tflite-py"><span class="std std-ref">Deploy a Framework-prequantized Model with TVM - Part 3 (TFLite)</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_prequantized_tflite.py</span></code>)</p></td>
-<td><p>02:29.938</p></td>
+<td><p>02:34.338</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_quantized.html#sphx-glr-how-to-deploy-models-deploy-quantized-py"><span class="std std-ref">Deploy a Quantized Model on Cuda</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_quantized.py</span></code>)</p></td>
-<td><p>01:39.936</p></td>
+<td><p>01:45.103</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_prequantized.html#sphx-glr-how-to-deploy-models-deploy-prequantized-py"><span class="std std-ref">Deploy a Framework-prequantized Model with TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_prequantized.py</span></code>)</p></td>
-<td><p>01:19.728</p></td>
+<td><p>01:16.487</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_model_on_adreno.html#sphx-glr-how-to-deploy-models-deploy-model-on-adreno-py"><span class="std std-ref">Deploy the Pretrained Model on Adrenoâ„¢</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_adreno.py</span></code>)</p></td>
-<td><p>00:58.394</p></td>
+<td><p>00:54.919</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_model_on_adreno_tvmc.html#sphx-glr-how-to-deploy-models-deploy-model-on-adreno-tvmc-py"><span class="std std-ref">Deploy the Pretrained Model on Adrenoâ„¢ with tvmc Interface</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_adreno_tvmc.py</span></code>)</p></td>
-<td><p>00:53.202</p></td>
+<td><p>00:52.118</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_model_on_android.html#sphx-glr-how-to-deploy-models-deploy-model-on-android-py"><span class="std std-ref">Deploy the Pretrained Model on Android</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_android.py</span></code>)</p></td>
-<td><p>00:44.689</p></td>
+<td><p>00:42.890</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_model_on_nano.html#sphx-glr-how-to-deploy-models-deploy-model-on-nano-py"><span class="std std-ref">Deploy the Pretrained Model on Jetson Nano</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_nano.py</span></code>)</p></td>
-<td><p>00:29.210</p></td>
+<td><p>00:28.694</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="deploy_model_on_rasp.html#sphx-glr-how-to-deploy-models-deploy-model-on-rasp-py"><span class="std std-ref">Deploy the Pretrained Model on Raspberry Pi</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_rasp.py</span></code>)</p></td>
-<td><p>00:28.670</p></td>
+<td><p>00:28.440</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="deploy_sparse.html#sphx-glr-how-to-deploy-models-deploy-sparse-py"><span class="std std-ref">Deploy a Hugging Face Pruned Model on CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_sparse.py</span></code>)</p></td>
diff --git a/docs/how_to/extend_tvm/bring_your_own_datatypes.html b/docs/how_to/extend_tvm/bring_your_own_datatypes.html
index f4c346d721..db83e4ecf4 100644
--- a/docs/how_to/extend_tvm/bring_your_own_datatypes.html
+++ b/docs/how_to/extend_tvm/bring_your_own_datatypes.html
@@ -624,7 +624,7 @@ In this alpha state of the Bring Your Own Datatypes framework, we have not imple
<span class="n">module</span><span class="p">,</span> <a href="https://docs.python.org/3/library/stdtypes.html#dict" title="builtins.dict" class="sphx-glr-backref-module-builtins sphx-glr-backref-type-py-class sphx-glr-backref-instance"><span class="n">params</span></a> <span class="o">=</span> <span class="n">get_mobilenet</span><span class="p">()</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip929e5e9d-8d88-4ea5-95fc-5c8e6ad4ee23 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zipeea5ca4e-641e-4da1-8968-6e0f02dd652e from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
</pre></div>
</div>
<p>It’s easy to execute MobileNet with native TVM:</p>
diff --git a/docs/how_to/extend_tvm/sg_execution_times.html b/docs/how_to/extend_tvm/sg_execution_times.html
index 4d4d60160a..09327af206 100644
--- a/docs/how_to/extend_tvm/sg_execution_times.html
+++ b/docs/how_to/extend_tvm/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-extend-tvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:58.769</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
+<p><strong>00:54.099</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -354,19 +354,19 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="bring_your_own_datatypes.html#sphx-glr-how-to-extend-tvm-bring-your-own-datatypes-py"><span class="std std-ref">Bring Your Own Datatypes to TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">bring_your_own_datatypes.py</span></code>)</p></td>
-<td><p>00:54.661</p></td>
+<td><p>00:50.187</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="use_pass_instrument.html#sphx-glr-how-to-extend-tvm-use-pass-instrument-py"><span class="std std-ref">How to Use TVM Pass Instrument</span></a> (<code class="docutils literal notranslate"><span class="pre">use_pass_instrument.py</span></code>)</p></td>
-<td><p>00:02.952</p></td>
+<td><p>00:02.801</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="use_pass_infra.html#sphx-glr-how-to-extend-tvm-use-pass-infra-py"><span class="std std-ref">How to Use TVM Pass Infra</span></a> (<code class="docutils literal notranslate"><span class="pre">use_pass_infra.py</span></code>)</p></td>
-<td><p>00:01.149</p></td>
+<td><p>00:01.105</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="low_level_custom_pass.html#sphx-glr-how-to-extend-tvm-low-level-custom-pass-py"><span class="std std-ref">Writing a Customized Pass</span></a> (<code class="docutils literal notranslate"><span class="pre">low_level_custom_pass.py</span></code>)</p></td>
-<td><p>00:00.008</p></td>
+<td><p>00:00.007</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/extend_tvm/use_pass_instrument.html b/docs/how_to/extend_tvm/use_pass_instrument.html
index 3c06a1b48a..a9a98e3ca2 100644
--- a/docs/how_to/extend_tvm/use_pass_instrument.html
+++ b/docs/how_to/extend_tvm/use_pass_instrument.html
@@ -531,10 +531,10 @@ profile the execution time of each passes.</p>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Printing results of timing profile...
-InferType: 23565us [23565us] (49.07%; 49.07%)
-FoldScaleAxis: 24454us [9us] (50.93%; 50.93%)
- FoldConstant: 24444us [1762us] (50.91%; 99.96%)
- InferType: 22683us [22683us] (47.24%; 92.79%)
+InferType: 22163us [22163us] (48.64%; 48.64%)
+FoldScaleAxis: 23400us [8us] (51.36%; 51.36%)
+ FoldConstant: 23391us [1749us] (51.34%; 99.96%)
+ InferType: 21643us [21643us] (47.50%; 92.52%)
</pre></div>
</div>
</div>
@@ -556,10 +556,10 @@ Refer to following sections and <a class="reference internal" href="../../refere
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Printing results of timing profile...
-InferType: 23804us [23804us] (49.22%; 49.22%)
-FoldScaleAxis: 24559us [8us] (50.78%; 50.78%)
- FoldConstant: 24551us [1875us] (50.76%; 99.97%)
- InferType: 22675us [22675us] (46.89%; 92.36%)
+InferType: 21773us [21773us] (48.20%; 48.20%)
+FoldScaleAxis: 23399us [6us] (51.80%; 51.80%)
+ FoldConstant: 23393us [1794us] (51.79%; 99.97%)
+ InferType: 21599us [21599us] (47.82%; 92.33%)
</pre></div>
</div>
<p>Register empty list to clear existing instruments.</p>
diff --git a/docs/how_to/optimize_operators/opt_conv_cuda.html b/docs/how_to/optimize_operators/opt_conv_cuda.html
index 59be08adbd..5d5db1f922 100644
--- a/docs/how_to/optimize_operators/opt_conv_cuda.html
+++ b/docs/how_to/optimize_operators/opt_conv_cuda.html
@@ -580,7 +580,7 @@ latency of convolution.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Convolution: </span><span class="si">%f</span><span class="s2"> ms"</span> <span class="o">%</span> <span class="p">(</span><span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">w</span><span class="p">,</span> <span class="n">b</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span> <span class="o">*</span> <span cl [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 53.512542 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 34.863105 ms
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-optimize-operators-opt-conv-cuda-py">
diff --git a/docs/how_to/optimize_operators/opt_conv_tensorcore.html b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
index 1b2c741221..157833cbd4 100644
--- a/docs/how_to/optimize_operators/opt_conv_tensorcore.html
+++ b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
@@ -862,7 +862,7 @@ be able to run on our build server</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"conv2d with tensor core: </span><span class="si">%f</span><span class="s2"> ms"</span> <span class="o">%</span> <span class="p">(</span><span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">w</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span> <span class="o">* [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 12.223469 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 12.234599 ms
</pre></div>
</div>
</div>
diff --git a/docs/how_to/optimize_operators/opt_gemm.html b/docs/how_to/optimize_operators/opt_gemm.html
index ee74ff08b2..754151b7aa 100644
--- a/docs/how_to/optimize_operators/opt_gemm.html
+++ b/docs/how_to/optimize_operators/opt_gemm.html
@@ -477,8 +477,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
<span class="nb">print</span><span class="p">(</span><span class="s2">"Baseline: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.019328
-Baseline: 3.498159
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.017909
+Baseline: 3.397584
</pre></div>
</div>
<p>In TVM, we can always inspect lower level IR to debug or optimize our schedule.
@@ -537,7 +537,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt1: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.329814
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.303157
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -594,7 +594,7 @@ vastly.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt2: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.353135
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.331700
</pre></div>
</div>
<p>Here is the generated IR after vectorization.</p>
@@ -649,7 +649,7 @@ the access pattern for A matrix is more cache friendly.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt3: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.134273
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.124434
</pre></div>
</div>
<p>Here is the generated IR after loop permutation.</p>
@@ -726,7 +726,7 @@ flattening.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt4: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.110898
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.110773
</pre></div>
</div>
<p>Here is the generated IR after array packing.</p>
@@ -804,7 +804,7 @@ write to C when all the block results are ready.</p>
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt5: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">evaluator</span><span class="p">(</span><span class="n">a</span><span class="p">,</span> <span class="n">b</span><span class="p">,</span> <span class="n">c</span><span class="p">)</span><span class="o">.</span><span class="n">mean</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.112132
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.111581
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -884,7 +884,7 @@ class Module:
<span class="nb">print</span><span class="p">(</span><span class="s2">"Opt6: </span><span class="si">%f</span><span class="s2">"</span> <span class="o">%</span> <span class="n">opt6_time</span><span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.148916
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.146901
</pre></div>
</div>
<p>Here is the generated IR after parallelization.</p>
diff --git a/docs/how_to/optimize_operators/sg_execution_times.html b/docs/how_to/optimize_operators/sg_execution_times.html
index ed7dbe98f3..74e6ab3c02 100644
--- a/docs/how_to/optimize_operators/sg_execution_times.html
+++ b/docs/how_to/optimize_operators/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-optimize-operators-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:36.796</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
+<p><strong>00:35.333</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -354,15 +354,15 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="opt_gemm.html#sphx-glr-how-to-optimize-operators-opt-gemm-py"><span class="std std-ref">How to optimize GEMM on CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_gemm.py</span></code>)</p></td>
-<td><p>00:33.689</p></td>
+<td><p>00:32.406</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="opt_conv_tensorcore.html#sphx-glr-how-to-optimize-operators-opt-conv-tensorcore-py"><span class="std std-ref">How to optimize convolution using TensorCores</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_conv_tensorcore.py</span></code>)</p></td>
-<td><p>00:01.894</p></td>
+<td><p>00:01.830</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="opt_conv_cuda.html#sphx-glr-how-to-optimize-operators-opt-conv-cuda-py"><span class="std std-ref">How to optimize convolution on GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">opt_conv_cuda.py</span></code>)</p></td>
-<td><p>00:01.212</p></td>
+<td><p>00:01.097</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
index 73626f2c03..50f6eb05c8 100644
--- a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
+++ b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-tune-with-autoscheduler-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>10:36.831</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
+<p><strong>10:16.843</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 85%" />
@@ -354,27 +354,27 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_conv2d_layer_cuda.html#sphx-glr-how-to-tune-with-autoscheduler-tune-conv2d-layer-cuda-py"><span class="std std-ref">Auto-scheduling a Convolution Layer for GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_conv2d_layer_cuda.py</span></code>)</p></td>
-<td><p>06:28.831</p></td>
+<td><p>06:10.766</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_network_x86.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-x86-py"><span class="std std-ref">Auto-scheduling a Neural Network for x86 CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_x86.py</span></code>)</p></td>
-<td><p>01:47.644</p></td>
+<td><p>01:42.680</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_network_cuda.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-cuda-py"><span class="std std-ref">Auto-scheduling a Neural Network for NVIDIA GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_cuda.py</span></code>)</p></td>
-<td><p>01:13.908</p></td>
+<td><p>01:10.583</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_sparse_x86.html#sphx-glr-how-to-tune-with-autoscheduler-tune-sparse-x86-py"><span class="std std-ref">Auto-scheduling Sparse Matrix Multiplication on CPU with Custom Sketch Rule</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_sparse_x86.py</span></code>)</p></td>
-<td><p>00:36.900</p></td>
+<td><p>00:44.969</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_network_arm.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-arm-py"><span class="std std-ref">Auto-scheduling a Neural Network for ARM CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_arm.py</span></code>)</p></td>
-<td><p>00:15.119</p></td>
+<td><p>00:14.425</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_network_mali.html#sphx-glr-how-to-tune-with-autoscheduler-tune-network-mali-py"><span class="std std-ref">Auto-scheduling a Neural Network for mali GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_network_mali.py</span></code>)</p></td>
-<td><p>00:14.429</p></td>
+<td><p>00:13.420</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html b/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
index 21e9da13dc..60cb7b8b81 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_conv2d_layer_cuda.html
@@ -511,85 +511,550 @@ class Module:
def main(data: T.Buffer((1, 512, 7, 7), "float32"), kernel: T.Buffer((512, 512, 3, 3), "float32"), bias: T.Buffer((1, 512, 1, 1), "float32"), compute: T.Buffer((1, 512, 7, 7), "float32")):
T.func_attr({"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True})
blockIdx_x = T.launch_thread("blockIdx.x", 32)
- conv2d_nchw = T.allocate([7], "float32", "local")
- pad_temp_shared = T.allocate([1568], "float32", "shared")
- kernel_shared = T.allocate([512], "float32", "shared")
- threadIdx_x = T.launch_thread("threadIdx.x", 112)
- conv2d_nchw_1 = T.Buffer((7,), data=conv2d_nchw, scope="local", align=16)
+ conv2d_nchw = T.allocate([8], "float32", "local")
+ pad_temp_shared = T.allocate([4032], "float32", "shared")
+ kernel_shared = T.allocate([3072], "float32", "shared")
+ threadIdx_x = T.launch_thread("threadIdx.x", 98)
+ conv2d_nchw_1 = T.Buffer((4,), data=conv2d_nchw, scope="local", align=8)
conv2d_nchw_1[0] = T.float32(0)
- conv2d_nchw_1[1] = T.float32(0)
conv2d_nchw_1[2] = T.float32(0)
- conv2d_nchw_1[3] = T.float32(0)
conv2d_nchw_1[4] = T.float32(0)
- conv2d_nchw_1[5] = T.float32(0)
conv2d_nchw_1[6] = T.float32(0)
- for rc_outer_outer, ry_outer_outer, rx_outer_outer in T.grid(16, 3, 3):
- cse_var_2: T.int32 = rc_outer_outer * 288
- cse_var_1: T.int32 = ry_outer_outer * 3
+ conv2d_nchw_1[1] = T.float32(0)
+ conv2d_nchw_1[3] = T.float32(0)
+ conv2d_nchw_1[5] = T.float32(0)
+ conv2d_nchw_1[7] = T.float32(0)
+ for rc_outer_outer, rx_outer_outer in T.grid(8, 3):
+ cse_var_2: T.int32 = rc_outer_outer * 3136
+ cse_var_1: T.int32 = rc_outer_outer * 576
threadIdx_x_1 = T.env_thread("threadIdx.x")
- pad_temp_shared_1 = T.Buffer((1568,), data=pad_temp_shared, scope="shared")
+ pad_temp_shared_1 = T.Buffer((4032,), data=pad_temp_shared, scope="shared")
data_1 = T.Buffer((25088,), data=data.data)
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer - 8], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 112] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 104], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 224] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 216], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 336] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 328], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 448] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 440], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 560] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 552], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 672] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 664], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 784] = T.if_then_else(1 <= threadIdx_x_1 % 49 // 7 + ry_outer_outer and threadIdx_x_1 % 49 // 7 + ry_outer_outer < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 776], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 896] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 2) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 888], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1008] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 4) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1000], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1120] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 6) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1112], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1232] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 1) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1224], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1344] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 3) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1336], T.float32(0))
- with T.launch_thread(threadIdx_x_1, 112):
- pad_temp_shared_1[threadIdx_x_1 + 1456] = T.if_then_else(1 <= ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 and ry_outer_outer + (threadIdx_x_1 // 7 + 5) % 7 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 % 7 and rx_outer_outer + threadIdx_x_1 % 7 < 8, data_1[rc_outer_outer * 1568 + ry_outer_outer * 7 + threadIdx_x_1 + rx_outer_outer + 1448], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2] = T.if_then_else(7 <= threadIdx_x_1 * 2 % 63 and threadIdx_x_1 * 2 % 63 < 56 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + threadIdx_x_1 * 2 // 63 * 49 + rx_outer_outer + threadIdx_x_1 * 2 % 63 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1] = T.if_then_else(7 <= (threadIdx_x_1 * 2 + 1) % 63 and (threadIdx_x_1 * 2 + 1) % 63 < 56 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1) // 63 * 49 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 63 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 196] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 1) % 9 and (threadIdx_x_1 * 2 // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 196) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 1) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 197] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 197) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 392] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 2) % 9 and (threadIdx_x_1 * 2 // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 392) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 2) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 393] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 393) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 588] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 3) % 9 and (threadIdx_x_1 * 2 // 7 + 3) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 588) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 3) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 589] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 589) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 784] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 4) % 9 and (threadIdx_x_1 * 2 // 7 + 4) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 784) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 4) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 785] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 785) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 980] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 5) % 9 and (threadIdx_x_1 * 2 // 7 + 5) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 980) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 5) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 981] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 981) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1176] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 6) % 9 and (threadIdx_x_1 * 2 // 7 + 6) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1176) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 6) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1177] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1177) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1372] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 7) % 9 and (threadIdx_x_1 * 2 // 7 + 7) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1372) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 7) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1373] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1373) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1568] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 8) % 9 and (threadIdx_x_1 * 2 // 7 + 8) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1568) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 8) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1569] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1569) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1764] = T.if_then_else(7 <= threadIdx_x_1 * 2 % 63 and threadIdx_x_1 * 2 % 63 < 56 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + threadIdx_x_1 * 2 // 63 * 49 + rx_outer_outer + threadIdx_x_1 * 2 % 63 + 1364], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1765] = T.if_then_else(7 <= (threadIdx_x_1 * 2 + 1) % 63 and (threadIdx_x_1 * 2 + 1) % 63 < 56 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1) // 63 * 49 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 63 + 1364], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1960] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 1) % 9 and (threadIdx_x_1 * 2 // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1960) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 1) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 1961] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1961) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2156] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 2) % 9 and (threadIdx_x_1 * 2 // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2156) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 2) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2157] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2157) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2352] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 3) % 9 and (threadIdx_x_1 * 2 // 7 + 3) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2352) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 3) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2353] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2353) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 3) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2548] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 4) % 9 and (threadIdx_x_1 * 2 // 7 + 4) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2548) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 4) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2549] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2549) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 4) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2744] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 5) % 9 and (threadIdx_x_1 * 2 // 7 + 5) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2744) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 5) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2745] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2745) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 5) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2940] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 6) % 9 and (threadIdx_x_1 * 2 // 7 + 6) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2940) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 6) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 2941] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 2941) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 6) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3136] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 7) % 9 and (threadIdx_x_1 * 2 // 7 + 7) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3136) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 7) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3137] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3137) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 7) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3332] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 8) % 9 and (threadIdx_x_1 * 2 // 7 + 8) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3332) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 8) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3333] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3333) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 8) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3528] = T.if_then_else(7 <= threadIdx_x_1 * 2 % 63 and threadIdx_x_1 * 2 % 63 < 56 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + threadIdx_x_1 * 2 // 63 * 49 + rx_outer_outer + threadIdx_x_1 * 2 % 63 + 2736], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3529] = T.if_then_else(7 <= (threadIdx_x_1 * 2 + 1) % 63 and (threadIdx_x_1 * 2 + 1) % 63 < 56 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 1) // 63 * 49 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 63 + 2736], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3724] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 1) % 9 and (threadIdx_x_1 * 2 // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3724) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 1) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3725] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3725) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 1) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
+ with T.launch_thread(threadIdx_x_1, 98):
+ if T.likely(threadIdx_x_1 < 56):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3920] = T.if_then_else(1 <= (threadIdx_x_1 * 2 // 7 + 2) % 9 and (threadIdx_x_1 * 2 // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + threadIdx_x_1 * 2 % 7 and rx_outer_outer + threadIdx_x_1 * 2 % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3920) // 63 * 49 + (threadIdx_x_1 * 2 // 7 + 2) % 9 * 7 + rx_outer_outer + threadIdx_x_1 * 2 % 7 - 8], T.float32(0))
+ if T.likely(threadIdx_x_1 < 56):
+ pad_temp_shared_1[threadIdx_x_1 * 2 + 3921] = T.if_then_else(1 <= ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 and ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 < 8 and 1 <= rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 and rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 < 8, data_1[cse_var_2 + (threadIdx_x_1 * 2 + 3921) // 63 * 49 + ((threadIdx_x_1 * 2 + 1) // 7 + 2) % 9 * 7 + rx_outer_outer + (threadIdx_x_1 * 2 + 1) % 7 - 8], T.float32(0))
threadIdx_x_2 = T.env_thread("threadIdx.x")
- kernel_shared_1 = T.Buffer((512,), data=kernel_shared, scope="shared")
+ kernel_shared_1 = T.Buffer((3072,), data=kernel_shared, scope="shared")
kernel_1 = T.Buffer((2359296,), data=kernel.data)
- with T.launch_thread(threadIdx_x_2, 112):
- kernel_shared_1[threadIdx_x_2 * 3] = kernel_1[blockIdx_x * 73728 + threadIdx_x_2 * 3 // 32 * 4608 + cse_var_2 + threadIdx_x_2 * 3 % 32 * 9 + cse_var_1 + rx_outer_outer]
- kernel_shared_1[threadIdx_x_2 * 3 + 1] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 1) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 1) % 32 * 9 + cse_var_1 + rx_outer_outer]
- kernel_shared_1[threadIdx_x_2 * 3 + 2] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 2) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 2) % 32 * 9 + cse_var_1 + rx_outer_outer]
- with T.launch_thread(threadIdx_x_2, 112):
- if T.likely(threadIdx_x_2 < 59):
- kernel_shared_1[threadIdx_x_2 * 3 + 336] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 336) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 16) % 32 * 9 + cse_var_1 + rx_outer_outer]
- if T.likely(threadIdx_x_2 < 59):
- kernel_shared_1[threadIdx_x_2 * 3 + 337] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 337) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 17) % 32 * 9 + cse_var_1 + rx_outer_outer]
- if T.likely(threadIdx_x_2 < 58):
- kernel_shared_1[threadIdx_x_2 * 3 + 338] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 * 3 + 338) // 32 * 4608 + cse_var_2 + (threadIdx_x_2 * 3 + 18) % 32 * 9 + cse_var_1 + rx_outer_outer]
- for rc_outer_inner in range(16):
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 1] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 2] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 3] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 4] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 5] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 6] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2]
- conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 49] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 50] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 51] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 52] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 53] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 54] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 98 + threadIdx_x % 7 * 7 + 55] * kernel_shared_1[threadIdx_x // 7 * 32 + rc_outer_inner * 2 + 1]
- for i3_inner in range(7):
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2] = kernel_1[blockIdx_x * 73728 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 98] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 98) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 98) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 196] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 196) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 12]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 294] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 294) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 102) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 392] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 392) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 24]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 490] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 490) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 106) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 588] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 588) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 36]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 686] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 686) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 110) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 784] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 784) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 48]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 882] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 882) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 114) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 980] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 980) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 60]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1078] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1078) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 118) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1176] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1176) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 72]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1274] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1274) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 122) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1372] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1372) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 84]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1470] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1470) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 126) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1568] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1568) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 96]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1666] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1666) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 130) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1764] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1764) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 108]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1862] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1862) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 134) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 1960] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 1960) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 120]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2058] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2058) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 138) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2156] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2156) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 132]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2254] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2254) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 142) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2352] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2352) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 144]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2450] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2450) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 146) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2548] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2548) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 156]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2646] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2646) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 150) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2744] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2744) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 168]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2842] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2842) // 192 * 4608 + cse_var_1 + (threadIdx_x_2 + 154) % 192 * 3 + rx_outer_outer]
+ with T.launch_thread(threadIdx_x_2, 98):
+ kernel_shared_1[threadIdx_x_2 + 2940] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 2940) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 180]
+ with T.launch_thread(threadIdx_x_2, 98):
+ if T.likely(threadIdx_x_2 < 34):
+ kernel_shared_1[threadIdx_x_2 + 3038] = kernel_1[blockIdx_x * 73728 + (threadIdx_x_2 + 3038) // 192 * 4608 + cse_var_1 + threadIdx_x_2 * 3 + rx_outer_outer + 474]
+ for rc_outer_inner in range(4):
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 768]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1536]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2304]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 769]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1537]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2305]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 770]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1538]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2306]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 3]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 771]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1539]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2307]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 4]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 772]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1540]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2308]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 5]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 773]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1541]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2309]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 6]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 774]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1542]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2310]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 7]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 775]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1543]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2311]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 8]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 776]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1544]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2312]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 9]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 777]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1545]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2313]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 10]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 778]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1546]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2314]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 11]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 779]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1547]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2315]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 12]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 780]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1548]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2316]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 13]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 781]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1549]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2317]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 14]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 782]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1550]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2318]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 15]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 783]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1551]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2319]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 16]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 784]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1552]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2320]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 17]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 785]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1553]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2321]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 18]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 786]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1554]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2322]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 19]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 787]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1555]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2323]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 20]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 788]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1556]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2324]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 21]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 789]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1557]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2325]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 22]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 790]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1558]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2326]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 23]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 791]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1559]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2327]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 24]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 792]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1560]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2328]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 25]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 793]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1561]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2329]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 26]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 794]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1562]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2330]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 27]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 795]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1563]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2331]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 28]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 796]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1564]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2332]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 29]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 797]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1565]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2333]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 30]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 798]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1566]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2334]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 31]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 799]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1567]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2335]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 32]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 800]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1568]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2336]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 33]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 801]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1569]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2337]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 34]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 802]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1570]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2338]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 35]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 803]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1571]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2339]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 36]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 804]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1572]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2340]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 37]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 805]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1573]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2341]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 38]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 806]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1574]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2342]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 39]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 807]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1575]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2343]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 40]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 808]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1576]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2344]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 41]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 809]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1577]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2345]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 42]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 810]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1578]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2346]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 43]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 811]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1579]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2347]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 44]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 812]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1580]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2348]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 45]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 813]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1581]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2349]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 46]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 814]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1582]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2350]
+ conv2d_nchw_1[0] = conv2d_nchw_1[0] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 47]
+ conv2d_nchw_1[2] = conv2d_nchw_1[2] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 815]
+ conv2d_nchw_1[4] = conv2d_nchw_1[4] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1583]
+ conv2d_nchw_1[6] = conv2d_nchw_1[6] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2351]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 192]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 960]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1728]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2496]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 193]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 961]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1729]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 7] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2497]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 194]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 962]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1730]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 14] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2498]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 195]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 963]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1731]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 63] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2499]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 196]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 964]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1732]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 70] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2500]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 197]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 965]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1733]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 77] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2501]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 198]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 966]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1734]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 126] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2502]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 199]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 967]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1735]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 133] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2503]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 200]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 968]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1736]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 140] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2504]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 201]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 969]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1737]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 189] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2505]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 202]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 970]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1738]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 196] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2506]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 203]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 971]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1739]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 203] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2507]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 204]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 972]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1740]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 252] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2508]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 205]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 973]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1741]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 259] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2509]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 206]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 974]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1742]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 266] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2510]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 207]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 975]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1743]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 315] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2511]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 208]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 976]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1744]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 322] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2512]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 209]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 977]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1745]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 329] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2513]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 210]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 978]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1746]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 378] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2514]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 211]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 979]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1747]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 385] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2515]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 212]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 980]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1748]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 392] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2516]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 213]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 981]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1749]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 441] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2517]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 214]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 982]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1750]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 448] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2518]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 215]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 983]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1751]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 455] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2519]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 216]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 984]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1752]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 504] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2520]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 217]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 985]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1753]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 511] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2521]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 218]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 986]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1754]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 518] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2522]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 219]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 987]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1755]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 567] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2523]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 220]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 988]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1756]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 574] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2524]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 221]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 989]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1757]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 581] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2525]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 222]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 990]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1758]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 630] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2526]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 223]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 991]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1759]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 637] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2527]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 224]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 992]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1760]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 644] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2528]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 225]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 993]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1761]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 693] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2529]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 226]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 994]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1762]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 700] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2530]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 227]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 995]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1763]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 707] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2531]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 228]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 996]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1764]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 756] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2532]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 229]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 997]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1765]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 763] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2533]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 230]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 998]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1766]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 770] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2534]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 231]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 999]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1767]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 819] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2535]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 232]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1000]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1768]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 826] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2536]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 233]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1001]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1769]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 833] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2537]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 234]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1002]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1770]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 882] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2538]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 235]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1003]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1771]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 889] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2539]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 236]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1004]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1772]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 896] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2540]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 237]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1005]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1773]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 945] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2541]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 238]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1006]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1774]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 952] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2542]
+ conv2d_nchw_1[1] = conv2d_nchw_1[1] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 239]
+ conv2d_nchw_1[3] = conv2d_nchw_1[3] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1007]
+ conv2d_nchw_1[5] = conv2d_nchw_1[5] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 1775]
+ conv2d_nchw_1[7] = conv2d_nchw_1[7] + pad_temp_shared_1[rc_outer_inner * 1008 + threadIdx_x % 49 + 959] * kernel_shared_1[threadIdx_x // 49 * 384 + rc_outer_inner * 48 + 2543]
+ for i1_inner in range(2):
compute_1 = T.Buffer((25088,), data=compute.data)
bias_1 = T.Buffer((512,), data=bias.data)
- compute_1[blockIdx_x * 784 + threadIdx_x * 7 + i3_inner] = T.max(conv2d_nchw_1[i3_inner] + bias_1[blockIdx_x * 16 + threadIdx_x // 7], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49] = T.max(conv2d_nchw_1[i1_inner] + bias_1[blockIdx_x * 16 + threadIdx_x // 49 * 2 + i1_inner], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49 + 196] = T.max(conv2d_nchw_1[i1_inner + 2] + bias_1[blockIdx_x * 16 + threadIdx_x // 49 * 2 + i1_inner + 4], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49 + 392] = T.max(conv2d_nchw_1[i1_inner + 4] + bias_1[blockIdx_x * 16 + threadIdx_x // 49 * 2 + i1_inner + 8], T.float32(0))
+ compute_1[blockIdx_x * 784 + threadIdx_x // 49 * 98 + i1_inner * 49 + threadIdx_x % 49 + 588] = T.max(conv2d_nchw_1[i1_inner + 6] + bias_1[blockIdx_x * 16 + threadIdx_x // 49 * 2 + i1_inner + 12], T.float32(0))
</pre></div>
</div>
</div>
@@ -623,7 +1088,7 @@ class Module:
<span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.387 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.229 ms
</pre></div>
</div>
</div>
@@ -653,20 +1118,20 @@ conv2d_nchw_nn_o_o_i, conv2d_nchw_nn_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o
conv2d_nchw_nn_o_o_o_i, conv2d_nchw_nn_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_i, factor=1)
conv2d_nchw_nn_o_o_o_o, conv2d_nchw_nn_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_o_o_i, factor=1)
conv2d_nchw_ff_o_i, conv2d_nchw_ff_i = s[conv2d_nchw].split(conv2d_nchw_ff, factor=1)
-conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=1)
-conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=16)
-conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=1)
+conv2d_nchw_ff_o_o_i, conv2d_nchw_ff_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_i, factor=2)
+conv2d_nchw_ff_o_o_o_i, conv2d_nchw_ff_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_i, factor=2)
+conv2d_nchw_ff_o_o_o_o, conv2d_nchw_ff_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_ff_o_o_o_i, factor=4)
conv2d_nchw_yy_o_i, conv2d_nchw_yy_i = s[conv2d_nchw].split(conv2d_nchw_yy, factor=1)
conv2d_nchw_yy_o_o_i, conv2d_nchw_yy_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_i, factor=1)
conv2d_nchw_yy_o_o_o_i, conv2d_nchw_yy_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_i, factor=7)
conv2d_nchw_yy_o_o_o_o, conv2d_nchw_yy_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_yy_o_o_o_i, factor=1)
-conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=7)
+conv2d_nchw_xx_o_i, conv2d_nchw_xx_i = s[conv2d_nchw].split(conv2d_nchw_xx, factor=1)
conv2d_nchw_xx_o_o_i, conv2d_nchw_xx_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_i, factor=1)
-conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=1)
+conv2d_nchw_xx_o_o_o_i, conv2d_nchw_xx_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_i, factor=7)
conv2d_nchw_xx_o_o_o_o, conv2d_nchw_xx_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_o_i, factor=1)
-conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=2)
-conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=16)
-conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=1)
+conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=16)
+conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=4)
+conv2d_nchw_ry_o_i, conv2d_nchw_ry_i = s[conv2d_nchw].split(conv2d_nchw_ry, factor=3)
conv2d_nchw_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=1)
conv2d_nchw_rx_o_i, conv2d_nchw_rx_i = s[conv2d_nchw].split(conv2d_nchw_rx, factor=1)
conv2d_nchw_rx_o_o, conv2d_nchw_rx_o_i = s[conv2d_nchw].split(conv2d_nchw_rx_o_i, factor=1)
@@ -674,14 +1139,14 @@ s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nc
compute_i0_o_i, compute_i0_i = s[compute].split(compute_i0, factor=1)
compute_i0_o_o_i, compute_i0_o_i = s[compute].split(compute_i0_o_i, factor=1)
compute_i0_o_o_o, compute_i0_o_o_i = s[compute].split(compute_i0_o_o_i, factor=1)
-compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=1)
-compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=16)
-compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=1)
+compute_i1_o_i, compute_i1_i = s[compute].split(compute_i1, factor=2)
+compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=2)
+compute_i1_o_o_o, compute_i1_o_o_i = s[compute].split(compute_i1_o_o_i, factor=4)
compute_i2_o_i, compute_i2_i = s[compute].split(compute_i2, factor=1)
compute_i2_o_o_i, compute_i2_o_i = s[compute].split(compute_i2_o_i, factor=7)
compute_i2_o_o_o, compute_i2_o_o_i = s[compute].split(compute_i2_o_o_i, factor=1)
-compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=7)
-compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=1)
+compute_i3_o_i, compute_i3_i = s[compute].split(compute_i3, factor=1)
+compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=7)
compute_i3_o_o_o, compute_i3_o_o_i = s[compute].split(compute_i3_o_o_i, factor=1)
s[compute].reorder(compute_i0_o_o_o, compute_i1_o_o_o, compute_i2_o_o_o, compute_i3_o_o_o, compute_i0_o_o_i, compute_i1_o_o_i, compute_i2_o_o_i, compute_i3_o_o_i, compute_i0_o_i, compute_i1_o_i, compute_i2_o_i, compute_i3_o_i, compute_i0_i, compute_i1_i, compute_i2_i, compute_i3_i)
s[conv2d_nchw].compute_at(s[compute], compute_i3_o_i)
@@ -699,16 +1164,16 @@ s[compute].bind(compute_i0_o_o_i_i1_o_o_i_fused_i2_o_o_i_fused_i3_o_o_i_fused, t
compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused = s[compute].fuse(compute_i0_o_i, compute_i1_o_i, compute_i2_o_i, compute_i3_o_i)
s[compute].bind(compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused, te.thread_axis("threadIdx.x"))
kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[kernel_shared].fuse(kernel_shared_ax0, kernel_shared_ax1, kernel_shared_ax2, kernel_shared_ax3)
-kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=3)
+kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
s[kernel_shared].vectorize(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
-kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=112)
+kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[kernel_shared].split(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=98)
s[kernel_shared].bind(kernel_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused = s[pad_temp_shared].fuse(pad_temp_shared_ax0, pad_temp_shared_ax1, pad_temp_shared_ax2, pad_temp_shared_ax3)
-pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=1)
+pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused, factor=2)
s[pad_temp_shared].vectorize(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_i)
-pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=112)
+pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_o, pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i = s[pad_temp_shared].split(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o, factor=98)
s[pad_temp_shared].bind(pad_temp_shared_ax0_ax1_fused_ax2_fused_ax3_fused_o_i, te.thread_axis("threadIdx.x"))
-s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 64)
+s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "auto_unroll_max_step", 1024)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
CUDA source code:
@@ -726,69 +1191,495 @@ CUDA source code:
#define int64_t long long
#define uint64_t unsigned long long
#endif
-extern "C" __global__ void __launch_bounds__(112) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
- float conv2d_nchw[7];
- __shared__ float pad_temp_shared[1568];
- __shared__ float kernel_shared[512];
+extern "C" __global__ void __launch_bounds__(98) default_function_kernel0(float* __restrict__ data, float* __restrict__ kernel, float* __restrict__ compute, float* __restrict__ bias) {
+ float conv2d_nchw[8];
+ __shared__ float pad_temp_shared[4032];
+ __shared__ float kernel_shared[3072];
conv2d_nchw[0] = 0.000000e+00f;
- conv2d_nchw[1] = 0.000000e+00f;
conv2d_nchw[2] = 0.000000e+00f;
- conv2d_nchw[3] = 0.000000e+00f;
conv2d_nchw[4] = 0.000000e+00f;
- conv2d_nchw[5] = 0.000000e+00f;
conv2d_nchw[6] = 0.000000e+00f;
- for (int rc_outer_outer = 0; rc_outer_outer < 16; ++rc_outer_outer) {
- for (int ry_outer_outer = 0; ry_outer_outer < 3; ++ry_outer_outer) {
- for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
- __syncthreads();
- pad_temp_shared[((int)threadIdx.x)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) - 8)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 104)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 216)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 328)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 440)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 552)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 672)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 664)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((1 <= (((((int)threadIdx.x) % 49) / 7) + ry_outer_outer)) && ((((((int)threadIdx.x) % 49) / 7) + ry_outer_outer) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 776)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 896)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 2) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 888)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1008)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 4) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1000)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1120)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 6) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1112)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1232)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 1) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1224)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1344)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 3) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1336)] : 0.000000e+00f);
- pad_temp_shared[(((int)threadIdx.x) + 1456)] = (((((1 <= (ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7))) && ((ry_outer_outer + (((((int)threadIdx.x) / 7) + 5) % 7)) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 1568) + (ry_outer_outer * 7)) + ((int)threadIdx.x)) + rx_outer_outer) + 1448)] : 0.000000e+00f);
- kernel_shared[(((int)threadIdx.x) * 3)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) * 3) >> 5) * 4608)) + (rc_outer_outer * 288)) + (((((int)threadIdx.x) * 3) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- kernel_shared[((((int)threadIdx.x) * 3) + 1)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 1) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 1) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- kernel_shared[((((int)threadIdx.x) * 3) + 2)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 2) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 2) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- if (((int)threadIdx.x) < 59) {
- kernel_shared[((((int)threadIdx.x) * 3) + 336)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 336) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 16) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- if (((int)threadIdx.x) < 59) {
- kernel_shared[((((int)threadIdx.x) * 3) + 337)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 337) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 17) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- if (((int)threadIdx.x) < 58) {
- kernel_shared[((((int)threadIdx.x) * 3) + 338)] = kernel[((((((((int)blockIdx.x) * 73728) + ((((((int)threadIdx.x) * 3) + 338) >> 5) * 4608)) + (rc_outer_outer * 288)) + ((((((int)threadIdx.x) * 3) + 18) & 31) * 9)) + (ry_outer_outer * 3)) + rx_outer_outer)];
- }
- __syncthreads();
- for (int rc_outer_inner = 0; rc_outer_inner < 16; ++rc_outer_inner) {
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7))] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 1)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 2)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 3)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 4)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 5)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 6)] * kernel_shared[(((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2))]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 49)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 50)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 51)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 52)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 53)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 54)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 98) + ((((int)threadIdx.x) % 7) * 7)) + 55)] * kernel_shared[((((((int)threadIdx.x) / 7) * 32) + (rc_outer_inner * 2)) + 1)]));
- }
+ conv2d_nchw[1] = 0.000000e+00f;
+ conv2d_nchw[3] = 0.000000e+00f;
+ conv2d_nchw[5] = 0.000000e+00f;
+ conv2d_nchw[7] = 0.000000e+00f;
+ for (int rc_outer_outer = 0; rc_outer_outer < 8; ++rc_outer_outer) {
+ for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
+ __syncthreads();
+ pad_temp_shared[(((int)threadIdx.x) * 2)] = (((((7 <= ((((int)threadIdx.x) * 2) % 63)) && (((((int)threadIdx.x) * 2) % 63) < 56)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + (((((int)threadIdx.x) * 2) / 63) * 49)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 63)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1)] = (((((7 <= (((((int)threadIdx.x) * 2) + 1) % 63)) && ((((((int)threadIdx.x) * 2) + 1) % 63) < 56)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1) / 63) * 49)) + rx_outer_outer) + (((((int)threadIdx.x) * 2) + 1) % 63)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 196)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 1) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 196) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) * 7)) + rx_outer_outer) + ((((int)thre [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 197)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 197) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) * 7)) + [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 392)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 2) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 392) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) * 7)) + rx_outer_outer) + ((((int)thre [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 393)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 393) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) * 7)) + [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 588)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 588) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 3) % 9) * 7)) + rx_outer_outer) + ((((int)thre [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 589)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 589) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9) * 7)) + [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 784)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 4) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 784) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 4) % 9) * 7)) + rx_outer_outer) + ((((int)thre [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 785)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 785) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9) * 7)) + [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 980)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 5) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 980) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 5) % 9) * 7)) + rx_outer_outer) + ((((int)thre [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 981)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 981) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9) * 7)) + [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1176)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1176) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 6) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1177)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1177) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1372)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 7) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1372) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 7) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1373)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1373) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1568)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 8) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1568) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 8) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1569)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1569) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1764)] = (((((7 <= ((((int)threadIdx.x) * 2) % 63)) && (((((int)threadIdx.x) * 2) % 63) < 56)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + (((((int)threadIdx.x) * 2) / 63) * 49)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 63)) + 1364)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1765)] = (((((7 <= (((((int)threadIdx.x) * 2) + 1) % 63)) && ((((((int)threadIdx.x) * 2) + 1) % 63) < 56)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1) / 63) * 49)) + rx_outer_outer) + (((((int)threadIdx.x) * 2) + 1) % 63)) + 1364)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1960)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 1) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1960) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 1961)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1961) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2156)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 2) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2156) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2157)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2157) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2352)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2352) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 3) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2353)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2353) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 3) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2548)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 4) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2548) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 4) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2549)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2549) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 4) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2744)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 5) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2744) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 5) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2745)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2745) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 5) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2940)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2940) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 6) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 2941)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 2941) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 6) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3136)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 7) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3136) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 7) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3137)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3137) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 7) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3332)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 8) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3332) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 8) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3333)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3333) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 8) % 9) * 7)) [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3528)] = (((((7 <= ((((int)threadIdx.x) * 2) % 63)) && (((((int)threadIdx.x) * 2) % 63) < 56)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + (((((int)threadIdx.x) * 2) / 63) * 49)) + rx_outer_outer) + ((((int)threadIdx.x) * 2) % 63)) + 2736)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3529)] = (((((7 <= (((((int)threadIdx.x) * 2) + 1) % 63)) && ((((((int)threadIdx.x) * 2) + 1) % 63) < 56)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 1) / 63) * 49)) + rx_outer_outer) + (((((int)threadIdx.x) * 2) + 1) % 63)) + 2736)] : 0.000000e+00f);
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3724)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 1) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3724) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 1) % 9) * 7)) + rx_outer_outer) + ((((int)th [...]
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3725)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3725) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 1) % 9) * 7)) [...]
+ if (((int)threadIdx.x) < 56) {
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3920)] = (((((1 <= ((((((int)threadIdx.x) * 2) / 7) + 2) % 9)) && (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3920) / 63) * 49)) + (((((((int)threadIdx.x) * 2) / 7) + 2) % 9) * 7)) + rx_outer_outer) + ((((int) [...]
+ }
+ if (((int)threadIdx.x) < 56) {
+ pad_temp_shared[((((int)threadIdx.x) * 2) + 3921)] = (((((1 <= (((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9)) && ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 2) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ((((((int)threadIdx.x) * 2) + 3921) / 63) * 49)) + ((((((((int)threadIdx.x) * 2) + 1) / 7) + 2) % 9) * 7) [...]
+ }
+ kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 73728) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 98)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 98) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 98) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 196)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 196) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 12)];
+ kernel_shared[(((int)threadIdx.x) + 294)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 294) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 102) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 392)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 392) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 24)];
+ kernel_shared[(((int)threadIdx.x) + 490)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 490) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 106) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 588)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 588) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 36)];
+ kernel_shared[(((int)threadIdx.x) + 686)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 686) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 110) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 784) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 48)];
+ kernel_shared[(((int)threadIdx.x) + 882)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 882) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 114) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 980)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 980) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 60)];
+ kernel_shared[(((int)threadIdx.x) + 1078)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1078) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 118) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1176)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1176) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 72)];
+ kernel_shared[(((int)threadIdx.x) + 1274)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1274) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 122) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1372)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1372) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 84)];
+ kernel_shared[(((int)threadIdx.x) + 1470)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1470) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 126) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1568)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1568) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 96)];
+ kernel_shared[(((int)threadIdx.x) + 1666)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1666) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 130) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1764)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1764) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 108)];
+ kernel_shared[(((int)threadIdx.x) + 1862)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1862) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 134) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1960)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 1960) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 120)];
+ kernel_shared[(((int)threadIdx.x) + 2058)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2058) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 138) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2156)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2156) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 132)];
+ kernel_shared[(((int)threadIdx.x) + 2254)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2254) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 142) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2352)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2352) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 144)];
+ kernel_shared[(((int)threadIdx.x) + 2450)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2450) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 146) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2548)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2548) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 156)];
+ kernel_shared[(((int)threadIdx.x) + 2646)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2646) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 150) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2744)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2744) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 168)];
+ kernel_shared[(((int)threadIdx.x) + 2842)] = kernel[(((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2842) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 154) % 192) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 2940)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 2940) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 180)];
+ if (((int)threadIdx.x) < 34) {
+ kernel_shared[(((int)threadIdx.x) + 3038)] = kernel[((((((((int)blockIdx.x) * 73728) + (((((int)threadIdx.x) + 3038) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 474)];
+ }
+ __syncthreads();
+ for (int rc_outer_inner = 0; rc_outer_inner < 4; ++rc_outer_inner) {
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[(((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48))]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 768)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1536)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2304)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 769)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1537)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2305)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 770)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1538)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2306)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 3)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 771)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1539)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2307)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 4)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 772)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1540)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2308)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 5)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 773)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1541)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2309)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 6)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 774)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1542)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2310)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 7)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 775)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1543)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2311)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 8)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 776)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1544)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2312)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 9)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 777)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1545)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2313)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 10)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 778)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1546)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2314)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 11)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 779)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1547)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2315)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 12)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 780)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1548)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2316)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 13)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 781)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1549)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2317)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 14)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 782)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1550)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2318)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 15)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 783)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1551)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2319)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 16)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 784)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1552)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2320)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 17)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 785)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1553)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2321)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 18)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 786)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1554)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2322)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 19)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 787)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1555)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2323)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 20)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 788)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1556)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2324)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 21)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 789)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1557)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2325)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 22)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 790)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1558)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2326)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 23)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 791)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1559)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2327)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 24)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 792)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1560)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2328)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 25)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 793)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1561)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2329)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 26)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 794)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1562)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2330)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 27)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 795)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1563)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2331)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 28)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 796)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1564)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2332)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 29)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 797)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1565)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2333)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 30)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 798)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1566)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2334)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 31)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 799)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1567)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2335)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 32)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 800)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1568)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2336)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 33)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 801)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1569)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2337)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 34)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 802)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1570)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2338)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 35)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 803)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1571)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2339)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 36)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 804)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1572)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2340)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 37)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 805)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1573)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2341)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 38)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 806)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1574)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2342)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 39)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 807)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1575)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2343)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 40)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 808)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1576)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2344)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 41)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 809)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1577)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2345)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 42)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 810)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1578)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2346)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 43)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 811)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1579)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2347)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 44)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 812)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1580)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2348)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 45)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 813)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1581)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2349)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 46)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 814)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1582)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2350)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 47)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 815)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1583)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2351)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 192)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 960)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1728)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49))] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2496)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 193)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 961)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1729)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 7)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2497)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 194)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 962)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1730)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 14)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2498)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 195)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 963)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1731)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 63)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2499)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 196)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 964)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1732)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 70)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2500)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 197)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 965)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1733)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 77)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2501)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 198)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 966)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1734)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 126)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2502)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 199)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 967)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1735)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 133)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2503)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 200)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 968)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1736)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 140)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2504)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 201)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 969)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1737)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 189)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2505)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 202)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 970)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1738)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 196)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2506)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 203)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 971)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1739)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 203)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2507)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 204)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 972)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1740)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 252)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2508)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 205)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 973)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1741)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 259)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2509)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 206)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 974)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1742)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 266)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2510)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 207)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 975)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1743)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 315)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2511)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 208)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 976)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1744)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 322)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2512)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 209)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 977)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1745)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 329)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2513)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 210)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 978)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1746)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 378)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2514)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 211)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 979)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1747)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 385)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2515)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 212)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 980)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1748)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 392)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2516)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 213)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 981)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1749)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 441)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2517)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 214)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 982)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1750)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 448)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2518)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 215)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 983)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1751)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 455)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2519)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 216)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 984)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1752)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 504)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2520)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 217)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 985)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1753)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 511)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2521)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 218)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 986)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1754)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 518)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2522)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 219)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 987)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1755)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 567)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2523)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 220)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 988)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1756)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 574)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2524)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 221)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 989)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1757)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 581)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2525)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 222)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 990)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1758)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 630)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2526)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 223)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 991)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1759)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 637)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2527)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 224)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 992)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1760)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 644)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2528)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 225)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 993)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1761)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 693)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2529)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 226)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 994)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1762)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 700)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2530)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 227)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 995)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1763)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 707)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2531)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 228)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 996)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1764)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 756)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2532)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 229)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 997)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1765)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 763)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2533)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 230)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 998)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1766)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 770)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2534)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 231)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 999)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1767)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 819)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2535)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 232)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1000)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1768)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 826)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2536)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 233)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1001)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1769)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 833)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2537)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 234)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1002)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1770)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 882)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2538)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 235)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1003)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1771)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 889)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2539)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 236)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1004)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1772)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 896)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2540)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 237)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1005)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1773)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 945)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2541)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 238)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1006)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1774)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 952)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2542)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 239)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1007)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 1775)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_outer_inner * 1008) + (((int)threadIdx.x) % 49)) + 959)] * kernel_shared[((((((int)threadIdx.x) / 49) * 384) + (rc_outer_inner * 48)) + 2543)]));
}
}
}
- for (int i3_inner = 0; i3_inner < 7; ++i3_inner) {
- compute[(((((int)blockIdx.x) * 784) + (((int)threadIdx.x) * 7)) + i3_inner)] = max((conv2d_nchw[i3_inner] + bias[((((int)blockIdx.x) * 16) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ for (int i1_inner = 0; i1_inner < 2; ++i1_inner) {
+ compute[((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49))] = max((conv2d_nchw[i1_inner] + bias[(((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49)) + 196)] = max((conv2d_nchw[(i1_inner + 2)] + bias[((((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner) + 4)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49)) + 392)] = max((conv2d_nchw[(i1_inner + 4)] + bias[((((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner) + 8)]), 0.000000e+00f);
+ compute[(((((((int)blockIdx.x) * 784) + ((((int)threadIdx.x) / 49) * 98)) + (i1_inner * 49)) + (((int)threadIdx.x) % 49)) + 588)] = max((conv2d_nchw[(i1_inner + 6)] + bias[((((((int)blockIdx.x) * 16) + ((((int)threadIdx.x) / 49) * 2)) + i1_inner) + 12)]), 0.000000e+00f);
}
}
</pre></div>
@@ -823,7 +1714,7 @@ In the example below we resume the status and do more 5 trials.</p>
Get devices for measurement successfully!
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 6 minutes 28.831 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 6 minutes 10.766 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-conv2d-layer-cuda-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/e3e540f3b477c0c52d8eb73e674e8ffd/tune_conv2d_layer_cuda.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">tune_conv2d_layer_cuda.py</span></code></a></p>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
index 20d2fc96cc..d0c664751d 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
@@ -921,7 +921,7 @@ so we can read the log file and load the best schedules.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 8.1199 8.1161 8.1291 8.1144 0.0065
+ 8.0999 8.0999 8.1066 8.0931 0.0055
</pre></div>
</div>
</div>
@@ -943,7 +943,7 @@ to learn how to use the RPC Tracker and RPC Server.
To use the RPC Tracker in auto-scheduler, replace the runner in <code class="code docutils literal notranslate"><span class="pre">TuningOptions</span></code>
with <a class="reference internal" href="../../reference/api/python/auto_scheduler.html#tvm.auto_scheduler.RPCRunner" title="tvm.auto_scheduler.RPCRunner"><code class="xref any py py-class docutils literal notranslate"><span class="pre">auto_scheduler.RPCRunner</span></code></a>.</p></li>
</ol>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 13.908 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 10.583 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-network-cuda-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/eafe360d52540634c9eea0fa89e804bd/tune_network_cuda.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">tune_network_cuda.py</span></code></a></p>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_network_x86.html b/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
index ac7ba36968..76e79f15b6 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
@@ -940,7 +940,7 @@ so we can read the log file and load the best schedules.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 770.8157 771.0289 771.6153 769.8030 0.7551
+ 762.8765 762.7099 764.5182 761.4014 1.2779
</pre></div>
</div>
</div>
@@ -962,7 +962,7 @@ to learn how to use the RPC Tracker and RPC Server.
To use the RPC Tracker in auto-scheduler, replace the runner in <code class="code docutils literal notranslate"><span class="pre">TuningOptions</span></code>
with <a class="reference internal" href="../../reference/api/python/auto_scheduler.html#tvm.auto_scheduler.RPCRunner" title="tvm.auto_scheduler.RPCRunner"><code class="xref any py py-class docutils literal notranslate"><span class="pre">auto_scheduler.RPCRunner</span></code></a>.</p></li>
</ol>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 47.644 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 42.680 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autoscheduler-tune-network-x86-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/e416b94ca1090b0897c0f6e0df95b911/tune_network_x86.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">tune_network_x86.py</span></code></a></p>
diff --git a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
index 3495b99212..427c63c3d6 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
@@ -637,25 +637,85 @@ class Module:
@T.prim_func
def main(placeholder: T.Buffer((128, 256), "float32"), placeholder_1: T.Buffer((4916, 16, 1), "float32"), placeholder_2: T.Buffer((4916,), "int32"), placeholder_3: T.Buffer((33,), "int32"), placeholder_4: T.Buffer((128, 512), "float32"), compute: T.Buffer((128, 512), "float32")):
T.func_attr({"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True})
- for i0_outer_i1_outer_fused in T.parallel(32):
- compute_1 = T.allocate([2048], "float32", "global")
- compute_2 = T.Buffer((2048,), data=compute_1)
+ for i0_outer_i1_outer_fused in T.parallel(128):
+ compute_1 = T.allocate([512], "float32", "global")
+ compute_2 = T.Buffer((512,), data=compute_1)
for i_outer_inner in range(32):
- for i_inner_init, j_init in T.grid(4, 16):
- compute_2[i_outer_inner * 64 + i_inner_init * 16 + j_init] = T.float32(0)
- for elem_idx, i_inner, j in T.grid(placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused], 4, 16):
+ cse_var_2: T.int32 = i0_outer_i1_outer_fused // 4
+ cse_var_1: T.int32 = i_outer_inner * 16
+ compute_2[cse_var_1] = T.float32(0)
+ compute_2[cse_var_1 + 1] = T.float32(0)
+ compute_2[cse_var_1 + 2] = T.float32(0)
+ compute_2[cse_var_1 + 3] = T.float32(0)
+ compute_2[cse_var_1 + 4] = T.float32(0)
+ compute_2[cse_var_1 + 5] = T.float32(0)
+ compute_2[cse_var_1 + 6] = T.float32(0)
+ compute_2[cse_var_1 + 7] = T.float32(0)
+ compute_2[cse_var_1 + 8] = T.float32(0)
+ compute_2[cse_var_1 + 9] = T.float32(0)
+ compute_2[cse_var_1 + 10] = T.float32(0)
+ compute_2[cse_var_1 + 11] = T.float32(0)
+ compute_2[cse_var_1 + 12] = T.float32(0)
+ compute_2[cse_var_1 + 13] = T.float32(0)
+ compute_2[cse_var_1 + 14] = T.float32(0)
+ compute_2[cse_var_1 + 15] = T.float32(0)
+ for elem_idx in range(placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
placeholder_5 = T.Buffer((33,), "int32", data=placeholder_3.data)
- if T.likely(elem_idx < placeholder_5[i0_outer_i1_outer_fused + 1] - placeholder_5[i0_outer_i1_outer_fused]):
- placeholder_6 = T.Buffer((78656,), data=placeholder_1.data)
- placeholder_7 = T.Buffer((32768,), data=placeholder.data)
- placeholder_8 = T.Buffer((4916,), "int32", data=placeholder_2.data)
- cse_var_1: T.int32 = i_outer_inner * 64 + i_inner * 16 + j
- compute_2[cse_var_1] = compute_2[cse_var_1] + placeholder_6[placeholder_5[i0_outer_i1_outer_fused] * 16 + elem_idx * 16 + j] * T.max(placeholder_7[i_outer_inner * 1024 + i_inner * 256 + placeholder_8[placeholder_5[i0_outer_i1_outer_fused] + elem_idx]], T.float32(0))
+ placeholder_6 = T.Buffer((78656,), data=placeholder_1.data)
+ placeholder_7 = T.Buffer((32768,), data=placeholder.data)
+ placeholder_8 = T.Buffer((4916,), "int32", data=placeholder_2.data)
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ compute_2[cse_var_1] = compute_2[cse_var_1] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx]], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_3: T.int32 = cse_var_1 + 1
+ compute_2[cse_var_3] = compute_2[cse_var_3] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 1] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx]], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_4: T.int32 = cse_var_1 + 2
+ compute_2[cse_var_4] = compute_2[cse_var_4] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 2] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx]], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_5: T.int32 = cse_var_1 + 3
+ compute_2[cse_var_5] = compute_2[cse_var_5] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 3] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx]], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_6: T.int32 = cse_var_1 + 4
+ compute_2[cse_var_6] = compute_2[cse_var_6] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 256], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_7: T.int32 = cse_var_1 + 5
+ compute_2[cse_var_7] = compute_2[cse_var_7] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 1] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 256], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_8: T.int32 = cse_var_1 + 6
+ compute_2[cse_var_8] = compute_2[cse_var_8] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 2] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 256], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_9: T.int32 = cse_var_1 + 7
+ compute_2[cse_var_9] = compute_2[cse_var_9] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 3] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 256], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_10: T.int32 = cse_var_1 + 8
+ compute_2[cse_var_10] = compute_2[cse_var_10] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 512], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_11: T.int32 = cse_var_1 + 9
+ compute_2[cse_var_11] = compute_2[cse_var_11] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 1] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 512], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_12: T.int32 = cse_var_1 + 10
+ compute_2[cse_var_12] = compute_2[cse_var_12] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 2] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 512], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_13: T.int32 = cse_var_1 + 11
+ compute_2[cse_var_13] = compute_2[cse_var_13] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 3] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 512], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_14: T.int32 = cse_var_1 + 12
+ compute_2[cse_var_14] = compute_2[cse_var_14] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 768], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_15: T.int32 = cse_var_1 + 13
+ compute_2[cse_var_15] = compute_2[cse_var_15] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 1] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 768], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_16: T.int32 = cse_var_1 + 14
+ compute_2[cse_var_16] = compute_2[cse_var_16] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 2] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 768], T.float32(0))
+ if T.likely(elem_idx < placeholder_5[cse_var_2 + 1] - placeholder_5[cse_var_2]):
+ cse_var_17: T.int32 = cse_var_1 + 15
+ compute_2[cse_var_17] = compute_2[cse_var_17] + placeholder_6[placeholder_5[cse_var_2] * 16 + elem_idx * 16 + i0_outer_i1_outer_fused % 4 * 4 + 3] * T.max(placeholder_7[i_outer_inner * 1024 + placeholder_8[placeholder_5[cse_var_2] + elem_idx] + 768], T.float32(0))
for i0_inner in range(128):
- cse_var_2: T.int32 = i0_inner * 512 + i0_outer_i1_outer_fused * 16
+ cse_var_18: T.int32 = i0_inner * 512 + i0_outer_i1_outer_fused * 4
compute_3 = T.Buffer((65536,), data=compute.data)
placeholder_5 = T.Buffer((65536,), data=placeholder_4.data)
- compute_3[cse_var_2:cse_var_2 + 16] = T.max(compute_2[i0_inner * 16:i0_inner * 16 + 16] + placeholder_5[cse_var_2:cse_var_2 + 16], T.Broadcast(T.float32(0), 16))
+ compute_3[cse_var_18:cse_var_18 + 4] = T.max(compute_2[i0_inner * 4:i0_inner * 4 + 4] + placeholder_5[cse_var_18:cse_var_18 + 4], T.Broadcast(T.float32(0), 4))
</pre></div>
</div>
</div>
@@ -689,7 +749,7 @@ class Module:
<span class="p">)</span>
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 1.485 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 2.434 ms
</pre></div>
</div>
<div class="admonition note">
diff --git a/docs/how_to/tune_with_autotvm/sg_execution_times.html b/docs/how_to/tune_with_autotvm/sg_execution_times.html
index 4e01a085dc..c9406e7dd8 100644
--- a/docs/how_to/tune_with_autotvm/sg_execution_times.html
+++ b/docs/how_to/tune_with_autotvm/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-tune-with-autotvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:40.527</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
+<p><strong>00:50.535</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -354,11 +354,11 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_conv2d_cuda.html#sphx-glr-how-to-tune-with-autotvm-tune-conv2d-cuda-py"><span class="std std-ref">Tuning High Performance Convolution on NVIDIA GPUs</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_conv2d_cuda.py</span></code>)</p></td>
-<td><p>00:40.491</p></td>
+<td><p>00:50.501</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_relay_x86.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-x86-py"><span class="std std-ref">Auto-tuning a Convolutional Network for x86 CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_x86.py</span></code>)</p></td>
-<td><p>00:00.022</p></td>
+<td><p>00:00.021</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_relay_cuda.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-cuda-py"><span class="std std-ref">Auto-tuning a Convolutional Network for NVIDIA GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_cuda.py</span></code>)</p></td>
@@ -366,11 +366,11 @@
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tune_relay_arm.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-arm-py"><span class="std std-ref">Auto-tuning a Convolutional Network for ARM CPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_arm.py</span></code>)</p></td>
-<td><p>00:00.005</p></td>
+<td><p>00:00.004</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="tune_relay_mobile_gpu.html#sphx-glr-how-to-tune-with-autotvm-tune-relay-mobile-gpu-py"><span class="std std-ref">Auto-tuning a Convolutional Network for Mobile GPU</span></a> (<code class="docutils literal notranslate"><span class="pre">tune_relay_mobile_gpu.py</span></code>)</p></td>
-<td><p>00:00.005</p></td>
+<td><p>00:00.004</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
index 02eb9c29e0..866385dc8e 100644
--- a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
+++ b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
@@ -695,9 +695,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 2, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 64, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5478804
-No: 2 GFLOPS: 114.52/114.52 result: MeasureResult(costs=(0.0020214115454545455,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.826885461807251, timestamp=1679626739.1992018) [('tile_f', [-1, 1, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6269615
-No: 3 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 1, 128]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2739872
+No: 2 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -819,8 +818,9 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 4, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4664196
-No: 4 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 8, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8372791
+No: 3 GFLOPS: 157.98/157.98 result: MeasureResult(costs=(0.0014653737777777778,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.0277414321899414, timestamp=1679627934.8009439) [('tile_f', [-1, 2, 2, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2394051
+No: 4 GFLOPS: 0.00/157.98 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -942,8 +942,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 512, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4573854
-No: 5 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 32, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 128]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4435362
+No: 5 GFLOPS: 0.00/157.98 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1065,8 +1065,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 64, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4178335
-No: 6 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 2, 32]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 256]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1540411
+No: 6 GFLOPS: 0.00/157.98 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1188,8 +1188,10 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 8, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 64]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7328010
-No: 7 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 128, 2, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6154791
+No: 7 GFLOPS: 20.05/157.98 result: MeasureResult(costs=(0.011545547454545457,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.0361216068267822, timestamp=1679627944.4602265) [('tile_f', [-1, 1, 2, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 8, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1463583
+No: 8 GFLOPS: 210.09/210.09 result: MeasureResult(costs=(0.0011019326483516483,), error_no=MeasureErrorNo.NO_ERROR, all_cost=6.238207817077637, timestamp=1679627945.3220174) [('tile_f', [-1, 2, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10020176
+No: 9 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1311,9 +1313,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 4, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9260858
-No: 8 GFLOPS: 2.80/114.52 result: MeasureResult(costs=(0.08256726824999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=8.456061601638794, timestamp=1679626752.5319996) [('tile_f', [-1, 64, 2, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5821050
-No: 9 GFLOPS: 0.00/114.52 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 2, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3385287
+No: 10 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1435,9 +1436,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 2, 16]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 64, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2606293
-No: 10 GFLOPS: 155.35/155.35 result: MeasureResult(costs=(0.0014902013766233768,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.0830042362213135, timestamp=1679626755.7926762) [('tile_f', [-1, 8, 4, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4879015
-No: 11 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 128, 4]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7724115
+No: 11 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1559,8 +1559,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 4, 1]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6578462
-No: 12 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 4, 16]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5932037
+No: 12 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1682,8 +1682,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 64, 1, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 64, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5055302
-No: 13 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 2, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7725665
+No: 13 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -1805,255 +1805,161 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 8, 16]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 4, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,342501
-No: 14 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 2, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3708513
+No: 14 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 742, in __call__
+ yield remote, remote.load_module(os.path.split(build_result.filename)[1])
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 706, in run_through_rpc
+ costs = time_f(*args).results
+ File "/workspace/python/tvm/runtime/module.py", line 357, in evaluator
+ blob = feval(*args)
File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
+ File "tvm/_ffi/_cython/./packed_func.pxi", line 262, in tvm._ffi._cy3.core.FuncCall
+ File "tvm/_ffi/_cython/./packed_func.pxi", line 251, in tvm._ffi._cy3.core.FuncCall3
File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
+ 4: TVMFuncCall
at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
+ 3: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
+ at ../include/tvm/runtime/packed_func.h:1217
+ 2: tvm::runtime::RPCWrappedFunc::operator()(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
+ at ../src/runtime/rpc/rpc_module.cc:129
+ 1: tvm::runtime::RPCClientSession::CallFunc(void*, TVMValue const*, int const*, int, std::function<void (tvm::runtime::TVMArgs)> const&)
+ at ../src/runtime/rpc/rpc_endpoint.cc:1012
+ 0: tvm::runtime::RPCEndpoint::CallFunc(void*, TVMValue const*, int const*, int, std::function<void (tvm::runtime::TVMArgs)>)
+ at ../src/runtime/rpc/rpc_endpoint.cc:804
+ File "../src/runtime/rpc/rpc_endpoint.cc", line 804
+TVMError:
+---------------------------------------------------------------
+An error occurred during the execution of TVM.
+For more information, please see: https://tvm.apache.org/docs/errors.html
+---------------------------------------------------------------
+ Check failed: (code == RPCCode::kReturn) is false: code=kShutdown
+
+During handling of the above exception, another exception occurred:
Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 1, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 256, 2]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5487563
-No: 15 GFLOPS: 0.00/155.35 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
- func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
- func = build(s, args, target=target, runtime=runtime)
- File "/workspace/python/tvm/driver/build_module.py", line 227, in build
- input_mod = lower(inputs, args, name=name, binds=binds)
- File "/workspace/python/tvm/driver/build_module.py", line 134, in lower
- return ffi.lower_schedule(inp, args, name, binds, simple_mode)
- File "tvm/_ffi/_cython/./packed_func.pxi", line 331, in tvm._ffi._cy3.core.PackedFuncBase.__call__
- File "tvm/_ffi/_cython/./packed_func.pxi", line 276, in tvm._ffi._cy3.core.FuncCall
- File "tvm/_ffi/_cython/./base.pxi", line 181, in tvm._ffi._cy3.core.CHECK_CALL
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 706, in run_through_rpc
+ costs = time_f(*args).results
+ File "/usr/lib/python3.7/contextlib.py", line 130, in __exit__
+ self.gen.throw(type, value, traceback)
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 746, in __call__
+ remote.remove(build_result.filename)
+ File "/workspace/python/tvm/rpc/client.py", line 144, in remove
+ self._remote_funcs["remove"] = self.get_function("tvm.rpc.server.remove")
+ File "/workspace/python/tvm/rpc/client.py", line 72, in get_function
+ return self._sess.get_function(name)
+ File "/workspace/python/tvm/runtime/module.py", line 171, in get_function
+ self.handle, c_str(name), ctypes.c_int(query_imports), ctypes.byref(ret_handle)
+ File "/workspace/python/tvm/_ffi/base.py", line 348, in check_call
+ raise get_last_ffi_error()
tvm._ffi.base.TVMError: Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
+ 52: 0xffffffffffffffff
+ 51: _start
+ 50: __libc_start_main
+ 49: _Py_UnixMain
+ 48: 0x0000000000650da0
+ 47: 0x0000000000650afa
+ 46: _PyFunction_FastCallDict
+ 45: _PyEval_EvalCodeWithName
+ 44: _PyEval_EvalFrameDefault
+ 43: _PyFunction_FastCallKeywords
+ 42: _PyEval_EvalCodeWithName
+ 41: _PyEval_EvalFrameDefault
+ 40: _PyMethodDef_RawFastCallKeywords
+ 39: 0x0000000000546369
+ 38: _PyEval_EvalCodeWithName
+ 37: _PyEval_EvalFrameDefault
+ 36: _PyFunction_FastCallKeywords
+ 35: _PyEval_EvalCodeWithName
+ 34: _PyEval_EvalFrameDefault
+ 33: _PyFunction_FastCallDict
+ 32: _PyEval_EvalCodeWithName
+ 31: _PyEval_EvalFrameDefault
+ 30: _PyObject_FastCallDict
+ 29: 0x00000000004c06e1
+ 28: _PyFunction_FastCallDict
+ 27: _PyEval_EvalFrameDefault
+ 26: _PyMethodDescr_FastCallKeywords
+ 25: 0x00000000005dcb58
+ 24: 0x00000000005dc83f
+ 23: 0x00000000004ba127
+ 22: _PyEval_EvalFrameDefault
+ 21: _PyFunction_FastCallKeywords
+ 20: _PyEval_EvalFrameDefault
+ 19: _PyFunction_FastCallKeywords
+ 18: _PyEval_EvalFrameDefault
+ 17: _PyFunction_FastCallKeywords
+ 16: _PyEval_EvalCodeWithName
+ 15: _PyEval_EvalFrameDefault
+ 14: 0x0000000000537c30
+ 13: _PyObject_FastCallKeywords
+ 12: 0x00007f8eb7416fa2
+ 11: _ctypes_callproc
+ 10: ffi_call
+ 9: ffi_call_unix64
+ 8: TVMModGetFunction
+ at ../src/runtime/c_runtime_api.cc:408
+ 7: tvm::runtime::ModuleNode::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, bool)
+ at ../src/runtime/module.cc:66
+ 6: tvm::runtime::RPCModuleNode::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, tvm::runtime::ObjectPtr<tvm::runtime::Object> const&)
+ at ../src/runtime/rpc/rpc_module.cc:185
+ 5: tvm::runtime::RPCClientSession::GetFunction(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
+ at ../src/runtime/rpc/rpc_endpoint.cc:1007
+ 4: tvm::runtime::TVMRetValue tvm::runtime::RPCEndpoint::SysCallRemote<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&>(tvm::runtime::RPCCode, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)
+ at ../src/runtime/rpc/rpc_endpoint.h:223
+ 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<int, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&>(int&&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) const
at ../include/tvm/runtime/packed_func.h:1621
2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
at ../include/tvm/runtime/packed_func.h:1217
1: Call
at ../include/tvm/runtime/packed_func.h:1213
0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel
+ at ../src/runtime/rpc/rpc_endpoint.cc:684
+ File "../src/runtime/rpc/rpc_endpoint.cc", line 684
+TVMError:
+---------------------------------------------------------------
+An error occurred during the execution of TVM.
+For more information, please see: https://tvm.apache.org/docs/errors.html
+---------------------------------------------------------------
+ Check failed: (code == RPCCode::kReturn) is false: code=1
Traceback (most recent call last):
- 24: TVMFuncCall
- at ../src/runtime/c_runtime_api.cc:477
- 23: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 22: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 21: operator()
- at ../include/tvm/runtime/packed_func.h:1734
- 20: unpack_call<tvm::IRModule, 5, tvm::<lambda(tvm::te::Schedule, const tvm::runtime::Array<tvm::runtime::ObjectRef>&, const tvm::runtime::String&, const tvm::runtime::Map<tvm::te::Tensor, tvm::tir::Buffer>&, bool)> >
- at ../include/tvm/runtime/packed_func.h:1674
- 19: run<>
- at ../include/tvm/runtime/packed_func.h:1634
- 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1634
- 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
- at ../include/tvm/runtime/packed_func.h:1649
- 13: operator()
- at ../src/driver/driver_api.cc:402
- 12: tvm::LowerSchedule(tvm::te::Schedule, tvm::runtime::Array<tvm::runtime::ObjectRef, void> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::unordered_map<tvm::te::Tensor, tvm::tir::Buffer, std::hash<tvm::te::Tensor>, std::equal_to<tvm::te::Tensor>, std::allocator<std::pair<tvm::te::Tensor const, tvm::tir::Buffer> > > const&, tvm::GlobalVarSupply, bool)
- at ../src/driver/driver_api.cc:388
- 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
- at ../src/driver/driver_api.cc:283
- 10: tvm::transform::Pass::operator()(tvm::IRModule) const
- at ../src/ir/transform.cc:258
- 9: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 8: tvm::transform::SequentialNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:451
- 7: tvm::transform::Pass::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/ir/transform.cc:274
- 6: tvm::tir::transform::PrimFuncPassNode::operator()(tvm::IRModule, tvm::transform::PassContext const&) const
- at ../src/tir/ir/transform.cc:100
- 5: tvm::runtime::TypedPackedFunc<tvm::tir::PrimFunc (tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext)>::operator()(tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext) const
- at ../include/tvm/runtime/packed_func.h:1753
- 4: tvm::tir::PrimFunc tvm::runtime::detail::typed_packed_call_dispatcher<tvm::tir::PrimFunc>::run<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::runtime::PackedFunc const&, tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&)
- at ../include/tvm/runtime/packed_func.h:1697
- 3: tvm::runtime::TVMRetValue tvm::runtime::PackedFunc::operator()<tvm::tir::PrimFunc, tvm::IRModule, tvm::transform::PassContext>(tvm::tir::PrimFunc&&, tvm::IRModule&&, tvm::transform::PassContext&&) const
- at ../include/tvm/runtime/packed_func.h:1621
- 2: tvm::runtime::PackedFuncObj::CallPacked(tvm::runtime::TVMArgs, tvm::runtime::TVMRetValue*) const
- at ../include/tvm/runtime/packed_func.h:1217
- 1: Call
- at ../include/tvm/runtime/packed_func.h:1213
- 0: operator()
- at ../src/runtime/c_runtime_api.cc:534
- File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
- raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 2, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 16]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1293394
-No: 16 GFLOPS: 259.48/259.48 result: MeasureResult(costs=(0.0008921655081967214,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.353220224380493, timestamp=1679626758.422771) [('tile_f', [-1, 1, 32, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9729540
-No: 17 GFLOPS: 0.00/259.48 result: Traceback (most recent call last):
+ 52: 0xffffffffffffffff
+ 51: _start
+ 50: __libc_start_main
+ 49: _Py_UnixMain
+ 48: 0x0000000000650da0
+ 47: 0x0000000000650afa
+ 46: _PyFunction_FastCallDict
+ 45: _PyEval_EvalCodeWithName
+ 44: _PyEval_EvalFrameDefault
+ 43: _PyFunction_FastCallKeywords
+ 42: _PyEval_EvalCodeWithName
+ 41: _PyEval_EvalFrameDefault
+ 40: _PyMethodDef_RawFastCallKeywords
+ 39: 0x0000000000546369
+ 38: _PyEval_EvalCodeWithName
+ 37: _PyEval_EvalFrameDefault
+ 36: _PyFunction_FastCallKeywords
+ 35: _PyEval_EvalCodeWithName
+ 34: _PyEval_EvalFrameDefault
+ 33: _PyFunction_FastCallDict
+ 32: _PyEval_EvalCodeWithName
+ 31: _PyEval_EvalFrameDefault
+ 30: _PyObject_FastCallDict
+ 29: 0x00000000004c06e1
+ 28: _PyFunction_FastCallDict
+ 27: _PyEval_EvalFrameDefault
+ 26: _PyMethodDescr_FastCallKeywords
+ 25: 0x00000000005dcb58
+ 24: 0x00000000005dc83f
+ 23: 0x00000000004ba127
+ 22: _PyEval_EvalFrameDefault
+ 21: _PyFunction_FastCallKeywords
+ 20: _PyEval_EvalFrameDefault
+ 19: _PyFunction_FastCall [('tile_f', [-1, 4, 1, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 8]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5909586
+No: 15 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2175,8 +2081,9 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 2, 16]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 16, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9016650
-No: 18 GFLOPS: 0.00/259.48 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 16, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 8, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,660600
+No: 16 GFLOPS: 1.07/210.09 result: MeasureResult(costs=(0.21648690775,), error_no=MeasureErrorNo.NO_ERROR, all_cost=5.180893421173096, timestamp=1679627955.7687864) [('tile_f', [-1, 32, 1, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,201001
+No: 17 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2298,8 +2205,8 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 128, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 32, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,53302
-No: 19 GFLOPS: 0.00/259.48 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 16, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 64, 8]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2052759
+No: 18 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2421,8 +2328,9 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 1, 128]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 16, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,82932
-No: 20 GFLOPS: 0.00/259.48 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 16, 1, 8]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 128, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9705660
+No: 19 GFLOPS: 10.49/210.09 result: MeasureResult(costs=(0.022066975833333335,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.2342774868011475, timestamp=1679627959.2058794) [('tile_f', [-1, 2, 1, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,8908206
+No: 20 GFLOPS: 0.00/210.09 result: Traceback (most recent call last):
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 592, in __call__
func, arg_info = _build_func_common(measure_input, self.runtime, **kwargs)
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 544, in _build_func_common
@@ -2544,7 +2452,7 @@ Traceback (most recent call last):
File "tvm/_ffi/_cython/./packed_func.pxi", line 56, in tvm._ffi._cy3.core.tvm_callback
File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 875, in verify_pass
raise InstantiationError("Skipped because of invalid gpu kernel")
-tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 4, 64]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9460496
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 64, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 128]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,5019615
</pre></div>
</div>
<p>Finally we can inspect the best config from log file, check correctness,
@@ -2583,9 +2491,9 @@ and measure running time.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Finish loading 20 records
Best config:
-[('tile_f', [-1, 1, 32, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9729540
+[('tile_f', [-1, 2, 1, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10020176
Finish loading 20 records
-Time cost of this operator: 0.001312
+Time cost of this operator: 0.001490
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-tune-with-autotvm-tune-conv2d-cuda-py">
diff --git a/docs/how_to/work_with_microtvm/micro_autotune.html b/docs/how_to/work_with_microtvm/micro_autotune.html
index d3b3a9dea4..79acdd2b49 100644
--- a/docs/how_to/work_with_microtvm/micro_autotune.html
+++ b/docs/how_to/work_with_microtvm/micro_autotune.html
@@ -649,10 +649,10 @@ the tuned operator.</p>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>########## Build without Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 314.2 98.697 (1, 2, 10, 10, 3) 2 1 [314.2]
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.151 0.99 (1, 6, 10, 10) 1 1 [3.151]
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.996 0.313 (1, 1, 10, 10, 3) 1 1 [0.996]
-Total_time - 318.347 - - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 319.3 98.755 (1, 2, 10, 10, 3) 2 1 [319.3]
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.057 0.945 (1, 6, 10, 10) 1 1 [3.057]
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.968 0.299 (1, 1, 10, 10, 3) 1 1 [0.968]
+Total_time - 323.325 - - - - -
</pre></div>
</div>
</div>
@@ -704,13 +704,13 @@ Total_time -
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>########## Build with Autotuning ##########
Node Name Ops Time(us) Time(%) Shape Inputs Outputs Measurements(us)
--------- --- -------- ------- ----- ------ ------- ----------------
-tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 136.8 98.078 (1, 6, 10, 10, 1) 2 1 [136.8]
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.821 1.305 (1, 6, 10, 10) 1 1 [1.821]
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.86 0.617 (1, 3, 10, 10, 1) 1 1 [0.86]
-Total_time - 139.481 - - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 173.8 97.599 (1, 6, 10, 10, 1) 2 1 [173.8]
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 2.708 1.521 (1, 6, 10, 10) 1 1 [2.708]
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 1.568 0.88 (1, 1, 10, 10, 3) 1 1 [1.568]
+Total_time - 178.076 - - - - -
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 28.427 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 21.031 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-autotune-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/9ccca8fd489a1486ac71b55a55c320c5/micro_autotune.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">micro_autotune.py</span></code></a></p>
diff --git a/docs/how_to/work_with_microtvm/micro_pytorch.html b/docs/how_to/work_with_microtvm/micro_pytorch.html
index 921d0ae5c3..87330cbff0 100644
--- a/docs/how_to/work_with_microtvm/micro_pytorch.html
+++ b/docs/how_to/work_with_microtvm/micro_pytorch.html
@@ -460,8 +460,8 @@ download a cat image and preprocess it to use as the model input.</p>
Downloading: "https://download.pytorch.org/models/quantized/mobilenet_v2_qnnpack_37f702c5.pth" to /workspace/.cache/torch/hub/checkpoints/mobilenet_v2_qnnpack_37f702c5.pth
0%| | 0.00/3.42M [00:00<?, ?B/s]
- 61%|###### | 2.09M/3.42M [00:00<00:00, 16.8MB/s]
-100%|##########| 3.42M/3.42M [00:00<00:00, 26.4MB/s]
+ 61%|###### | 2.09M/3.42M [00:00<00:00, 9.77MB/s]
+100%|##########| 3.42M/3.42M [00:00<00:00, 15.7MB/s]
/workspace/python/tvm/relay/frontend/pytorch_utils.py:47: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
return LooseVersion(torch_ver) > ver
/venv/apache-tvm-py3.7/lib/python3.7/site-packages/setuptools/_distutils/version.py:346: DeprecationWarning: distutils Version classes are deprecated. Use packaging.version instead.
@@ -587,7 +587,7 @@ via the host <cite>main.cc`</cite> or if a Zephyr emulated board is selected as
Torch top-1 id: 282, class name: tiger cat
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 22.970 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 17.499 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-pytorch-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/12b9ecc04c41abaa12022061771821d1/micro_pytorch.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">micro_pytorch.py</span></code></a></p>
diff --git a/docs/how_to/work_with_microtvm/micro_train.html b/docs/how_to/work_with_microtvm/micro_train.html
index b5a360f158..277d85ea56 100644
--- a/docs/how_to/work_with_microtvm/micro_train.html
+++ b/docs/how_to/work_with_microtvm/micro_train.html
@@ -528,7 +528,7 @@ take about <strong>2 minutes</strong> to download the Stanford Cars, while COCO
<a href="https://docs.python.org/3/library/shutil.html#shutil.move" title="shutil.move" class="sphx-glr-backref-module-shutil sphx-glr-backref-type-py-function"><span class="n">shutil</span><span class="o">.</span><span class="n">move</span></a><span class="p">(</span><span class="sa">f</span><span class="s2">"</span><span class="si">{</span><a href="https://docs.python.org/3/library/stdtypes.html#str" title="builtins.str" class="sphx-glr-backref-module-builtins sphx-glr-backref-typ [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>'/tmp/tmpr8om6t_s/images/random'
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>'/tmp/tmp9g460_h3/images/random'
</pre></div>
</div>
</div>
@@ -588,8 +588,8 @@ objects to other stuff? We can display some examples from our datasets using <co
<span class="n">plt</span><span class="o">.</span><span class="n">axis</span><span class="p">(</span><span class="s2">"off"</span><span class="p">)</span>
</pre></div>
</div>
-<img src="../../_images/sphx_glr_micro_train_001.png" srcset="../../_images/sphx_glr_micro_train_001.png" alt="[0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0]" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/tmp/tmpr8om6t_s/images/target contains 8144 images
-/tmp/tmpr8om6t_s/images/random contains 5000 images
+<img src="../../_images/sphx_glr_micro_train_001.png" srcset="../../_images/sphx_glr_micro_train_001.png" alt="[0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0]" class = "sphx-glr-single-img"/><div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/tmp/tmp9g460_h3/images/target contains 8144 images
+/tmp/tmp9g460_h3/images/random contains 5000 images
</pre></div>
</div>
</div>
@@ -701,13 +701,13 @@ the time on our validation set).</p>
</pre></div>
</div>
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Epoch 1/3
-328/328 - 42s - loss: 0.2768 - accuracy: 0.9157 - val_loss: 0.1518 - val_accuracy: 0.9517 - 42s/epoch - 128ms/step
+328/328 - 40s - loss: 0.2110 - accuracy: 0.9230 - val_loss: 0.1209 - val_accuracy: 0.9577 - 40s/epoch - 122ms/step
Epoch 2/3
-328/328 - 35s - loss: 0.1149 - accuracy: 0.9561 - val_loss: 0.1394 - val_accuracy: 0.9513 - 35s/epoch - 108ms/step
+328/328 - 35s - loss: 0.0927 - accuracy: 0.9640 - val_loss: 0.1028 - val_accuracy: 0.9630 - 35s/epoch - 106ms/step
Epoch 3/3
-328/328 - 35s - loss: 0.0679 - accuracy: 0.9748 - val_loss: 0.1042 - val_accuracy: 0.9653 - 35s/epoch - 107ms/step
+328/328 - 35s - loss: 0.0615 - accuracy: 0.9788 - val_loss: 0.1040 - val_accuracy: 0.9675 - 35s/epoch - 105ms/step
-<keras.callbacks.History object at 0x7f228f798990>
+<keras.callbacks.History object at 0x7fd033351990>
</pre></div>
</div>
</div>
@@ -971,7 +971,7 @@ as intended.</p>
<p>From here, we could modify the model to read live images from the camera - we have another
Arduino tutorial for how to do that <a class="reference external" href="https://github.com/guberti/tvm-arduino-demos/tree/master/examples/person_detection">on GitHub</a>. Alternatively, we could also
<a class="reference external" href="https://tvm.apache.org/docs/how_to/work_with_microtvm/micro_autotune.html">use TVM’s autotuning capabilities</a> to dramatically improve the model’s performance.</p>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 4 minutes 31.483 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 4 minutes 25.665 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-train-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/b52cec46baf4f78d6bcd94cbe269c8a6/micro_train.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">micro_train.py</span></code></a></p>
diff --git a/docs/how_to/work_with_microtvm/sg_execution_times.html b/docs/how_to/work_with_microtvm/sg_execution_times.html
index 45eb7a3ebe..7fe28eeb3b 100644
--- a/docs/how_to/work_with_microtvm/sg_execution_times.html
+++ b/docs/how_to/work_with_microtvm/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-work-with-microtvm-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>07:50.419</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
+<p><strong>07:30.160</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 82%" />
@@ -354,27 +354,27 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_train.html#sphx-glr-how-to-work-with-microtvm-micro-train-py"><span class="std std-ref">5. Training Vision Models for microTVM on Arduino</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_train.py</span></code>)</p></td>
-<td><p>04:31.483</p></td>
+<td><p>04:25.665</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_autotune.html#sphx-glr-how-to-work-with-microtvm-micro-autotune-py"><span class="std std-ref">6. Model Tuning with microTVM</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_autotune.py</span></code>)</p></td>
-<td><p>01:28.427</p></td>
+<td><p>01:21.031</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_pytorch.html#sphx-glr-how-to-work-with-microtvm-micro-pytorch-py"><span class="std std-ref">4. microTVM PyTorch Tutorial</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_pytorch.py</span></code>)</p></td>
-<td><p>01:22.970</p></td>
+<td><p>01:17.499</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_aot.html#sphx-glr-how-to-work-with-microtvm-micro-aot-py"><span class="std std-ref">3. microTVM Ahead-of-Time (AOT) Compilation</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_aot.py</span></code>)</p></td>
-<td><p>00:10.957</p></td>
+<td><p>00:10.356</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_custom_ide.html#sphx-glr-how-to-work-with-microtvm-micro-custom-ide-py"><span class="std std-ref">9. Bring microTVM to your own development environment</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_custom_ide.py</span></code>)</p></td>
-<td><p>00:08.651</p></td>
+<td><p>00:08.011</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_tflite.html#sphx-glr-how-to-work-with-microtvm-micro-tflite-py"><span class="std std-ref">2. microTVM TFLite Tutorial</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_tflite.py</span></code>)</p></td>
-<td><p>00:07.931</p></td>
+<td><p>00:07.598</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="micro_ethosu.html#sphx-glr-how-to-work-with-microtvm-micro-ethosu-py"><span class="std std-ref">7. Running TVM on bare metal Arm(R) Cortex(R)-M55 CPU and Ethos(TM)-U55 NPU with CMSIS-NN</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_ethosu.py</span></code>)</p></td>
diff --git a/docs/how_to/work_with_relay/sg_execution_times.html b/docs/how_to/work_with_relay/sg_execution_times.html
index c0156e2dc6..079939b37a 100644
--- a/docs/how_to/work_with_relay/sg_execution_times.html
+++ b/docs/how_to/work_with_relay/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-work-with-relay-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:38.525</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
+<p><strong>00:36.777</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -354,19 +354,19 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="using_pipeline_executor.html#sphx-glr-how-to-work-with-relay-using-pipeline-executor-py"><span class="std std-ref">Using Pipeline Executor in Relay</span></a> (<code class="docutils literal notranslate"><span class="pre">using_pipeline_executor.py</span></code>)</p></td>
-<td><p>00:33.774</p></td>
+<td><p>00:32.040</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="using_external_lib.html#sphx-glr-how-to-work-with-relay-using-external-lib-py"><span class="std std-ref">Using External Libraries in Relay</span></a> (<code class="docutils literal notranslate"><span class="pre">using_external_lib.py</span></code>)</p></td>
-<td><p>00:03.004</p></td>
+<td><p>00:02.928</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="build_gcn.html#sphx-glr-how-to-work-with-relay-build-gcn-py"><span class="std std-ref">Building a Graph Convolutional Network</span></a> (<code class="docutils literal notranslate"><span class="pre">build_gcn.py</span></code>)</p></td>
-<td><p>00:01.740</p></td>
+<td><p>00:01.802</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="using_relay_viz.html#sphx-glr-how-to-work-with-relay-using-relay-viz-py"><span class="std std-ref">Use Relay Visualizer to Visualize Relay</span></a> (<code class="docutils literal notranslate"><span class="pre">using_relay_viz.py</span></code>)</p></td>
-<td><p>00:00.007</p></td>
+<td><p>00:00.006</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/how_to/work_with_schedules/intrin_math.html b/docs/how_to/work_with_schedules/intrin_math.html
index 234e64ec7b..fd0491defa 100644
--- a/docs/how_to/work_with_schedules/intrin_math.html
+++ b/docs/how_to/work_with_schedules/intrin_math.html
@@ -540,7 +540,7 @@ The following example customizes CUDA lowering rule for <code class="code docuti
<a href="../../reference/api/python/ir.html#tvm.ir.register_intrin_lowering" title="tvm.ir.register_intrin_lowering" class="sphx-glr-backref-module-tvm-ir sphx-glr-backref-type-py-function"><span class="n">register_intrin_lowering</span></a><span class="p">(</span><span class="s2">"tir.exp"</span><span class="p">,</span> <span class="n">target</span><span class="o">=</span><span class="s2">"cuda"</span><span class="p">,</span> <span class="n">f</span><span class="o">= [...]
</pre></div>
</div>
-<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span><function my_cuda_math_rule at 0x7f1edd6a7710>
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span><function my_cuda_math_rule at 0x7fced0208ef0>
</pre></div>
</div>
<p>Register the rule to TVM with override option to override existing rule.
diff --git a/docs/how_to/work_with_schedules/sg_execution_times.html b/docs/how_to/work_with_schedules/sg_execution_times.html
index 934baadb32..db340c954f 100644
--- a/docs/how_to/work_with_schedules/sg_execution_times.html
+++ b/docs/how_to/work_with_schedules/sg_execution_times.html
@@ -345,7 +345,7 @@
<div class="section" id="computation-times">
<span id="sphx-glr-how-to-work-with-schedules-sg-execution-times"></span><h1>Computation times<a class="headerlink" href="#computation-times" title="Permalink to this headline">¶</a></h1>
-<p><strong>00:08.721</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
+<p><strong>00:08.728</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -354,35 +354,35 @@
</colgroup>
<tbody>
<tr class="row-odd"><td><p><a class="reference internal" href="intrin_math.html#sphx-glr-how-to-work-with-schedules-intrin-math-py"><span class="std std-ref">Intrinsics and Math Functions</span></a> (<code class="docutils literal notranslate"><span class="pre">intrin_math.py</span></code>)</p></td>
-<td><p>00:05.745</p></td>
+<td><p>00:05.962</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tensorize.html#sphx-glr-how-to-work-with-schedules-tensorize-py"><span class="std std-ref">Use Tensorize to Leverage Hardware Intrinsics</span></a> (<code class="docutils literal notranslate"><span class="pre">tensorize.py</span></code>)</p></td>
-<td><p>00:01.410</p></td>
+<td><p>00:01.266</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="reduction.html#sphx-glr-how-to-work-with-schedules-reduction-py"><span class="std std-ref">Reduction</span></a> (<code class="docutils literal notranslate"><span class="pre">reduction.py</span></code>)</p></td>
-<td><p>00:00.642</p></td>
+<td><p>00:00.619</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="scan.html#sphx-glr-how-to-work-with-schedules-scan-py"><span class="std std-ref">Scan and Recurrent Kernel</span></a> (<code class="docutils literal notranslate"><span class="pre">scan.py</span></code>)</p></td>
-<td><p>00:00.631</p></td>
+<td><p>00:00.603</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="extern_op.html#sphx-glr-how-to-work-with-schedules-extern-op-py"><span class="std std-ref">External Tensor Functions</span></a> (<code class="docutils literal notranslate"><span class="pre">extern_op.py</span></code>)</p></td>
-<td><p>00:00.138</p></td>
+<td><p>00:00.130</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tedd.html#sphx-glr-how-to-work-with-schedules-tedd-py"><span class="std std-ref">Use Tensor Expression Debug Display (TEDD) for Visualization</span></a> (<code class="docutils literal notranslate"><span class="pre">tedd.py</span></code>)</p></td>
-<td><p>00:00.066</p></td>
+<td><p>00:00.063</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="schedule_primitives.html#sphx-glr-how-to-work-with-schedules-schedule-primitives-py"><span class="std std-ref">Schedule Primitives in TVM</span></a> (<code class="docutils literal notranslate"><span class="pre">schedule_primitives.py</span></code>)</p></td>
-<td><p>00:00.058</p></td>
+<td><p>00:00.055</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="tuple_inputs.html#sphx-glr-how-to-work-with-schedules-tuple-inputs-py"><span class="std std-ref">Compute and Reduce with Tuple Inputs</span></a> (<code class="docutils literal notranslate"><span class="pre">tuple_inputs.py</span></code>)</p></td>
-<td><p>00:00.032</p></td>
+<td><p>00:00.029</p></td>
<td><p>0.0 MB</p></td>
</tr>
</tbody>
diff --git a/docs/objects.inv b/docs/objects.inv
index b00712a008..c79fd7bdde 100644
Binary files a/docs/objects.inv and b/docs/objects.inv differ
diff --git a/docs/reference/api/doxygen/annotated.html b/docs/reference/api/doxygen/annotated.html
index d3dfaec958..27bbfe0425 100644
--- a/docs/reference/api/doxygen/annotated.html
+++ b/docs/reference/api/doxygen/annotated.html
@@ -876,14 +876,14 @@ $(function() {
<tr id="row_1_10_28_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BroadcastNode.html" target="_self">BroadcastNode</a></td><td class="desc">Create a vector where all the elements are value </td></tr>
<tr id="row_1_10_29_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Buffer.html" target="_self">Buffer</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Buffer.html" title="Buffer is a symbolic n-darray structure. It is a composition of primitive symbolic types...">Buffer</a> is a symbolic n-darray structure. It is a [...]
<tr id="row_1_10_30_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferLoad.html" target="_self">BufferLoad</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1BufferLoadNode.html" title="Load value from the high dimension buffer. ">BufferLoadNode</a> </td></tr>
-<tr id="row_1_10_31_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferLoadNode.html" target="_self">BufferLoadNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Load.html" title="Managed reference to LoadNode. ">Load</a> value from the high dimension buffer </td></tr>
+<tr id="row_1_10_31_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferLoadNode.html" target="_self">BufferLoadNode</a></td><td class="desc">Load value from the high dimension buffer </td></tr>
<tr id="row_1_10_32_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferNode.html" target="_self">BufferNode</a></td><td class="desc">Node to represent a buffer </td></tr>
<tr id="row_1_10_33_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferRealize.html" target="_self">BufferRealize</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1BufferRealizeNode.html" title="Annotate the region where the buffer need to be read and write in the body. We only need to allocat [...]
<tr id="row_1_10_34_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferRealizeNode.html" target="_self">BufferRealizeNode</a></td><td class="desc">Annotate the region where the buffer need to be read and write in the body. We only need to allocate the space for the corresponding region </td></tr>
<tr id="row_1_10_35_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferRegion.html" target="_self">BufferRegion</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1BufferRegionNode.html" title="Representing the region of multi-dimensional buffer access. ">BufferRegionNode</a> </td></tr>
<tr id="row_1_10_36_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferRegionNode.html" target="_self">BufferRegionNode</a></td><td class="desc">Representing the region of multi-dimensional buffer access </td></tr>
<tr id="row_1_10_37_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferStore.html" target="_self">BufferStore</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1BufferStoreNode.html" title="Store value to the high dimension buffer. ">BufferStoreNode</a> </td></tr>
-<tr id="row_1_10_38_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferStoreNode.html" target="_self">BufferStoreNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Store.html" title="Managed reference to StoreNode. ">Store</a> value to the high dimension buffer </td></tr>
+<tr id="row_1_10_38_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1BufferStoreNode.html" target="_self">BufferStoreNode</a></td><td class="desc">Store value to the high dimension buffer </td></tr>
<tr id="row_1_10_39_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Call.html" target="_self">Call</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1CallNode.html" title="Call node. ">CallNode</a> </td></tr>
<tr id="row_1_10_40_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1CallNode.html" target="_self">CallNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Call.html" title="Managed reference to CallNode. ">Call</a> node </td></tr>
<tr id="row_1_10_41_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Cast.html" target="_self">Cast</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1CastNode.html" title="Cast value from one data type to another. ">CastNode</a> </td></tr>
@@ -941,86 +941,82 @@ $(function() {
<tr id="row_1_10_93_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LetNode.html" target="_self">LetNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Let.html" title="Managed reference to LetNode. ">Let</a> binding. Bind var to value then evaluate body </td></tr>
<tr id="row_1_10_94_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LetStmt.html" target="_self">LetStmt</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1LetStmtNode.html" title="Let binding, bind var to value, then run body. ">LetStmtNode</a> </td></tr>
<tr id="row_1_10_95_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LetStmtNode.html" target="_self">LetStmtNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Let.html" title="Managed reference to LetNode. ">Let</a> binding, bind var to value, then run body </td></tr>
-<tr id="row_1_10_96_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Load.html" target="_self">Load</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1LoadNode.html" title="Load the value from buffer_var. ">LoadNode</a> </td></tr>
-<tr id="row_1_10_97_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LoadNode.html" target="_self">LoadNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Load.html" title="Managed reference to LoadNode. ">Load</a> the value from buffer_var </td></tr>
-<tr id="row_1_10_98_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LoopRV.html" target="_self">LoopRV</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1LoopRVNode.html" title="A random variable that evaluates to a TensorIR for loop. ">LoopRVNode</a> </td></tr>
-<tr id="row_1_10_99_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LoopRVNode.html" target="_self">LoopRVNode</a></td><td class="desc">A random variable that evaluates to a TensorIR for loop </td></tr>
-<tr id="row_1_10_100_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LT.html" target="_self">LT</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1LTNode.html" title="a < b ">LTNode</a> </td></tr>
-<tr id="row_1_10_101_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LTNode.html" target="_self">LTNode</a></td><td class="desc">< b </td></tr>
-<tr id="row_1_10_102_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MatchBufferRegion.html" target="_self">MatchBufferRegion</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1MatchBufferRegionNode.html" title="Match introduces a constraint that the source buffer region can be remapped to the dat [...]
-<tr id="row_1_10_103_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MatchBufferRegionNode.html" target="_self">MatchBufferRegionNode</a></td><td class="desc">Match introduces a constraint that the source buffer region can be remapped to the data layout specified by the buffer field. The constraint can be checked in later part of [...]
-<tr id="row_1_10_104_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Max.html" target="_self">Max</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1MaxNode.html" title="max(a, b) ">MaxNode</a> </td></tr>
-<tr id="row_1_10_105_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MaxNode.html" target="_self">MaxNode</a></td><td class="desc">Max(a, b) </td></tr>
-<tr id="row_1_10_106_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="structtvm_1_1tir_1_1MemCpyDetails.html" target="_self">MemCpyDetails</a></td><td class="desc">Helper struct for return value of IdentifyMemCpy </td></tr>
-<tr id="row_1_10_107_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Min.html" target="_self">Min</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1MinNode.html" title="min(a, b) ">MinNode</a> </td></tr>
-<tr id="row_1_10_108_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MinNode.html" target="_self">MinNode</a></td><td class="desc">Min(a, b) </td></tr>
-<tr id="row_1_10_109_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Mod.html" target="_self">Mod</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ModNode.html" title="a % b in the C semnatics. ">ModNode</a> </td></tr>
-<tr id="row_1_10_110_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ModNode.html" target="_self">ModNode</a></td><td class="desc">% b in the C semnatics </td></tr>
-<tr id="row_1_10_111_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Mul.html" target="_self">Mul</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1MulNode.html" title="a * b ">MulNode</a> </td></tr>
-<tr id="row_1_10_112_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MulNode.html" target="_self">MulNode</a></td><td class="desc"><ul>
+<tr id="row_1_10_96_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LoopRV.html" target="_self">LoopRV</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1LoopRVNode.html" title="A random variable that evaluates to a TensorIR for loop. ">LoopRVNode</a> </td></tr>
+<tr id="row_1_10_97_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LoopRVNode.html" target="_self">LoopRVNode</a></td><td class="desc">A random variable that evaluates to a TensorIR for loop </td></tr>
+<tr id="row_1_10_98_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LT.html" target="_self">LT</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1LTNode.html" title="a < b ">LTNode</a> </td></tr>
+<tr id="row_1_10_99_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1LTNode.html" target="_self">LTNode</a></td><td class="desc">< b </td></tr>
+<tr id="row_1_10_100_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MatchBufferRegion.html" target="_self">MatchBufferRegion</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1MatchBufferRegionNode.html" title="Match introduces a constraint that the source buffer region can be remapped to the dat [...]
+<tr id="row_1_10_101_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MatchBufferRegionNode.html" target="_self">MatchBufferRegionNode</a></td><td class="desc">Match introduces a constraint that the source buffer region can be remapped to the data layout specified by the buffer field. The constraint can be checked in later part of [...]
+<tr id="row_1_10_102_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Max.html" target="_self">Max</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1MaxNode.html" title="max(a, b) ">MaxNode</a> </td></tr>
+<tr id="row_1_10_103_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MaxNode.html" target="_self">MaxNode</a></td><td class="desc">Max(a, b) </td></tr>
+<tr id="row_1_10_104_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="structtvm_1_1tir_1_1MemCpyDetails.html" target="_self">MemCpyDetails</a></td><td class="desc">Helper struct for return value of IdentifyMemCpy </td></tr>
+<tr id="row_1_10_105_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Min.html" target="_self">Min</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1MinNode.html" title="min(a, b) ">MinNode</a> </td></tr>
+<tr id="row_1_10_106_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MinNode.html" target="_self">MinNode</a></td><td class="desc">Min(a, b) </td></tr>
+<tr id="row_1_10_107_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Mod.html" target="_self">Mod</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ModNode.html" title="a % b in the C semnatics. ">ModNode</a> </td></tr>
+<tr id="row_1_10_108_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ModNode.html" target="_self">ModNode</a></td><td class="desc">% b in the C semnatics </td></tr>
+<tr id="row_1_10_109_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Mul.html" target="_self">Mul</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1MulNode.html" title="a * b ">MulNode</a> </td></tr>
+<tr id="row_1_10_110_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1MulNode.html" target="_self">MulNode</a></td><td class="desc"><ul>
<li>b </li>
</ul>
</td></tr>
-<tr id="row_1_10_113_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1NE.html" target="_self">NE</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1NENode.html" title="a != b ">NENode</a> </td></tr>
-<tr id="row_1_10_114_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1NENode.html" target="_self">NENode</a></td><td class="desc">!= b </td></tr>
-<tr id="row_1_10_115_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Not.html" target="_self">Not</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1NotNode.html" title="!a ">NotNode</a> </td></tr>
-<tr id="row_1_10_116_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1NotNode.html" target="_self">NotNode</a></td><td class="desc">!a </td></tr>
-<tr id="row_1_10_117_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Or.html" target="_self">Or</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1OrNode.html" title="a || b ">OrNode</a> </td></tr>
-<tr id="row_1_10_118_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1OrNode.html" target="_self">OrNode</a></td><td class="desc">|| b </td></tr>
-<tr id="row_1_10_119_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Prefetch.html" target="_self">Prefetch</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1PrefetchNode.html" title="A prefetch hint for a buffer. ">PrefetchNode</a> </td></tr>
-<tr id="row_1_10_120_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1PrefetchNode.html" target="_self">PrefetchNode</a></td><td class="desc">A prefetch hint for a buffer </td></tr>
-<tr id="row_1_10_121_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1PrimFunc.html" target="_self">PrimFunc</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1PrimFuncNode.html" title="Primitive functions that contains TIR statements. ">PrimFuncNode</a> </td></tr>
-<tr id="row_1_10_122_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1PrimFuncNode.html" target="_self">PrimFuncNode</a></td><td class="desc">Primitive functions that contains TIR statements </td></tr>
-<tr id="row_1_10_123_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerLoad.html" target="_self">ProducerLoad</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ProducerLoadNode.html" title="Load value from the result produced by the producer. ">ProducerLoadNode</a> </td></tr>
-<tr id="row_1_10_124_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerLoadNode.html" target="_self">ProducerLoadNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Load.html" title="Managed reference to LoadNode. ">Load</a> value from the result produced by the producer </td></tr>
-<tr id="row_1_10_125_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerRealize.html" target="_self">ProducerRealize</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ProducerRealizeNode.html" title="Annotate the bounds where the data produced by the producer need to be written and read in b [...]
-<tr id="row_1_10_126_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerRealizeNode.html" target="_self">ProducerRealizeNode</a></td><td class="desc">Annotate the bounds where the data produced by the producer need to be written and read in body. We will need to allocate space for the corresponding regions </td></tr>
-<tr id="row_1_10_127_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerStore.html" target="_self">ProducerStore</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ProducerStoreNode.html" title="Store value into mult-dimensional array that will be read by the consumer of the producer. ">Produ [...]
-<tr id="row_1_10_128_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerStoreNode.html" target="_self">ProducerStoreNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Store.html" title="Managed reference to StoreNode. ">Store</a> value into mult-dimensional array that will be read by the consumer of the prod [...]
-<tr id="row_1_10_129_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Ramp.html" target="_self">Ramp</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1RampNode.html" title="Construct a vector with lanes elements where its i-th element equals base + i * stride. This is useful to construct a index f [...]
-<tr id="row_1_10_130_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1RampNode.html" target="_self">RampNode</a></td><td class="desc">Construct a vector with lanes elements where its i-th element equals base + i * stride. This is useful to construct a index for a continuous vector load </td></tr>
-<tr id="row_1_10_131_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Reduce.html" target="_self">Reduce</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ReduceNode.html" title="Reduction operator operator. ">ReduceNode</a> </td></tr>
-<tr id="row_1_10_132_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ReduceNode.html" target="_self">ReduceNode</a></td><td class="desc">Reduction operator operator </td></tr>
-<tr id="row_1_10_133_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Schedule.html" target="_self">Schedule</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ScheduleNode.html" title="The user-facing schedule class. ">ScheduleNode</a> </td></tr>
-<tr id="row_1_10_134_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ScheduleNode.html" target="_self">ScheduleNode</a></td><td class="desc">The user-facing schedule class </td></tr>
-<tr id="row_1_10_135_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ScheduleState.html" target="_self">ScheduleState</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ScheduleStateNode.html" title="The state of scheduling, which exposes a Replace method as the primary interface for all the sched [...]
-<tr id="row_1_10_136_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ScheduleStateNode.html" target="_self">ScheduleStateNode</a></td><td class="desc">The state of scheduling, which exposes a <code>Replace</code> method as the primary interface for all the scheduling primitives to manipulate the TensorIR </td></tr>
-<tr id="row_1_10_137_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Select.html" target="_self">Select</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1SelectNode.html" title="return true_value if condition is true, otherwise return false_value. ">SelectNode</a> </td></tr>
-<tr id="row_1_10_138_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SelectNode.html" target="_self">SelectNode</a></td><td class="desc">Return true_value if condition is true, otherwise return false_value </td></tr>
-<tr id="row_1_10_139_" class="even" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span id="arr_1_10_139_" class="arrow" onclick="toggleFolder('1_10_139_')">►</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SeqStmt.html" target="_self">SeqStmt</a></td><td class="desc">Sequence statement </td></tr>
-<tr id="row_1_10_139_0_" class="even" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SeqStmt_1_1Flattener.html" target="_self">Flattener</a></td><td class="desc">Helper class to flatten sequence of arguments into Array </td></tr>
-<tr id="row_1_10_140_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SeqStmtNode.html" target="_self">SeqStmtNode</a></td><td class="desc">The container of seq statement. Represent a sequence of statements </td></tr>
-<tr id="row_1_10_141_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Shuffle.html" target="_self">Shuffle</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ShuffleNode.html" title="Shuffle instruction. vec = concat(vectors) result = (vec[indices[0]], vec[indices[1]] ...">ShuffleNode</a> </td></tr>
-<tr id="row_1_10_142_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ShuffleNode.html" target="_self">ShuffleNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Shuffle.html" title="Managed reference to ShuffleNode. ">Shuffle</a> instruction. vec = concat(vectors) result = (vec[indices[0]], vec[indices[1]] ...) </ [...]
-<tr id="row_1_10_143_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SizeVar.html" target="_self">SizeVar</a></td><td class="desc">Named variable represents a tensor index size </td></tr>
-<tr id="row_1_10_144_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SizeVarNode.html" target="_self">SizeVarNode</a></td><td class="desc">A variable node represent a tensor index size, whose value must be non-negative </td></tr>
-<tr id="row_1_10_145_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Stmt.html" target="_self">Stmt</a></td><td class="desc">Container of all statements </td></tr>
-<tr id="row_1_10_146_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtExprMutator.html" target="_self">StmtExprMutator</a></td><td class="desc">Mutator that recursively mutates stmts and exprs on them </td></tr>
-<tr id="row_1_10_147_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtExprVisitor.html" target="_self">StmtExprVisitor</a></td><td class="desc">Visitor that recursively visit stmts and exprs on them </td></tr>
-<tr id="row_1_10_148_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtFunctor.html" target="_self">StmtFunctor</a></td><td class="desc">Same as <a class="el" href="classtvm_1_1tir_1_1ExprFunctor.html" title="A dynamical functor that dispatches on in the first Expr argument. You can use this as a more powerfu...">ExprFunctor</a> [...]
-<tr id="row_1_10_149_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtFunctor_3_01R_07const_01Stmt_01_6n_00_01Args_8_8_8_01args_08_4.html" target="_self">StmtFunctor< R(const Stmt &n, Args... args)></a></td><td class="desc"></td></tr>
-<tr id="row_1_10_150_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtMutator.html" target="_self">StmtMutator</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1StmtMutator.html" title="StmtMutator that mutates the statements. ">StmtMutator</a> that mutates the statements </td></tr>
-<tr id="row_1_10_151_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtNode.html" target="_self">StmtNode</a></td><td class="desc">Base node of all statements </td></tr>
-<tr id="row_1_10_152_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtSRef.html" target="_self">StmtSRef</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1StmtSRefNode.html" title="An object that refers to schedulable elements (block/for-loop) in TensorIR, aka "sref". ">StmtSRefNode< [...]
-<tr id="row_1_10_153_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtSRefNode.html" target="_self">StmtSRefNode</a></td><td class="desc">An object that refers to schedulable elements (block/for-loop) in TensorIR, aka "sref" </td></tr>
-<tr id="row_1_10_154_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtVisitor.html" target="_self">StmtVisitor</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1StmtVisitor.html" title="StmtVisitor. ">StmtVisitor</a> </td></tr>
-<tr id="row_1_10_155_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Store.html" target="_self">Store</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1StoreNode.html" title="Store value to the buffer. ">StoreNode</a> </td></tr>
-<tr id="row_1_10_156_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StoreNode.html" target="_self">StoreNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Store.html" title="Managed reference to StoreNode. ">Store</a> value to the buffer </td></tr>
-<tr id="row_1_10_157_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StringImm.html" target="_self">StringImm</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1StringImmNode.html" title="String constants, only used in asserts. ">StringImmNode</a> </td></tr>
-<tr id="row_1_10_158_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StringImmNode.html" target="_self">StringImmNode</a></td><td class="desc">String constants, only used in asserts </td></tr>
-<tr id="row_1_10_159_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Sub.html" target="_self">Sub</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1SubNode.html" title="a - b ">SubNode</a> </td></tr>
-<tr id="row_1_10_160_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SubNode.html" target="_self">SubNode</a></td><td class="desc"><ul>
+<tr id="row_1_10_111_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1NE.html" target="_self">NE</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1NENode.html" title="a != b ">NENode</a> </td></tr>
+<tr id="row_1_10_112_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1NENode.html" target="_self">NENode</a></td><td class="desc">!= b </td></tr>
+<tr id="row_1_10_113_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Not.html" target="_self">Not</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1NotNode.html" title="!a ">NotNode</a> </td></tr>
+<tr id="row_1_10_114_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1NotNode.html" target="_self">NotNode</a></td><td class="desc">!a </td></tr>
+<tr id="row_1_10_115_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Or.html" target="_self">Or</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1OrNode.html" title="a || b ">OrNode</a> </td></tr>
+<tr id="row_1_10_116_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1OrNode.html" target="_self">OrNode</a></td><td class="desc">|| b </td></tr>
+<tr id="row_1_10_117_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Prefetch.html" target="_self">Prefetch</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1PrefetchNode.html" title="A prefetch hint for a buffer. ">PrefetchNode</a> </td></tr>
+<tr id="row_1_10_118_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1PrefetchNode.html" target="_self">PrefetchNode</a></td><td class="desc">A prefetch hint for a buffer </td></tr>
+<tr id="row_1_10_119_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1PrimFunc.html" target="_self">PrimFunc</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1PrimFuncNode.html" title="Primitive functions that contains TIR statements. ">PrimFuncNode</a> </td></tr>
+<tr id="row_1_10_120_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1PrimFuncNode.html" target="_self">PrimFuncNode</a></td><td class="desc">Primitive functions that contains TIR statements </td></tr>
+<tr id="row_1_10_121_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerLoad.html" target="_self">ProducerLoad</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ProducerLoadNode.html" title="Load value from the result produced by the producer. ">ProducerLoadNode</a> </td></tr>
+<tr id="row_1_10_122_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerLoadNode.html" target="_self">ProducerLoadNode</a></td><td class="desc">Load value from the result produced by the producer </td></tr>
+<tr id="row_1_10_123_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerRealize.html" target="_self">ProducerRealize</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ProducerRealizeNode.html" title="Annotate the bounds where the data produced by the producer need to be written and read in b [...]
+<tr id="row_1_10_124_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerRealizeNode.html" target="_self">ProducerRealizeNode</a></td><td class="desc">Annotate the bounds where the data produced by the producer need to be written and read in body. We will need to allocate space for the corresponding regions </td></tr>
+<tr id="row_1_10_125_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerStore.html" target="_self">ProducerStore</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ProducerStoreNode.html" title="Store value into mult-dimensional array that will be read by the consumer of the producer. ">Produ [...]
+<tr id="row_1_10_126_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ProducerStoreNode.html" target="_self">ProducerStoreNode</a></td><td class="desc">Store value into mult-dimensional array that will be read by the consumer of the producer </td></tr>
+<tr id="row_1_10_127_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Ramp.html" target="_self">Ramp</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1RampNode.html" title="Construct a vector with lanes elements where its i-th element equals base + i * stride. This is useful to construct a index f [...]
+<tr id="row_1_10_128_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1RampNode.html" target="_self">RampNode</a></td><td class="desc">Construct a vector with lanes elements where its i-th element equals base + i * stride. This is useful to construct a index for a continuous vector load </td></tr>
+<tr id="row_1_10_129_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Reduce.html" target="_self">Reduce</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ReduceNode.html" title="Reduction operator operator. ">ReduceNode</a> </td></tr>
+<tr id="row_1_10_130_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ReduceNode.html" target="_self">ReduceNode</a></td><td class="desc">Reduction operator operator </td></tr>
+<tr id="row_1_10_131_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Schedule.html" target="_self">Schedule</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ScheduleNode.html" title="The user-facing schedule class. ">ScheduleNode</a> </td></tr>
+<tr id="row_1_10_132_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ScheduleNode.html" target="_self">ScheduleNode</a></td><td class="desc">The user-facing schedule class </td></tr>
+<tr id="row_1_10_133_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ScheduleState.html" target="_self">ScheduleState</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ScheduleStateNode.html" title="The state of scheduling, which exposes a Replace method as the primary interface for all the sched [...]
+<tr id="row_1_10_134_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ScheduleStateNode.html" target="_self">ScheduleStateNode</a></td><td class="desc">The state of scheduling, which exposes a <code>Replace</code> method as the primary interface for all the scheduling primitives to manipulate the TensorIR </td></tr>
+<tr id="row_1_10_135_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Select.html" target="_self">Select</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1SelectNode.html" title="return true_value if condition is true, otherwise return false_value. ">SelectNode</a> </td></tr>
+<tr id="row_1_10_136_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SelectNode.html" target="_self">SelectNode</a></td><td class="desc">Return true_value if condition is true, otherwise return false_value </td></tr>
+<tr id="row_1_10_137_" class="even" style="display:none;"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span id="arr_1_10_137_" class="arrow" onclick="toggleFolder('1_10_137_')">►</span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SeqStmt.html" target="_self">SeqStmt</a></td><td class="desc">Sequence statement </td></tr>
+<tr id="row_1_10_137_0_" class="even" style="display:none;"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SeqStmt_1_1Flattener.html" target="_self">Flattener</a></td><td class="desc">Helper class to flatten sequence of arguments into Array </td></tr>
+<tr id="row_1_10_138_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SeqStmtNode.html" target="_self">SeqStmtNode</a></td><td class="desc">The container of seq statement. Represent a sequence of statements </td></tr>
+<tr id="row_1_10_139_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Shuffle.html" target="_self">Shuffle</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1ShuffleNode.html" title="Shuffle instruction. vec = concat(vectors) result = (vec[indices[0]], vec[indices[1]] ...">ShuffleNode</a> </td></tr>
+<tr id="row_1_10_140_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1ShuffleNode.html" target="_self">ShuffleNode</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1Shuffle.html" title="Managed reference to ShuffleNode. ">Shuffle</a> instruction. vec = concat(vectors) result = (vec[indices[0]], vec[indices[1]] ...) </ [...]
+<tr id="row_1_10_141_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SizeVar.html" target="_self">SizeVar</a></td><td class="desc">Named variable represents a tensor index size </td></tr>
+<tr id="row_1_10_142_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SizeVarNode.html" target="_self">SizeVarNode</a></td><td class="desc">A variable node represent a tensor index size, whose value must be non-negative </td></tr>
+<tr id="row_1_10_143_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Stmt.html" target="_self">Stmt</a></td><td class="desc">Container of all statements </td></tr>
+<tr id="row_1_10_144_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtExprMutator.html" target="_self">StmtExprMutator</a></td><td class="desc">Mutator that recursively mutates stmts and exprs on them </td></tr>
+<tr id="row_1_10_145_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtExprVisitor.html" target="_self">StmtExprVisitor</a></td><td class="desc">Visitor that recursively visit stmts and exprs on them </td></tr>
+<tr id="row_1_10_146_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtFunctor.html" target="_self">StmtFunctor</a></td><td class="desc">Same as <a class="el" href="classtvm_1_1tir_1_1ExprFunctor.html" title="A dynamical functor that dispatches on in the first Expr argument. You can use this as a more powerfu...">ExprFunctor</a> [...]
+<tr id="row_1_10_147_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtFunctor_3_01R_07const_01Stmt_01_6n_00_01Args_8_8_8_01args_08_4.html" target="_self">StmtFunctor< R(const Stmt &n, Args... args)></a></td><td class="desc"></td></tr>
+<tr id="row_1_10_148_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtMutator.html" target="_self">StmtMutator</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1StmtMutator.html" title="StmtMutator that mutates the statements. ">StmtMutator</a> that mutates the statements </td></tr>
+<tr id="row_1_10_149_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtNode.html" target="_self">StmtNode</a></td><td class="desc">Base node of all statements </td></tr>
+<tr id="row_1_10_150_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtSRef.html" target="_self">StmtSRef</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1StmtSRefNode.html" title="An object that refers to schedulable elements (block/for-loop) in TensorIR, aka "sref". ">StmtSRefNode< [...]
+<tr id="row_1_10_151_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtSRefNode.html" target="_self">StmtSRefNode</a></td><td class="desc">An object that refers to schedulable elements (block/for-loop) in TensorIR, aka "sref" </td></tr>
+<tr id="row_1_10_152_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StmtVisitor.html" target="_self">StmtVisitor</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1StmtVisitor.html" title="StmtVisitor. ">StmtVisitor</a> </td></tr>
+<tr id="row_1_10_153_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StringImm.html" target="_self">StringImm</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1StringImmNode.html" title="String constants, only used in asserts. ">StringImmNode</a> </td></tr>
+<tr id="row_1_10_154_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1StringImmNode.html" target="_self">StringImmNode</a></td><td class="desc">String constants, only used in asserts </td></tr>
+<tr id="row_1_10_155_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Sub.html" target="_self">Sub</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1SubNode.html" title="a - b ">SubNode</a> </td></tr>
+<tr id="row_1_10_156_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1SubNode.html" target="_self">SubNode</a></td><td class="desc"><ul>
<li>b </li>
</ul>
</td></tr>
-<tr id="row_1_10_161_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1TensorIntrin.html" target="_self">TensorIntrin</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1TensorIntrinNode.html" title="Tensor intrinsics for tensorization. ">TensorIntrinNode</a> </td></tr>
-<tr id="row_1_10_162_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1TensorIntrinNode.html" target="_self">TensorIntrinNode</a></td><td class="desc">Tensor intrinsics for tensorization </td></tr>
-<tr id="row_1_10_163_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Trace.html" target="_self">Trace</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1TraceNode.html" title="An execution trace of a scheduling program. ">TraceNode</a> </td></tr>
-<tr id="row_1_10_164_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1TraceNode.html" target="_self">TraceNode</a></td><td class="desc">An execution trace of a scheduling program </td></tr>
-<tr id="row_1_10_165_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Var.html" target="_self">Var</a></td><td class="desc">Named variable in TIR </td></tr>
-<tr id="row_1_10_166_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1VarNode.html" target="_self">VarNode</a></td><td class="desc">A variable node in the IR </td></tr>
-<tr id="row_1_10_167_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1While.html" target="_self">While</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1WhileNode.html" title="A While loop. ">WhileNode</a> </td></tr>
-<tr id="row_1_10_168_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1WhileNode.html" target="_self">WhileNode</a></td><td class="desc">A <a class="el" href="classtvm_1_1tir_1_1While.html" title="Managed reference to WhileNode. ">While</a> loop </td></tr>
+<tr id="row_1_10_157_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1TensorIntrin.html" target="_self">TensorIntrin</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1TensorIntrinNode.html" title="Tensor intrinsics for tensorization. ">TensorIntrinNode</a> </td></tr>
+<tr id="row_1_10_158_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1TensorIntrinNode.html" target="_self">TensorIntrinNode</a></td><td class="desc">Tensor intrinsics for tensorization </td></tr>
+<tr id="row_1_10_159_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Trace.html" target="_self">Trace</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1TraceNode.html" title="An execution trace of a scheduling program. ">TraceNode</a> </td></tr>
+<tr id="row_1_10_160_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1TraceNode.html" target="_self">TraceNode</a></td><td class="desc">An execution trace of a scheduling program </td></tr>
+<tr id="row_1_10_161_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1Var.html" target="_self">Var</a></td><td class="desc">Named variable in TIR </td></tr>
+<tr id="row_1_10_162_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1VarNode.html" target="_self">VarNode</a></td><td class="desc">A variable node in the IR </td></tr>
+<tr id="row_1_10_163_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1While.html" target="_self">While</a></td><td class="desc">Managed reference to <a class="el" href="classtvm_1_1tir_1_1WhileNode.html" title="A While loop. ">WhileNode</a> </td></tr>
+<tr id="row_1_10_164_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="classtvm_1_1tir_1_1WhileNode.html" target="_self">WhileNode</a></td><td class="desc">A <a class="el" href="classtvm_1_1tir_1_1While.html" title="Managed reference to WhileNode. ">While</a> loop </td></tr>
<tr id="row_1_11_" class="even" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_11_" class="arrow" onclick="toggleFolder('1_11_')">►</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi.html" target="_self">topi</a></td><td class="desc"></td></tr>
<tr id="row_1_11_0_" class="even" style="display:none;"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">C</span></span><a class="el" href="structtvm_1_1topi_1_1EinsumEquation.html" target="_self">EinsumEquation</a></td><td class="desc"></td></tr>
<tr id="row_1_12_" class="even" style="display:none;"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_12_" class="arrow" onclick="toggleFolder('1_12_')">►</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
diff --git a/docs/reference/api/doxygen/builtin_8h.html b/docs/reference/api/doxygen/builtin_8h.html
index 3081fa41fb..d8b70ee946 100644
--- a/docs/reference/api/doxygen/builtin_8h.html
+++ b/docs/reference/api/doxygen/builtin_8h.html
@@ -329,10 +329,10 @@ Functions</h2></td></tr>
<tr class="memdesc:a9603510b7eff067c253eb340310c2996"><td class="mdescLeft"> </td><td class="mdescRight">Create an Nd memory allocation with storage scope. <a href="namespacetvm_1_1tir_1_1builtin.html#a9603510b7eff067c253eb340310c2996">More...</a><br /></td></tr>
<tr class="separator:a9603510b7eff067c253eb340310c2996"><td class="memSeparator" colspan="2"> </td></tr>
<tr class="memitem:a6ea87b06a1eb1050b7503ef6cc1bab34"><td class="memItemLeft" align="right" valign="top">const Op & </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1tir_1_1builtin.html#a6ea87b06a1eb1050b7503ef6cc1bab34">tvm::tir::builtin::texture2d_store</a> ()</td></tr>
-<tr class="memdesc:a6ea87b06a1eb1050b7503ef6cc1bab34"><td class="mdescLeft"> </td><td class="mdescRight"><a class="el" href="classtvm_1_1tir_1_1Store.html" title="Managed reference to StoreNode. ">Store</a> to texture 2d memory. <a href="namespacetvm_1_1tir_1_1builtin.html#a6ea87b06a1eb1050b7503ef6cc1bab34">More...</a><br /></td></tr>
+<tr class="memdesc:a6ea87b06a1eb1050b7503ef6cc1bab34"><td class="mdescLeft"> </td><td class="mdescRight">Store to texture 2d memory. <a href="namespacetvm_1_1tir_1_1builtin.html#a6ea87b06a1eb1050b7503ef6cc1bab34">More...</a><br /></td></tr>
<tr class="separator:a6ea87b06a1eb1050b7503ef6cc1bab34"><td class="memSeparator" colspan="2"> </td></tr>
<tr class="memitem:ae52feb3e7828262a1b8a31062642ae36"><td class="memItemLeft" align="right" valign="top">const Op & </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1tir_1_1builtin.html#ae52feb3e7828262a1b8a31062642ae36">tvm::tir::builtin::texture2d_load</a> ()</td></tr>
-<tr class="memdesc:ae52feb3e7828262a1b8a31062642ae36"><td class="mdescLeft"> </td><td class="mdescRight"><a class="el" href="classtvm_1_1tir_1_1Load.html" title="Managed reference to LoadNode. ">Load</a> from texture 2d memory. <a href="namespacetvm_1_1tir_1_1builtin.html#ae52feb3e7828262a1b8a31062642ae36">More...</a><br /></td></tr>
+<tr class="memdesc:ae52feb3e7828262a1b8a31062642ae36"><td class="mdescLeft"> </td><td class="mdescRight">Load from texture 2d memory. <a href="namespacetvm_1_1tir_1_1builtin.html#ae52feb3e7828262a1b8a31062642ae36">More...</a><br /></td></tr>
<tr class="separator:ae52feb3e7828262a1b8a31062642ae36"><td class="memSeparator" colspan="2"> </td></tr>
<tr class="memitem:afd453e860a10562ac2692473edde6707"><td class="memItemLeft" align="right" valign="top">const Op & </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1tir_1_1builtin.html#afd453e860a10562ac2692473edde6707">tvm::tir::builtin::dma_copy</a> ()</td></tr>
<tr class="memdesc:afd453e860a10562ac2692473edde6707"><td class="mdescLeft"> </td><td class="mdescRight">Initiate a non-blocking DMA copy from source to destination. <a href="namespacetvm_1_1tir_1_1builtin.html#afd453e860a10562ac2692473edde6707">More...</a><br /></td></tr>
diff --git a/docs/reference/api/doxygen/c__backend__api_8h_source.html b/docs/reference/api/doxygen/c__backend__api_8h_source.html
index 26e79d4868..86397bd94b 100644
--- a/docs/reference/api/doxygen/c__backend__api_8h_source.html
+++ b/docs/reference/api/doxygen/c__backend__api_8h_source.html
@@ -66,13 +66,13 @@ $(function() {
<div class="title">c_backend_api.h</div> </div>
</div><!--header-->
<div class="contents">
-<a href="c__backend__api_8h.html">Go to the documentation of this file.</a><div class="fragment"><div class="line"><a name="l00001"></a><span class="lineno"> 1</span> <span class="comment">/*</span></div><div class="line"><a name="l00002"></a><span class="lineno"> 2</span> <span class="comment"> * Licensed to the Apache Software Foundation (ASF) under one</span></div><div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="comment"> * or [...]
+<a href="c__backend__api_8h.html">Go to the documentation of this file.</a><div class="fragment"><div class="line"><a name="l00001"></a><span class="lineno"> 1</span> <span class="comment">/*</span></div><div class="line"><a name="l00002"></a><span class="lineno"> 2</span> <span class="comment"> * Licensed to the Apache Software Foundation (ASF) under one</span></div><div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="comment"> * or [...]
<div class="ttc" id="c__backend__api_8h_html_a56a654a9aeba2f1ccf3e10918dd88ec5"><div class="ttname"><a href="c__backend__api_8h.html#a56a654a9aeba2f1ccf3e10918dd88ec5">TVMBackendGetFuncFromEnv</a></div><div class="ttdeci">int TVMBackendGetFuncFromEnv(void *mod_node, const char *func_name, TVMFunctionHandle *out)</div><div class="ttdoc">Backend function for modules to get function from its environment mod_node (its imports and global fu...</div></div>
<div class="ttc" id="c__backend__api_8h_html_ae70bd3ee026eb55b438ada05b08f0ce8"><div class="ttname"><a href="c__backend__api_8h.html#ae70bd3ee026eb55b438ada05b08f0ce8">TVMBackendRunOnce</a></div><div class="ttdeci">int TVMBackendRunOnce(void **handle, int(*f)(void *), void *cdata, int nbytes)</div><div class="ttdoc">Simple static initialization function. Run f once and set handle to be not null. This function is mai...</div></div>
<div class="ttc" id="c__backend__api_8h_html_a07eaf7d1b748d99aa7715c7adbdea231"><div class="ttname"><a href="c__backend__api_8h.html#a07eaf7d1b748d99aa7715c7adbdea231">TVMBackendAllocWorkspace</a></div><div class="ttdeci">void * TVMBackendAllocWorkspace(int device_type, int device_id, uint64_t nbytes, int dtype_code_hint, int dtype_bits_hint)</div><div class="ttdoc">Backend function to allocate temporal workspace. </div></div>
<div class="ttc" id="unionTVMValue_html"><div class="ttname"><a href="unionTVMValue.html">TVMValue</a></div><div class="ttdoc">Union type of values being passed through API and function calls. </div><div class="ttdef"><b>Definition:</b> c_runtime_api.h:208</div></div>
<div class="ttc" id="structTVMParallelGroupEnv_html"><div class="ttname"><a href="structTVMParallelGroupEnv.html">TVMParallelGroupEnv</a></div><div class="ttdoc">Environment for TVM parallel task. </div><div class="ttdef"><b>Definition:</b> c_backend_api.h:119</div></div>
-<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a61b1ef1047fb722a4e5ec2167c9963d7"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a61b1ef1047fb722a4e5ec2167c9963d7">tvm::tir::attr::device_id</a></div><div class="ttdeci">constexpr const char * device_id</div><div class="ttdoc">The allocation device for global malloc in host. </div><div class="ttdef"><b>Definition:</b> stmt.h:1405</div></div>
+<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a61b1ef1047fb722a4e5ec2167c9963d7"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a61b1ef1047fb722a4e5ec2167c9963d7">tvm::tir::attr::device_id</a></div><div class="ttdeci">constexpr const char * device_id</div><div class="ttdoc">The allocation device for global malloc in host. </div><div class="ttdef"><b>Definition:</b> stmt.h:1339</div></div>
<div class="ttc" id="c__runtime__api_8h_html_acf57d257a6e0841d84ebbd2a339d183e"><div class="ttname"><a href="c__runtime__api_8h.html#acf57d257a6e0841d84ebbd2a339d183e">TVMFunctionHandle</a></div><div class="ttdeci">void * TVMFunctionHandle</div><div class="ttdoc">Handle to packed function handle. </div><div class="ttdef"><b>Definition:</b> c_runtime_api.h:229</div></div>
<div class="ttc" id="c__backend__api_8h_html_a6ff8662943d0f003d55d9046cd24daf8"><div class="ttname"><a href="c__backend__api_8h.html#a6ff8662943d0f003d55d9046cd24daf8">TVMBackendParallelLaunch</a></div><div class="ttdeci">int TVMBackendParallelLaunch(FTVMParallelLambda flambda, void *cdata, int num_task)</div><div class="ttdoc">Backend function for running parallel jobs. </div></div>
<div class="ttc" id="c__backend__api_8h_html_a0c57deb5acb9338ec778d91bd6e42191"><div class="ttname"><a href="c__backend__api_8h.html#a0c57deb5acb9338ec778d91bd6e42191">TVMBackendRegisterSystemLibSymbol</a></div><div class="ttdeci">int TVMBackendRegisterSystemLibSymbol(const char *name, void *ptr)</div><div class="ttdoc">Backend function to register system-wide library symbol. </div></div>
diff --git a/docs/reference/api/doxygen/c__runtime__api_8h_source.html b/docs/reference/api/doxygen/c__runtime__api_8h_source.html
index b64487542b..13489bdec4 100644
--- a/docs/reference/api/doxygen/c__runtime__api_8h_source.html
+++ b/docs/reference/api/doxygen/c__runtime__api_8h_source.html
@@ -70,7 +70,7 @@ $(function() {
<div class="ttc" id="unionTVMValue_html_aa1c40fa9e74fbf97541fd9735062c4cc"><div class="ttname"><a href="unionTVMValue.html#aa1c40fa9e74fbf97541fd9735062c4cc">TVMValue::v_int64</a></div><div class="ttdeci">int64_t v_int64</div><div class="ttdef"><b>Definition:</b> c_runtime_api.h:209</div></div>
<div class="ttc" id="c__runtime__api_8h_html_a07954203342f2787acf988c4c351d9c3"><div class="ttname"><a href="c__runtime__api_8h.html#a07954203342f2787acf988c4c351d9c3">TVMPackedCFunc</a></div><div class="ttdeci">int(* TVMPackedCFunc)(TVMValue *args, int *type_codes, int num_args, TVMRetValueHandle ret, void *resource_handle)</div><div class="ttdoc">C type of packed function. </div><div class="ttdef"><b>Definition:</b> c_runtime_api.h:369</div></div>
<div class="ttc" id="c__runtime__api_8h_html_a477111f9accd70633dc5f2e7139b6cf4"><div class="ttname"><a href="c__runtime__api_8h.html#a477111f9accd70633dc5f2e7139b6cf4">TVMAPISetLastError</a></div><div class="ttdeci">void TVMAPISetLastError(const char *msg)</div><div class="ttdoc">Used for implementing C API function. Set last error message before return. </div></div>
-<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a7e4e7cd47471a9089022214d63d24206"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a7e4e7cd47471a9089022214d63d24206">tvm::tir::attr::device_type</a></div><div class="ttdeci">constexpr const char * device_type</div><div class="ttdoc">The device type. </div><div class="ttdef"><b>Definition:</b> stmt.h:1407</div></div>
+<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a7e4e7cd47471a9089022214d63d24206"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a7e4e7cd47471a9089022214d63d24206">tvm::tir::attr::device_type</a></div><div class="ttdeci">constexpr const char * device_type</div><div class="ttdoc">The device type. </div><div class="ttdef"><b>Definition:</b> stmt.h:1341</div></div>
<div class="ttc" id="c__runtime__api_8h_html_a57cbccb14c35a0e62dbc1b911188fcefa474961a2b52fee7b33e529015162e884"><div class="ttname"><a href="c__runtime__api_8h.html#a57cbccb14c35a0e62dbc1b911188fcefa474961a2b52fee7b33e529015162e884">TVMDeviceExtType_End</a></div><div class="ttdef"><b>Definition:</b> c_runtime_api.h:125</div></div>
<div class="ttc" id="c__runtime__api_8h_html_af420f0d74004615c03bb40270ad5d489"><div class="ttname"><a href="c__runtime__api_8h.html#af420f0d74004615c03bb40270ad5d489">TVMFuncFree</a></div><div class="ttdeci">int TVMFuncFree(TVMFunctionHandle func)</div><div class="ttdoc">Free the function when it is no longer needed. </div></div>
<div class="ttc" id="c__runtime__api_8h_html_a591e48e52098965e235657dab56bc096"><div class="ttname"><a href="c__runtime__api_8h.html#a591e48e52098965e235657dab56bc096">TVMArrayFree</a></div><div class="ttdeci">int TVMArrayFree(TVMArrayHandle handle)</div><div class="ttdoc">Free the TVM Array. </div></div>
@@ -102,7 +102,7 @@ $(function() {
<div class="ttc" id="c__runtime__api_8h_html_a8a01e20ffd5784465df0759c950154b5"><div class="ttname"><a href="c__runtime__api_8h.html#a8a01e20ffd5784465df0759c950154b5">TVMFuncCall</a></div><div class="ttdeci">int TVMFuncCall(TVMFunctionHandle func, TVMValue *arg_values, int *type_codes, int num_args, TVMValue *ret_val, int *ret_type_code)</div><div class="ttdoc">Call a Packed TVM Function. </div></div>
<div class="ttc" id="c__runtime__api_8h_html_aa91f776ed41a36790409f78725f81419"><div class="ttname"><a href="c__runtime__api_8h.html#aa91f776ed41a36790409f78725f81419">TVMStreamStreamSynchronize</a></div><div class="ttdeci">int TVMStreamStreamSynchronize(int device_type, int device_id, TVMStreamHandle src, TVMStreamHandle dst)</div><div class="ttdoc">Synchronize two streams of execution. </div></div>
<div class="ttc" id="c__runtime__api_8h_html_a6fee438ab81fbf4235375d9f17f36d6f"><div class="ttname"><a href="c__runtime__api_8h.html#a6fee438ab81fbf4235375d9f17f36d6f">TVMCbArgToReturn</a></div><div class="ttdeci">int TVMCbArgToReturn(TVMValue *value, int *code)</div><div class="ttdoc">Inplace translate callback argument value to return value. This is only needed for non-POD arguments...</div></div>
-<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a61b1ef1047fb722a4e5ec2167c9963d7"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a61b1ef1047fb722a4e5ec2167c9963d7">tvm::tir::attr::device_id</a></div><div class="ttdeci">constexpr const char * device_id</div><div class="ttdoc">The allocation device for global malloc in host. </div><div class="ttdef"><b>Definition:</b> stmt.h:1405</div></div>
+<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a61b1ef1047fb722a4e5ec2167c9963d7"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a61b1ef1047fb722a4e5ec2167c9963d7">tvm::tir::attr::device_id</a></div><div class="ttdeci">constexpr const char * device_id</div><div class="ttdoc">The allocation device for global malloc in host. </div><div class="ttdef"><b>Definition:</b> stmt.h:1339</div></div>
<div class="ttc" id="c__runtime__api_8h_html_ab98d6b66089da8b33b662ccdb243b26b"><div class="ttname"><a href="c__runtime__api_8h.html#ab98d6b66089da8b33b662ccdb243b26b">TVMFuncRemoveGlobal</a></div><div class="ttdeci">int TVMFuncRemoveGlobal(const char *name)</div><div class="ttdoc">Remove a global function. </div></div>
<div class="ttc" id="c__runtime__api_8h_html_ad3bd42da244a0e32ac82d7428e01a010"><div class="ttname"><a href="c__runtime__api_8h.html#ad3bd42da244a0e32ac82d7428e01a010">TVMFuncGetGlobal</a></div><div class="ttdeci">int TVMFuncGetGlobal(const char *name, TVMFunctionHandle *out)</div><div class="ttdoc">Get a global function. </div></div>
<div class="ttc" id="c__runtime__api_8h_html_a190e81769e805cca153514137a66e793a463445886242b0357db59f74241ce028"><div class="ttname"><a href="c__runtime__api_8h.html#a190e81769e805cca153514137a66e793a463445886242b0357db59f74241ce028">kTVMExtEnd</a></div><div class="ttdef"><b>Definition:</b> c_runtime_api.h:198</div></div>
diff --git a/docs/reference/api/doxygen/classes.html b/docs/reference/api/doxygen/classes.html
index 23142af82a..22297a1982 100644
--- a/docs/reference/api/doxygen/classes.html
+++ b/docs/reference/api/doxygen/classes.html
@@ -65,260 +65,260 @@ $(function() {
<div class="qindex"><a class="qindex" href="#letter_a">a</a> | <a class="qindex" href="#letter_b">b</a> | <a class="qindex" href="#letter_c">c</a> | <a class="qindex" href="#letter_d">d</a> | <a class="qindex" href="#letter_e">e</a> | <a class="qindex" href="#letter_f">f</a> | <a class="qindex" href="#letter_g">g</a> | <a class="qindex" href="#letter_h">h</a> | <a class="qindex" href="#letter_i">i</a> |& [...]
<table class="classindex">
<tr><td rowspan="2" valign="bottom"><a name="letter_a"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  a  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv3DTransposeAttrs.html">Conv3DTransposeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IntImm.html">IntImm</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternVar.html">PatternVar</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::r [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv3DWinogradAttrs.html">Conv3DWinogradAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IntImmNode.html">IntImmNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternVarNode.html">PatternVarNode</a> (<a class="el" href="namespacetvm_1_1relay [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AccessAnalyzer.html">AccessAnalyzer</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ConvGemmWeightTransformAttrs.html">ConvGemmWeightTransformAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1In [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AccessAnalyzerNode.html">AccessAnalyzerNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ConvWinogradWeightTransformAttrs.html">ConvWinogradWeightTransformAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtv [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AdaptivePool1DAttrs.html">AdaptivePool1DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CorrelationAttrs.html">CorrelationAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntSetNode.html">IntSetNode</a> (<a class="e [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AdaptivePool2DAttrs.html">AdaptivePool2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1CostModel.html">CostModel</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1IRBuilder.html">I [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AdaptivePool3DAttrs.html">AdaptivePool3DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CostModel.html">CostModel</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1IRBuilderFrame [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Add.html">Add</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1CostModelNode.html">CostModelNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1IRBuilderFrameNode.html">IRBuilderFrameNode</a> [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AddNode.html">AddNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CostModelNode.html">CostModelNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1IRBuilderNode.html">IRBuilderNode</a> [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ADT.html">ADT</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1CountNode.html">CountNode</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IRDocsifier.html">IRDocsifier</ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ADTObj.html">ADTObj</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CropAndResizeAttrs.html">CropAndResizeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IRDocsifierFunctor.html">IRDocsifierFunctor</a> ( [...]
+</td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv3DTransposeAttrs.html">Conv3DTransposeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IntImm.html">IntImm</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternVisitor.html">PatternVisitor</a> (<a class="el" href="namespacetvm_1_1relay.html [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv3DWinogradAttrs.html">Conv3DWinogradAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IntImmNode.html">IntImmNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternWildcard.html">PatternWildcard</a> (<a class="el" href="namespacetvm_1_1rel [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AccessAnalyzer.html">AccessAnalyzer</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ConvGemmWeightTransformAttrs.html">ConvGemmWeightTransformAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1In [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AccessAnalyzerNode.html">AccessAnalyzerNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ConvWinogradWeightTransformAttrs.html">ConvWinogradWeightTransformAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtv [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AdaptivePool1DAttrs.html">AdaptivePool1DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CorrelationAttrs.html">CorrelationAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntSetNode.html">IntSetNode</a> (<a class="e [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AdaptivePool2DAttrs.html">AdaptivePool2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1CostModel.html">CostModel</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1IRBuilder.html">I [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AdaptivePool3DAttrs.html">AdaptivePool3DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CostModel.html">CostModel</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1IRBuilderFrame [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Add.html">Add</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1CostModelNode.html">CostModelNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1IRBuilderFrameNode.html">IRBuilderFrameNode</a> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AddNode.html">AddNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CostModelNode.html">CostModelNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1IRBuilderNode.html">IRBuilderNode</a> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ADT.html">ADT</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1CountNode.html">CountNode</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IRDocsifier.html">IRDocsifier</ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ADTObj.html">ADTObj</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CropAndResizeAttrs.html">CropAndResizeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IRDocsifierFunctor.html">IRDocsifierFunctor</a> ( [...]
<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AffineGridAttrs.html">AffineGridAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_d"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  d  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IRDocsifierNode.html">IRDocsifierNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1tir_1_1usmp_1_1PoolAllocationNode.html">PoolAllocationNode</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1S [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AffineType.html">AffineType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IRModule.html">IRModule</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1PoolInfo.html">PoolInfo</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="str [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AffineTypeNode.html">AffineTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1Database.html">Database</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1ir_1_1IRModuleFrame.html">IRModuleFrame</a> (<a class=" [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AllClassNonMaximumSuppressionAttrs.html">AllClassNonMaximumSuppressionAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1DatabaseNode.html">DatabaseNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Allocate.html">Allocate</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DataProducer.html">DataProducer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IRModuleNode.html">IRModuleNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)  & [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AllocateConst.html">AllocateConst</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DataProducerNode.html">DataProducerNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1is__specialized.html">is_specialized</a> (<a class="el" href="namesp [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AllocateConstFrame.html">AllocateConstFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1DataType.html">DataType</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1de [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AllocateConstFrameNode.html">AllocateConstFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DataTypeLegalizer.html">DataTypeLegalizer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="s [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AllocateConstNode.html">AllocateConstNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DataTypePattern.html">DataTypePattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1is__valid__iterator_3_01Optional_3_01T_01_4_00_01IterTy [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1usmp_1_1AllocatedPoolInfo.html">AllocatedPoolInfo</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DataTypePatternNode.html">DataTypePatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1IterAdapter.html">IterAdap [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1tir_1_1usmp_1_1AllocatedPoolInfoNode.html">AllocatedPoolInfoNode</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DebugAttrs.html">DebugAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Map_1_1iterator.html">Map::iterato [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AllocateFrame.html">AllocateFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DeclBuffer.html">DeclBuffer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1MapNode_1_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AllocateFrameNode.html">AllocateFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1DeclBufferFrame.html">DeclBufferFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)  [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AllocateNode.html">AllocateNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1DeclBufferFrameNode.html">DeclBufferFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1suppo [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1vm_1_1Allocator.html">Allocator</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DeclBufferNode.html">DeclBufferNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1IteratorNode.html">IteratorNode</a> (<a clas [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AllocStorageAttrs.html">AllocStorageAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DeformableConv2DAttrs.html">DeformableConv2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1auto__scheduler_1_1AttachMapNode_1_1IterKeyHas [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AllocTensorAttrs.html">AllocTensorAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DenseAttrs.html">DenseAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMapExpr.html">IterMapExpr</a> (<a class="el" href="namespa [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1AltPattern.html">AltPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1DenseMapNode.html">DenseMapNode</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMapExprNode.html">IterMapExprNode</a> (<a class="el" href="nam [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1AltPatternNode.html">AltPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DensePackAttrs.html">DensePackAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMapResult.html">IterMapResult</a> (<a class="el" href=" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1Analyzer.html">Analyzer</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Dependency.html">Dependency</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMapResultNode.html">IterMapResultNode</a> (<a class="el" href="namespacetvm_1_1ari [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1And.html">And</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DependencyNode.html">DependencyNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMark.html">IterMark</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)&# [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AndNode.html">AndNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1qnn_1_1DequantizeAttrs.html">DequantizeAttrs</a> (<a class="el" href="namespacetvm_1_1relay_1_1qnn.html">tvm::relay::qnn</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMarkNode.html">IterMarkNode</a> (<a class="el" href="n [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AnnotationStep.html">AnnotationStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1DeviceAPI.html">DeviceAPI</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterSplitExpr.html">IterSplitExpr</ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AnnotationStepNode.html">AnnotationStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DeviceCopyAttrs.html">DeviceCopyAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterSplitExprNode.ht [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Any.html">Any</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1DeviceWrapper.html">DeviceWrapper</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterSumExpr.html">IterSumExpr</a> (<a class="el [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AnyNode.html">AnyNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1profiling_1_1DeviceWrapperNode.html">DeviceWrapperNode</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterSumExprNode.html">IterSumE [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ArangeAttrs.html">ArangeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPattern.html">DFPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IterVar.html">IterVar</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm: [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ArgInfo.html">ArgInfo</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternCallback.html">DFPatternCallback</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1IterVarAttr.html">IterVarAttr</a> (<a class=" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ArgInfoNode.html">ArgInfoNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternCallbackNode.html">DFPatternCallbackNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1IterVarAttrNode.html">IterVar [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ArgReduceAttrs.html">ArgReduceAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternFunctor.html">DFPatternFunctor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IterVarNode.html">IterVarNode</a> (<a class="el" href="na [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ArgsortAttrs.html">ArgsortAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternFunctor_3_01R_07const_01DFPattern_01_6n_00_01Args_8_8_8_08_4.html">DFPatternFunctor< R(const DFPattern &n, Args...)></a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a c [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Array.html">Array</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternNode.html">DFPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1IterVarRelationNode.html">IterVarRelationNode</a> (<a class="el" href="namesp [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IRDocsifierNode.html">IRDocsifierNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1PoolInfoNode.html">PoolInfoNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1StageNode.html">StageNode</a> (<a class="el" href="nam [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1AffineType.html">AffineType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IRModule.html">IRModule</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1PoolInfoProperties.html">PoolInfoProperties</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1AffineTypeNode.html">AffineTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1Database.html">Database</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1ir_1_1IRModuleFrame.html">IRModuleFrame</a> (<a class=" [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AllClassNonMaximumSuppressionAttrs.html">AllClassNonMaximumSuppressionAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1DatabaseNode.html">DatabaseNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Allocate.html">Allocate</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DataProducer.html">DataProducer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IRModuleNode.html">IRModuleNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)  & [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AllocateConst.html">AllocateConst</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DataProducerNode.html">DataProducerNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1is__specialized.html">is_specialized</a> (<a class="el" href="namesp [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AllocateConstFrame.html">AllocateConstFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1DataType.html">DataType</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1de [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AllocateConstFrameNode.html">AllocateConstFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DataTypeLegalizer.html">DataTypeLegalizer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="s [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AllocateConstNode.html">AllocateConstNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DataTypePattern.html">DataTypePattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1is__valid__iterator_3_01Optional_3_01T_01_4_00_01IterTy [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1usmp_1_1AllocatedPoolInfo.html">AllocatedPoolInfo</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DataTypePatternNode.html">DataTypePatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1IterAdapter.html">IterAdap [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1tir_1_1usmp_1_1AllocatedPoolInfoNode.html">AllocatedPoolInfoNode</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DebugAttrs.html">DebugAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1MapNode_1_1iterator.html">MapNode: [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AllocateFrame.html">AllocateFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DeclBuffer.html">DeclBuffer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Map_1_1ite [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AllocateFrameNode.html">AllocateFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1DeclBufferFrame.html">DeclBufferFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)  [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AllocateNode.html">AllocateNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1DeclBufferFrameNode.html">DeclBufferFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1suppo [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1vm_1_1Allocator.html">Allocator</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DeclBufferNode.html">DeclBufferNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1IteratorNode.html">IteratorNode</a> (<a clas [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AllocStorageAttrs.html">AllocStorageAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DeformableConv2DAttrs.html">DeformableConv2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1auto__scheduler_1_1AttachMapNode_1_1IterKeyHas [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AllocTensorAttrs.html">AllocTensorAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DenseAttrs.html">DenseAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMapExpr.html">IterMapExpr</a> (<a class="el" href="namespa [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1AltPattern.html">AltPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1DenseMapNode.html">DenseMapNode</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMapExprNode.html">IterMapExprNode</a> (<a class="el" href="nam [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1AltPatternNode.html">AltPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DensePackAttrs.html">DensePackAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMapResult.html">IterMapResult</a> (<a class="el" href=" [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1Analyzer.html">Analyzer</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Dependency.html">Dependency</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMapResultNode.html">IterMapResultNode</a> (<a class="el" href="namespacetvm_1_1ari [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1And.html">And</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DependencyNode.html">DependencyNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMark.html">IterMark</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)&# [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AndNode.html">AndNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1qnn_1_1DequantizeAttrs.html">DequantizeAttrs</a> (<a class="el" href="namespacetvm_1_1relay_1_1qnn.html">tvm::relay::qnn</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterMarkNode.html">IterMarkNode</a> (<a class="el" href="n [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AnnotationStep.html">AnnotationStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1DeviceAPI.html">DeviceAPI</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterSplitExpr.html">IterSplitExpr</ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AnnotationStepNode.html">AnnotationStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DeviceCopyAttrs.html">DeviceCopyAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterSplitExprNode.ht [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Any.html">Any</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1DeviceWrapper.html">DeviceWrapper</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterSumExpr.html">IterSumExpr</a> (<a class="el [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AnyNode.html">AnyNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1profiling_1_1DeviceWrapperNode.html">DeviceWrapperNode</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IterSumExprNode.html">IterSumE [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ArangeAttrs.html">ArangeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPattern.html">DFPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IterVar.html">IterVar</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm: [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ArgInfo.html">ArgInfo</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternCallback.html">DFPatternCallback</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1IterVarAttr.html">IterVarAttr</a> (<a class=" [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ArgInfoNode.html">ArgInfoNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternCallbackNode.html">DFPatternCallbackNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1IterVarAttrNode.html">IterVar [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ArgReduceAttrs.html">ArgReduceAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternFunctor.html">DFPatternFunctor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IterVarNode.html">IterVarNode</a> (<a class="el" href="na [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ArgsortAttrs.html">ArgsortAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternFunctor_3_01R_07const_01DFPattern_01_6n_00_01Args_8_8_8_08_4.html">DFPatternFunctor< R(const DFPattern &n, Args...)></a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a c [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Array.html">Array</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternNode.html">DFPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1IterVarRelationNode.html">IterVarRelationNode</a> (<a class="el" href="namesp [...]
<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ArrayAccessor.html">ArrayAccessor</a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DFPatternVisitor.html">DFPatternVisitor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_l"></a><table border="0" cel [...]
-</td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ProfilerNode.html">ProfilerNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1StringImmNode.html">StringImmNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ArrayAccessor_3_01const_01char_01_5_00_01_1_1tvm_1_1runtime_1_1String_01_4.html">ArrayAccessor< const char *, ::tvm::runtime::String ></a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DFTAttrs.html">DFTAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>) & [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1SimpleObjAllocator_1_1ArrayHandler.html">SimpleObjAllocator::ArrayHandler</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1Diagnostic.html">Diagnostic</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1L2NormalizeAttrs.html">L2NormalizeAttrs</a> (<a [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1ArrayIndexPath.html">ArrayIndexPath</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticBuilder.html">DiagnosticBuilder</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1LambdaDoc.html">LambdaDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">t [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1ArrayIndexPathNode.html">ArrayIndexPathNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticContext.html">DiagnosticContext</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1LambdaDocNode.html">LambdaDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ArrayIterator.html">ArrayIterator</a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticContextNode.html">DiagnosticContextNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1LaunchThre [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ArrayNode.html">ArrayNode</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticNode.html">DiagnosticNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1LaunchThreadFrameNode.html">LaunchThreadFrameNode</a> (<a class="e [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AssertDoc.html">AssertDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticRenderer.html">DiagnosticRenderer</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1LayerNormAttrs.html">LayerNormAttrs</a> (<a class="e [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ProgramBuilderNode.html">ProgramBuilderNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Sub.html">Sub</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td></tr>
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ArrayAccessor_3_01const_01char_01_5_00_01_1_1tvm_1_1runtime_1_1String_01_4.html">ArrayAccessor< const char *, ::tvm::runtime::String ></a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DFTAttrs.html">DFTAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>) & [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1SimpleObjAllocator_1_1ArrayHandler.html">SimpleObjAllocator::ArrayHandler</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1Diagnostic.html">Diagnostic</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1L2NormalizeAttrs.html">L2NormalizeAttrs</a> (<a [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1ArrayIndexPath.html">ArrayIndexPath</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticBuilder.html">DiagnosticBuilder</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1LambdaDoc.html">LambdaDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">t [...]
</td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AssertDocNode.html">AssertDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticRendererNode.html">DiagnosticRendererNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Layout.html">Layout</a> (<a class="el" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AssertFrame.html">AssertFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DictAttrs.html">DictAttrs</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LayoutAxis.html">LayoutAxis</a> (<a cla [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AssertFrameNode.html">AssertFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DictAttrsNode.html">DictAttrsNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LayoutNode.html">Layout [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AssertStmt.html">AssertStmt</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DictDoc.html">DictDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1LayoutTransformAttrs.html">LayoutTransformAttrs</a> ( [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AssertStmtNode.html">AssertStmtNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DictDocNode.html">DictDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LE.html">LE</a> (<a class="el" href="nam [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AssignDoc.html">AssignDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DilateAttrs.html">DilateAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1LeakyReluAttrs.html">LeakyReluAttrs</a> [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AssignDocNode.html">AssignDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Dilation2DAttrs.html">Dilation2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1tir_1_1LENode.html">LENode</a> (< [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AttachMap.html">AttachMap</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Div.html">Div</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Let.html">Let</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::t [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AttachMapNode.html">AttachMapNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DivNode.html">DivNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Let.html">Let</a> (<a class="el" href="namespacetvm_1 [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AttrAccessDoc.html">AttrAccessDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1Doc.html">Doc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1 [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AttrAccessDocNode.html">AttrAccessDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DocNode.html">DocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrDocEntry.html">AttrDocEntry</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DocStringDoc.html">DocStringDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LetNode.html">LetNode</a> (<a cla [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrDocVisitor.html">AttrDocVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DocStringDocNode.html">DocStringDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1LetNode.html">LetNo [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1AttrError.html">AttrError</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DominatorPattern.html">DominatorPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1LetPattern.html">LetPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::rela [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrExistVisitor.html">AttrExistVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DominatorPatternNode.html">DominatorPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1LetPatternNode.html">LetPatternNode</a> ( [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttrFieldInfo.html">AttrFieldInfo</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DropoutAttrs.html">DropoutAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LetStmt.html">LetStmt</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)  [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttrFieldInfoNode.html">AttrFieldInfoNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1DurationNode.html">DurationNode</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LetStmtNode.html">LetStmtNode</a> (<a clas [...]
-</td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1TempExprNode.html">TempExprNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AttrFrame.html">AttrFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DynExpandDimsAttrs.html">DynExpandDimsAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1suppor [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1ArrayIndexPathNode.html">ArrayIndexPathNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticContext.html">DiagnosticContext</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1LambdaDocNode.html">LambdaDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ArrayIterator.html">ArrayIterator</a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticContextNode.html">DiagnosticContextNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1LaunchThre [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ArrayNode.html">ArrayNode</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticNode.html">DiagnosticNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1LaunchThreadFrameNode.html">LaunchThreadFrameNode</a> (<a class="e [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AssertDoc.html">AssertDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticRenderer.html">DiagnosticRenderer</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1LayerNormAttrs.html">LayerNormAttrs</a> (<a class="e [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AssertDocNode.html">AssertDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DiagnosticRendererNode.html">DiagnosticRendererNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Layout.html">Layout</a> (<a class="el" [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AssertFrame.html">AssertFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DictAttrs.html">DictAttrs</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LayoutAxis.html">LayoutAxis</a> (<a cla [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AssertFrameNode.html">AssertFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1DictAttrsNode.html">DictAttrsNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LayoutNode.html">Layout [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AssertStmt.html">AssertStmt</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DictDoc.html">DictDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1LayoutTransformAttrs.html">LayoutTransformAttrs</a> ( [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AssertStmtNode.html">AssertStmtNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DictDocNode.html">DictDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LE.html">LE</a> (<a class="el" href="nam [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AssignDoc.html">AssignDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DilateAttrs.html">DilateAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1LeakyReluAttrs.html">LeakyReluAttrs</a> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AssignDocNode.html">AssignDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Dilation2DAttrs.html">Dilation2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1tir_1_1LENode.html">LENode</a> (< [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AttachMap.html">AttachMap</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Div.html">Div</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Let.html">Let</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::t [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1AttachMapNode.html">AttachMapNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1DivNode.html">DivNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Let.html">Let</a> (<a class="el" href="namespacetvm_1 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AttrAccessDoc.html">AttrAccessDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1Doc.html">Doc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1AttrAccessDocNode.html">AttrAccessDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DocNode.html">DocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrDocEntry.html">AttrDocEntry</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DocStringDoc.html">DocStringDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LetNode.html">LetNode</a> (<a cla [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrDocVisitor.html">AttrDocVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1DocStringDocNode.html">DocStringDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1LetNode.html">LetNo [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1AttrError.html">AttrError</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DominatorPattern.html">DominatorPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1LetPattern.html">LetPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::rela [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrExistVisitor.html">AttrExistVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1DominatorPatternNode.html">DominatorPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1LetPatternNode.html">LetPatternNode</a> ( [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1TensorAffineType.html">TensorAffineType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttrFieldInfo.html">AttrFieldInfo</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DropoutAttrs.html">DropoutAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LetStmt.html">LetStmt</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)  [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttrFieldInfoNode.html">AttrFieldInfoNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1DurationNode.html">DurationNode</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LetStmtNode.html">LetStmtNode</a> (<a clas [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AttrFrame.html">AttrFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1DynExpandDimsAttrs.html">DynExpandDimsAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1suppor [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1te_1_1TensorComputeOpNode.html">TensorComputeOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td></tr>
<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1AttrFrameNode.html">AttrFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_e"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  e  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ListDoc.html">ListDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1qnn_1_1QuantizeAttrs.html">QuantizeAttrs</a> (<a class="el" href="namespacetvm_1_1relay_1_1qnn.html">tvm::relay::qnn</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1TensorAffineType.html">TensorA [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttributeAccessPath.html">AttributeAccessPath</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ListDocNode.html">ListDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_r"></a><table border="0" cellspacing="0" cellpadding="0"> [...]
-</td><td valign="top"><a class="el" href="classtvm_1_1TensorAffineTypeNode.html">TensorAffineTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttributeAccessPathNode.html">AttributeAccessPathNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1EinsumAttrs.html">EinsumAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1LiteralDoc.html">LiteralDoc</a> (<a class="el" href="namespa [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1detail_1_1AttrInitEntry.html">AttrInitEntry</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1topi_1_1EinsumEquation.html">EinsumEquation</a> (<a class="el" href="namespacetvm_1_1topi.html">tvm::topi</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1LiteralDocNode.html">LiteralDocNode</a> (<a clas [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrInitVisitor.html">AttrInitVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1ElseFrame.html">ElseFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Loa [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrNonDefaultVisitor.html">AttrNonDefaultVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1ElseFrameNode.html">ElseFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="cl [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1detail_1_1AttrNopEntry.html">AttrNopEntry</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1EnvFunc.html">EnvFunc</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1LocalBuilder.html">LocalBuilder</a> (<a class="el" href="namespacetvm_1_1auto__scheduler. [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrNormalVisitor.html">AttrNormalVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1EnvFuncNode.html">EnvFuncNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1LocalBuilderNode.html">LocalBuilderNode</a> (<a class="el" href="namespa [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1AttrPattern.html">AttrPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1EQ.html">EQ</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1LocalRunner.html">LocalRunner</a> (<a class="el" href="namespacetvm_1_1auto__scheduler [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1AttrPatternNode.html">AttrPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1EQNode.html">EQNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1LocalRunnerNode.html">LocalRunnerNode</a> (<a class="el" href="namespa [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttrRegistry.html">AttrRegistry</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ErrorBuilder.html">ErrorBuilder</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LoopRV.html">LoopRV</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)  [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttrRegistryMap.html">AttrRegistryMap</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ErrorReporter.html">ErrorReporter</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LoopRVNode.html">LoopRVNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttrRegistryMapContainerMap.html">AttrRegistryMapContainerMap</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Evaluate.html">Evaluate</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1LRNAttrs.html">LRNAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1Attrs.html">Attrs</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1EvaluateNode.html">EvaluateNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LT.html">LT</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttrsNode.html">AttrsNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1vm_1_1Executable.html">Executable</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LTNode.html">LTNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a> [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrsSEqualVisitor.html">AttrsSEqualVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Executor.html">Executor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_m"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div c [...]
-</td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1RecClosureObj.html">RecClosureObj</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1TensorNode.html">TensorNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrsSHashVisitor.html">AttrsSHashVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExecutorNode.html">ExecutorNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1RecordReader.html">RecordReader</a> (<a class [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AttrStmt.html">AttrStmt</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExecutorRegEntry.html">ExecutorRegEntry</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Map.html">Map</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AttrStmtNode.html">AttrStmtNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ExpandDimsAttrs.html">ExpandDimsAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1MapNode.html">MapNode</a> (<a class="el" href="namespacetvm_1_1ru [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1detail_1_1AttrTriggerNonDefaultEntry.html">AttrTriggerNonDefaultEntry</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1tir_1_1ExprDeepEqual.html">ExprDeepEqual</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MapValuePath.html">MapValuePath</a> (<a class="e [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1AttrVisitor.html">AttrVisitor</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ExprDoc.html">ExprDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MapValuePathNode.html">MapValuePathNode</a> (<a class="el" href="namespacetvm.ht [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AutoSchedulerLayoutTransformAttrs.html">AutoSchedulerLayoutTransformAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ExprDocNode.html">ExprDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1re [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AvgPool1DAttrs.html">AvgPool1DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ExprFunctor.html">ExprFunctor</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MatchBufferRegion.html">MatchBufferRegion</a> (<a class="el" href="namesp [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AvgPool2DAttrs.html">AvgPool2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprFunctor.html">ExprFunctor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MatchBufferRegionNode.html">MatchBufferRegionNode</a> (<a class="el [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AvgPool3DAttrs.html">AvgPool3DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprFunctor_3_01R_07const_01Expr_01_6n_00_01Args_8_8_8_08_4.html">ExprFunctor< R(const Expr &n, Args...)></a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href=" [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ListDoc.html">ListDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1te_1_1TensorDom.html">TensorDom</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td></tr>
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttributeAccessPath.html">AttributeAccessPath</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ListDocNode.html">ListDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Ramp.html">Ramp</a> (<a class="el" href="namespac [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttributeAccessPathNode.html">AttributeAccessPathNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1EinsumAttrs.html">EinsumAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1LiteralDoc.html">LiteralDoc</a> (<a class="el" href="namespa [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1detail_1_1AttrInitEntry.html">AttrInitEntry</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1topi_1_1EinsumEquation.html">EinsumEquation</a> (<a class="el" href="namespacetvm_1_1topi.html">tvm::topi</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1LiteralDocNode.html">LiteralDocNode</a> (<a clas [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrInitVisitor.html">AttrInitVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1ElseFrame.html">ElseFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__sche [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrNonDefaultVisitor.html">AttrNonDefaultVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1ElseFrameNode.html">ElseFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="cl [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1detail_1_1AttrNopEntry.html">AttrNopEntry</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1EnvFunc.html">EnvFunc</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1LocalRunner.html">LocalRunner</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.ht [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrNormalVisitor.html">AttrNormalVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1EnvFuncNode.html">EnvFuncNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1LocalRunnerNode.html">LocalRunnerNode</a> (<a class="el" href="namespace [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1AttrPattern.html">AttrPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1EQ.html">EQ</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LoopRV.html">LoopRV</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)  &# [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1AttrPatternNode.html">AttrPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1EQNode.html">EQNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LoopRVNode.html">LoopRVNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttrRegistry.html">AttrRegistry</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ErrorBuilder.html">ErrorBuilder</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1LRNAttrs.html">LRNAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttrRegistryMap.html">AttrRegistryMap</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ErrorReporter.html">ErrorReporter</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LT.html">LT</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttrRegistryMapContainerMap.html">AttrRegistryMapContainerMap</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Evaluate.html">Evaluate</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1LTNode.html">LTNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1Attrs.html">Attrs</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1EvaluateNode.html">EvaluateNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_m"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  m  </div></td></tr></table>
+</td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1RecClosureObj.html">RecClosureObj</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1TensorTypeNode.html">TensorTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttrsNode.html">AttrsNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1vm_1_1Executable.html">Executable</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1RecordReader.html">RecordReader</a> (<a class="el" href="namespacetvm_1 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrsSEqualVisitor.html">AttrsSEqualVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Executor.html">Executor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Map.html">Map</a> (<a class="el" href="namespacetvm_1_1runt [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1detail_1_1AttrsSHashVisitor.html">AttrsSHashVisitor</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExecutorNode.html">ExecutorNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1MapNode.html">MapNode</a> (<a class="el" href="namesp [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AttrStmt.html">AttrStmt</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExecutorRegEntry.html">ExecutorRegEntry</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MapValuePath.html">MapValuePath</a> (<a class="el" href="namespacetvm.html">tvm</a> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1AttrStmtNode.html">AttrStmtNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ExpandDimsAttrs.html">ExpandDimsAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MapValuePathNode.html">MapValuePathNode</a> (<a class="el" href="namespacetv [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1detail_1_1AttrTriggerNonDefaultEntry.html">AttrTriggerNonDefaultEntry</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1tir_1_1ExprDeepEqual.html">ExprDeepEqual</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Match.html">Match</a> (<a class="el" hr [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1AttrVisitor.html">AttrVisitor</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ExprDoc.html">ExprDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MatchBufferRegion.html">MatchBufferRegion</a> (<a class="el" href="namesp [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AutoSchedulerLayoutTransformAttrs.html">AutoSchedulerLayoutTransformAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ExprDocNode.html">ExprDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1ti [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AvgPool1DAttrs.html">AvgPool1DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ExprFunctor.html">ExprFunctor</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1MatchNode.html">MatchNode</a> (<a class="el" href="namespacetvm_1_1rela [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AvgPool2DAttrs.html">AvgPool2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprFunctor.html">ExprFunctor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MatmulAttrs.html">MatmulAttrs</a> (<a class="el" href="namespace [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1AvgPool3DAttrs.html">AvgPool3DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprFunctor_3_01R_07const_01Expr_01_6n_00_01Args_8_8_8_08_4.html">ExprFunctor< R(const Expr &n, Args...)></a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href=" [...]
<tr><td rowspan="2" valign="bottom"><a name="letter_b"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  b  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ExprFunctor_3_01R_07const_01PrimExpr_01_6n_00_01Args_8_8_8_08_4.html">ExprFunctor< R(const PrimExpr &n, Args...)></a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MatmulAttrs.html">MatmulAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="str [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ExprMutator.html">ExprMutator</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MatrixSetDiagAttrs.html">MatrixSetDiagAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1ReflectionVTable.html">ReflectionVTable</a> (<a class="el" href="namespa [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseAttrsNode.html">BaseAttrsNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprMutator.html">ExprMutator</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Max.html">Max</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)    [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1te_1_1BaseComputeOpNode.html">BaseComputeOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprPattern.html">ExprPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MaxNode.html">MaxNode</a> (<a class="el" href="namespacetvm_1_1tir.html [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseExpr.html">BaseExpr</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprPatternNode.html">ExprPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MaxPool1DAttrs.html">MaxPool1DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm:: [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseExprNode.html">BaseExprNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprRewriter.html">ExprRewriter</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MaxPool2DAttrs.html">MaxPool2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseFunc.html">BaseFunc</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ExprStmtDoc.html">ExprStmtDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MaxPool3DAttrs.html">MaxPool3DAttrs</a> (<a class="el" href="namespa [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseFuncNode.html">BaseFuncNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ExprStmtDocNode.html">ExprStmtDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MeasureCallback.html">MeasureCallback</a> (< [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseTensorType.html">BaseTensorType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprVisitor.html">ExprVisitor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureCallback.html">MeasureCallback</a> (<a class="el" href="namespacetvm_1_1au [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseTensorTypeNode.html">BaseTensorTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ExprVisitor.html">ExprVisitor</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MeasureCallbackNode.html">MeasureCallbackNode</a> (<a class="el" href="namespace [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseValueEqual.html">BaseValueEqual</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ExternOp.html">ExternOp</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureCallbackNode.html">MeasureCallbackNode</a> (<a class="el" href="namespacetvm_1_1auto__sch [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1BaseValueHash.html">BaseValueHash</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ExternOpNode.html">ExternOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MeasureCandidate.html">MeasureCandidate</a> (<a class="el" href="namespacetvm_1_1meta__sche [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BatchMatmulAttrs.html">BatchMatmulAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ExtractedTask.html">ExtractedTask</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MeasureCandidateNode.h [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BatchNormAttrs.html">BatchNormAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ExtractedTaskNode.html">ExtractedTaskNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureInput.html [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BatchToSpaceNDAttrs.html">BatchToSpaceNDAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncObj_1_1Extractor.html">PackedFuncObj::Extractor</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureInp [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ExprFunctor_3_01R_07const_01PrimExpr_01_6n_00_01Args_8_8_8_08_4.html">ExprFunctor< R(const PrimExpr &n, Args...)></a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Max.html">Max</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1RefRe [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ExprMutator.html">ExprMutator</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MaxNode.html">MaxNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1RefReadNode.html">RefReadNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::rela [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseAttrsNode.html">BaseAttrsNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprMutator.html">ExprMutator</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MaxPool1DAttrs.html">MaxPool1DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1te_1_1BaseComputeOpNode.html">BaseComputeOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprPattern.html">ExprPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MaxPool2DAttrs.html">MaxPool2DAttrs</a> (<a class="el" href="namespa [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseExpr.html">BaseExpr</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprPatternNode.html">ExprPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MaxPool3DAttrs.html">MaxPool3DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm:: [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseExprNode.html">BaseExprNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprRewriter.html">ExprRewriter</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MeasureCallback.html">MeasureCallback</a> (<a class="el" href="namespacetvm_1_1meta_ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseFunc.html">BaseFunc</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ExprStmtDoc.html">ExprStmtDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureCallback.html">MeasureCallback</a> (<a class="el" hr [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseFuncNode.html">BaseFuncNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ExprStmtDocNode.html">ExprStmtDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MeasureCallbackNode.html">MeasureCallbackNod [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseTensorType.html">BaseTensorType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ExprVisitor.html">ExprVisitor</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureCallbackNode.html">MeasureCallbackNode</a> (<a class="el" href="namespacetvm_1_1 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseTensorTypeNode.html">BaseTensorTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ExprVisitor.html">ExprVisitor</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MeasureCandidate.html">MeasureCandidate</a> (<a class="el" href="namespace [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseValueEqual.html">BaseValueEqual</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ExternOp.html">ExternOp</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MeasureCandidateNode.html">MeasureCandidateNode</a> (<a class="el" href="namespacetvm_1_1meta__sc [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1BaseValueHash.html">BaseValueHash</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ExternOpNode.html">ExternOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureInput.html">MeasureInput</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.h [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BatchMatmulAttrs.html">BatchMatmulAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ExtractedTask.html">ExtractedTask</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureInputNode.html [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BatchNormAttrs.html">BatchNormAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1ExtractedTaskNode.html">ExtractedTaskNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureResult.htm [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BatchToSpaceNDAttrs.html">BatchToSpaceNDAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncObj_1_1Extractor.html">PackedFuncObj::Extractor</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureRes [...]
<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BiasAddAttrs.html">BiasAddAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_f"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  f  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureResult.html">MeasureResult</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ReorderStep.html">ReorderStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1 [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BijectiveLayout.html">BijectiveLayout</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1MeasureResultNode.html">MeasureResultNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ReorderStepNode.htm [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BijectiveLayoutNode.html">BijectiveLayoutNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1FeatureExtractor.html">FeatureExtractor</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1tir_1_1MemCpyDetails.html">MemCpyD [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BinaryConv2DAttrs.html">BinaryConv2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1FeatureExtractorNode.html">FeatureExtractorNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MemoryInfo.html">MemoryI [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BinaryDenseAttrs.html">BinaryDenseAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1FeatureSet.html">FeatureSet</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MemoryInfoNode.html">MemoryInfoNode</a> (<a class="el" href="namespacetv [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BinaryOpNode.html">BinaryOpNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1FIFOBufferAttrs.html">FIFOBufferAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1vm_1_1MemoryManager.html">MemoryManager</a> (<a class="el" href=" [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BitPackAttrs.html">BitPackAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1FixedPointMultiplyAttrs.html">FixedPointMultiplyAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structMemoryManagerInterface.html">MemoryManagerInterface</a> [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Block.html">Block</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1FixedPointMultiplyPerAxisAttrs.html">FixedPointMultiplyPerAxisAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MeshgridAttrs.html">MeshgridAttrs</a> (<a class="e [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1BlockFrame.html">BlockFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1SeqStmt_1_1Flattener.html">SeqStmt::Flattener</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1 [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1BlockFrameNode.html">BlockFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1FloatImm.html">FloatImm</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1MetadataArray.html" [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1tir_1_1BlockInfo.html">BlockInfo</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1FloatImmNode.html">FloatImmNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1MetadataArrayNode.html">MetadataArrayNode</a> (<a class="el" href="namespacetvm_1_1runtime [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1BlockInitFrame.html">BlockInitFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1FloorDiv.html">FloorDiv</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1 [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1BlockInitFrameNode.html">BlockInitFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1FloorDivNode.html">FloorDivNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockNode.html">BlockNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1FloorMod.html">FloorMod</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1MetadataNode.html">MetadataNode</a> (<a class="el" href="namespacetvm_1_1runtime [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockRealize.html">BlockRealize</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1FloorModNode.html">FloorModNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MetaScheduleLayoutTransformAttrs.html">MetaScheduleLayoutTransformAttrs</a> (<a [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockRealizeNode.html">BlockRealizeNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FollowFusedSplitStep.html">FollowFusedSplitStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1Metric [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockRV.html">BlockRV</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FollowFusedSplitStepNode.html">FollowFusedSplitStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1MetricCollectorN [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockRVNode.html">BlockRVNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FollowSplitStep.html">FollowSplitStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Min.html">Min</a> (<a class="el" href="name [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockScope.html">BlockScope</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FollowSplitStepNode.html">FollowSplitStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MinNode.html">MinNode</a> (<a class=" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockScopeNode.html">BlockScopeNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1For.html">For</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MirrorPadAttrs.html">MirrorPadAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm: [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1Bool.html">Bool</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ForDoc.html">ForDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MissingArrayElementPath.html">MissingArrayElementPath</a> (<a class="el" href="namespacetvm.html [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Broadcast.html">Broadcast</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ForDocNode.html">ForDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MissingArrayElementPathNode.html">MissingArrayElementPathNo [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1qnn_1_1BroadcastAttrs.html">BroadcastAttrs</a> (<a class="el" href="namespacetvm_1_1relay_1_1qnn.html">tvm::relay::qnn</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1ForFrame.html">ForFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BroadcastNode.html">BroadcastNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1ForFrameNode.html">ForFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MissingMapEntryPa [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Buffer.html">Buffer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ForNode.html">ForNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1MixedModeMutator.html">MixedModeMutator</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::rela [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1vm_1_1Buffer.html">Buffer</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1Frame.html">Frame</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1MixedModeVisitor.html">MixedModeVisit [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1usmp_1_1BufferInfo.html">BufferInfo</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1micro__rpc_1_1FrameBuffer.html">FrameBuffer</a> (<a class="el" href="namespacetvm_1_1runtime_1_1micro__rpc.html">tvm::runtime::micro_rpc</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Mod.html">Mod</ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1usmp_1_1BufferInfoAnalysis.html">BufferInfoAnalysis</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1FrameNode.html">FrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ModNode.html"> [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1tir_1_1usmp_1_1BufferInfoAnalysisNode.html">BufferInfoAnalysisNode</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1micro__rpc_1_1Framer.html">Framer</a> (<a class="el" href="namespacetvm_1_1runtime_1_1micro__rpc.html">tvm::runtime::micro_rpc</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1 [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1tir_1_1usmp_1_1BufferInfoNode.html">BufferInfoNode</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ShapeTupleObj_1_1FromStd.html">ShapeTupleObj::FromStd</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ModularSetAnalyzer. [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferLoad.html">BufferLoad</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1StringObj_1_1FromStd.html">StringObj::FromStd</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ModularSetNode.html">ModularSetNode</a> (<a class="el" hre [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferLoadNode.html">BufferLoadNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Function.html">Function</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Module.html">Module</a> (<a class="el" href="namespacetvm_1_1runtime.html">t [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferNode.html">BufferNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1FunctionDoc.html">FunctionDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ModuleNode.html">ModuleNode</a> (<a class="e [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferRealize.html">BufferRealize</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1FunctionDocNode.html">FunctionDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Mul.html">Mul</a> (<a class="el" h [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferRealizeNode.html">BufferRealizeNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1FunctionNode.html">FunctionNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MulNode.html">MulNode</a> (<a class="el" href="namespacetvm_1_1tir [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferRegion.html">BufferRegion</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1FunctionPattern.html">FunctionPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MultiBoxPriorAttrs.html">MultiBoxPriorAttrs</a> (<a class="el" href [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferRegionNode.html">BufferRegionNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1FunctionPatternNode.html">FunctionPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MultiBoxTransformLocAttrs.html">MultiBoxTransformLo [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferStore.html">BufferStore</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1FuncType.html">FuncType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MultinomialAttrs.html">MultinomialAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>) [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferStoreNode.html">BufferStoreNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1FuncTypeNode.html">FuncTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1Mutator.html">Mutator</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html"> [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1Builder.html">Builder</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1Fuse.html">Fuse</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MutatorNode.html">MutatorNode</a> (<a class="el" href="namespacetvm_ [...]
-</td><td valign="top"><a class="el" href="classtvm_1_1TypeCall.html">TypeCall</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderInput.html">BuilderInput</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1FuseNode.html">FuseNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_n"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div [...]
-</td><td valign="top"><a class="el" href="classtvm_1_1TypeCallNode.html">TypeCallNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderInputNode.html">BuilderInputNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FuseStep.html">FuseStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ScanOp.html">Sca [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderNode.html">BuilderNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FuseStepNode.html">FuseStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_ [...]
+</td><td valign="top"><a class="el" href="structtvm_1_1tir_1_1MemCpyDetails.html">MemCpyDetails</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1RepeatAttrs.html">RepeatAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1TupleDoc.html">TupleDoc</a> (<a class="el" href="namespacet [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BijectiveLayout.html">BijectiveLayout</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MemoryInfo.html">MemoryInfo</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1Report.html">Report</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling. [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BijectiveLayoutNode.html">BijectiveLayoutNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1FeatureExtractor.html">FeatureExtractor</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MemoryInfoNode.html">MemoryInfoNode [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BinaryConv2DAttrs.html">BinaryConv2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1FeatureExtractorNode.html">FeatureExtractorNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1vm_1_1MemoryM [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BinaryDenseAttrs.html">BinaryDenseAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1FeatureSet.html">FeatureSet</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structMemoryManagerInterface.html">MemoryManagerInterface</a>   </td> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BinaryOpNode.html">BinaryOpNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1FIFOBufferAttrs.html">FIFOBufferAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MeshgridAttrs.html">MeshgridAttrs</a> (<a class="el" href="namespa [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1BitPackAttrs.html">BitPackAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1FixedPointMultiplyAttrs.html">FixedPointMultiplyAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1Metadata.html">Metadata</a> (< [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Block.html">Block</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1FixedPointMultiplyPerAxisAttrs.html">FixedPointMultiplyPerAxisAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1MetadataArray.html">MetadataArray</a> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1BlockFrame.html">BlockFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1SeqStmt_1_1Flattener.html">SeqStmt::Flattener</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1BlockFrameNode.html">BlockFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1FloatImm.html">FloatImm</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1MetadataBase.html"> [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1tir_1_1BlockInfo.html">BlockInfo</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1FloatImmNode.html">FloatImmNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1MetadataBaseNode.html">MetadataBaseNode</a> (<a class="el" href="namespacetvm_1_1runtime_1 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1BlockInitFrame.html">BlockInitFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1FloorDiv.html">FloorDiv</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1BlockInitFrameNode.html">BlockInitFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1FloorDivNode.html">FloorDivNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockNode.html">BlockNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1FloorMod.html">FloorMod</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1MetricCollector.html">MetricCollector</a> (<a class="el" href="namespacetvm_1_1 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockRealize.html">BlockRealize</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1FloorModNode.html">FloorModNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1profiling_1_1MetricCollectorNode.html">MetricCollectorNode</a> (<a class="el" [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockRealizeNode.html">BlockRealizeNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FollowFusedSplitStep.html">FollowFusedSplitStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Min.html">Min</a> (<a c [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockRV.html">BlockRV</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FollowFusedSplitStepNode.html">FollowFusedSplitStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MinNode.html">MinNode</a> (<a cla [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockRVNode.html">BlockRVNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FollowSplitStep.html">FollowSplitStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MirrorPadAttrs.html">MirrorPadAttrs</a> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockScope.html">BlockScope</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FollowSplitStepNode.html">FollowSplitStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MissingArrayElementPath.html">MissingArrayEl [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BlockScopeNode.html">BlockScopeNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1For.html">For</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MissingArrayElementPathNode.html">MissingArrayElementPathNode</a> (<a class="el" href="namespacetvm.htm [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1Bool.html">Bool</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ForDoc.html">ForDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MissingMapEntryPath.html">MissingMapEntryPath</a> (<a class="el" href="namespacetvm.html">tvm</a [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Broadcast.html">Broadcast</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ForDocNode.html">ForDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1MissingMapEntryPathNode.html">MissingMapEntryPathNode</a> ( [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1qnn_1_1BroadcastAttrs.html">BroadcastAttrs</a> (<a class="el" href="namespacetvm_1_1relay_1_1qnn.html">tvm::relay::qnn</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1ForFrame.html">ForFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BroadcastNode.html">BroadcastNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1ForFrameNode.html">ForFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1MixedMod [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Buffer.html">Buffer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ForNode.html">ForNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Mod.html">Mod</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1vm_1_1Buffer.html">Buffer</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1Frame.html">Frame</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ModNode.html">ModNode</a> (<a class="el [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1usmp_1_1BufferInfo.html">BufferInfo</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1micro__rpc_1_1FrameBuffer.html">FrameBuffer</a> (<a class="el" href="namespacetvm_1_1runtime_1_1micro__rpc.html">tvm::runtime::micro_rpc</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ModularSet.ht [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1usmp_1_1BufferInfoAnalysis.html">BufferInfoAnalysis</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1FrameNode.html">FrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ModularSetAn [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1tir_1_1usmp_1_1BufferInfoAnalysisNode.html">BufferInfoAnalysisNode</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1micro__rpc_1_1Framer.html">Framer</a> (<a class="el" href="namespacetvm_1_1runtime_1_1micro__rpc.html">tvm::runtime::micro_rpc</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1 [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1tir_1_1usmp_1_1BufferInfoNode.html">BufferInfoNode</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp.html">tvm::tir::usmp</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ShapeTupleObj_1_1FromStd.html">ShapeTupleObj::FromStd</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Module.html">Modu [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferLoad.html">BufferLoad</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1StringObj_1_1FromStd.html">StringObj::FromStd</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ModuleNode.html">ModuleNode</a> (<a class="el" href="nam [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferLoadNode.html">BufferLoadNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Function.html">Function</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Mul.html">Mul</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)&# [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferNode.html">BufferNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1FunctionDoc.html">FunctionDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1MulNode.html">MulNode</a> (<a class="el" href="n [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferRealize.html">BufferRealize</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1FunctionDocNode.html">FunctionDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MultiBoxPriorAttrs.html">MultiB [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferRealizeNode.html">BufferRealizeNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1FunctionNode.html">FunctionNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MultiBoxTransformLocAttrs.html">MultiBoxTransformLocAttrs</a> ( [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferRegion.html">BufferRegion</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1FunctionPattern.html">FunctionPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1MultinomialAttrs.html">MultinomialAttrs</a> (<a class="el" href="na [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferRegionNode.html">BufferRegionNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1FunctionPatternNode.html">FunctionPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1Mutator.html">Mutator</a> (<a class="el" hr [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferStore.html">BufferStore</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1FuncType.html">FuncType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1MutatorNode.html">MutatorNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::met [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1TypeConstraint.html">TypeConstraint</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1BufferStoreNode.html">BufferStoreNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1FuncTypeNode.html">FuncTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_n"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  n  [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1TypeConstraintNode.html">TypeConstraintNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
+<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1Builder.html">Builder</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1Fuse.html">Fuse</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ScanOp.html">ScanOp</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderInput.html">BuilderInput</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1FuseNode.html">FuseNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1details_1_1Namer.html">Namer</a> (<a class= [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderInputNode.html">BuilderInputNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FuseStep.html">FuseStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1NameSupply.html">NameS [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderNode.html">BuilderNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1FuseStepNode.html">FuseStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1NameSupplyNode.html">Nam [...]
<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderResult.html">BuilderResult</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_g"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  g  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="classtvm_1_1NameSupply.html">NameSupply</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ScanOpNode.html">ScanOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1TypeData.html">TypeData</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderResultNode.html">BuilderResultNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1NameSupplyNode.html">NameSupplyNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ScatterElementsAttrs.html">ScatterElementsAttrs</a> [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1BuildResult.html">BuildResult</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GatherAttrs.html">GatherAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1NDArray.html">NDArray</a> (<a class="el" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1BuildResultNode.html">BuildResultNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GatherNDAttrs.html">GatherNDAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1NDArrayContainerTrait.html">NDArrayCon [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1NDArray.html">NDArray</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ScatterNDAttrs.html">ScatterNDAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1TypedPackedFunc.html">TypedPackedFunc</a> (<a class="el" href="n [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1BuilderResultNode.html">BuilderResultNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1NDArrayContainerTrait.html">NDArrayContainerTrait</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Schedule.html">Schedule</a> (<a class="e [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1BuildResult.html">BuildResult</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GatherAttrs.html">GatherAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1NdarraySizeAttrs.html">NdarraySizeAttrs</ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1BuildResultNode.html">BuildResultNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GatherNDAttrs.html">GatherNDAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1NE.html">NE</a> (<a class="el" h [...]
<tr><td rowspan="2" valign="bottom"><a name="letter_c"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  c  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1GE.html">GE</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1NdarraySizeAttrs.html">NdarraySizeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1Schedule.html">Schedule</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)& [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1GenericFunc.html">GenericFunc</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1NE.html">NE</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ScheduleContext.html">ScheduleContext</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td>< [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CacheReadStep.html">CacheReadStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GenericFuncNode.html">GenericFuncNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1NENode.html">NENode</a> (<a class="el" href="namespacetvm_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CacheReadStepNode.html">CacheReadStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1GENode.html">GENode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1NLLLossAttrs.html">NLLLossAttrs</a> (<a class= [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CacheWriteStep.html">CacheWriteStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GetValidCountsAttrs.html">GetValidCountsAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1NodeFunctor.html">NodeFunctor [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CacheWriteStepNode.html">CacheWriteStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GlobalPool2DAttrs.html">GlobalPool2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1NodeFunctor_3_01R_07const [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Call.html">Call</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalTypeVar.html">GlobalTypeVar</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1NonMaximumSuppressionAttrs.html">NonMaximumSuppressionAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Call.html">Call</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalTypeVarNode.html">GlobalTypeVarNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1NormalAttrs.html">NormalAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>) [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1CallDoc.html">CallDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classGlobalVar.html">GlobalVar</a>   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Not.html">Not</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" h [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1CallDocNode.html">CallDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalVar.html">GlobalVar</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1NotNode.html">NotNode</a> (<a class="el" href="namespacetvm_1_1tir.ht [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1profiling_1_1CallFrame.html">CallFrame</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalVarNode.html">GlobalVarNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1NullOptType.html">NullOptType</a> (<a class="el" h [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CallLoweredAttrs.html">CallLoweredAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalVarSupply.html">GlobalVarSupply</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_o"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah"> & [...]
-</td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1SearchCallback.html">SearchCallback</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1TypeName_3_01int64__t_01_4.html">TypeName< int64_t ></a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CallNode.html">CallNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalVarSupplyNode.html">GlobalVarSupplyNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1SearchCallbackNode.html">SearchCallbackNode</a> (<a class="el" href="namespacetvm_1 [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1CallNode.html">CallNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1usmp_1_1algo_1_1GreedyBase.html">GreedyBase</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp_1_1algo.html">tvm::tir::usmp::algo</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ObjAllocatorBase.html">ObjAllocatorBase< [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1CallPattern.html">CallPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GridSampleAttrs.html">GridSampleAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Object.html">Object</a> (<a class="el" href="namespacetvm_1_1 [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1CallPatternNode.html">CallPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GroupNormAttrs.html">GroupNormAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectEqual.html">ObjectEqual</a> (<a class="el" href= [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1CanonicalSimplifier.html">CanonicalSimplifier</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1GT.html">GT</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectHash.html">ObjectHash</a> (<a class="el" href="namespacetvm_1_1runtime. [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Cast.html">Cast</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1GTNode.html">GTNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1ObjectPath.html">ObjectPath</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="to [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1GE.html">GE</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1NENode.html">NENode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ScheduleNode.html">ScheduleNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)    [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1GenericFunc.html">GenericFunc</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1NLLLossAttrs.html">NLLLossAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1ScheduleNode.html">ScheduleNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>) [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CacheReadStep.html">CacheReadStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GenericFuncNode.html">GenericFuncNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1NodeFunctor.html">NodeFunctor</a> (<a class="el" href="namespacet [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CacheReadStepNode.html">CacheReadStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1GENode.html">GENode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1NodeFunctor_3_01R_07const_01ObjectRef_01_6n_00_01Args_8_ [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CacheWriteStep.html">CacheWriteStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GetValidCountsAttrs.html">GetValidCountsAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1NonMaximumSuppressi [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1CacheWriteStepNode.html">CacheWriteStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GlobalPool2DAttrs.html">GlobalPool2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1NormalAttrs.htm [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Call.html">Call</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalTypeVar.html">GlobalTypeVar</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Not.html">Not</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="to [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Call.html">Call</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalTypeVarNode.html">GlobalTypeVarNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1NotNode.html">NotNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1CallDoc.html">CallDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classGlobalVar.html">GlobalVar</a>   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1NullOptType.html">NullOptType</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1CallDocNode.html">CallDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalVar.html">GlobalVar</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_o"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class=" [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1SearchCallback.html">SearchCallback</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1TypeName_3_01void_01_5_01_4.html">TypeName< void * ></a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td></tr>
+<tr><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1profiling_1_1CallFrame.html">CallFrame</a> (<a class="el" href="namespacetvm_1_1runtime_1_1profiling.html">tvm::runtime::profiling</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalVarNode.html">GlobalVarNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1SearchCallbackNode.html">SearchCallbackNode [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CallLoweredAttrs.html">CallLoweredAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalVarSupply.html">GlobalVarSupply</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ObjAllocatorBase.html">ObjAllocatorBase</a> (<a class="el" href="namespacetv [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CallNode.html">CallNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1GlobalVarSupplyNode.html">GlobalVarSupplyNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Object.html">Object</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1CallNode.html">CallNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1usmp_1_1algo_1_1GreedyBase.html">GreedyBase</a> (<a class="el" href="namespacetvm_1_1tir_1_1usmp_1_1algo.html">tvm::tir::usmp::algo</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectEqual.html">ObjectEqual</a> (<a c [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1CallPattern.html">CallPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GridSampleAttrs.html">GridSampleAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectHash.html">ObjectHash</a> (<a class="el" href="namespa [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1CallPatternNode.html">CallPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1GroupNormAttrs.html">GroupNormAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1ObjectPath.html">ObjectPath</a> (<a class="el" href="namespacetvm. [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1CanonicalSimplifier.html">CanonicalSimplifier</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1GT.html">GT</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1ObjectPathNode.html">ObjectPathNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)& [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Cast.html">Cast</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1GTNode.html">GTNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1ObjectPathPair.html">ObjectPathPair</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td va [...]
<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CastAttrs.html">CastAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_h"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  h  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="classtvm_1_1ObjectPathNode.html">ObjectPathNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1meta__schedule_1_1SearchStrategyNode.html">SearchStrategyNode</a> (<a class="el" href="namespacetvm_1_1meta__schedule.html">tvm::meta_schedule</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1TypeRelationNode.html">TypeRelationNode</a> (<a class="el" [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CastHintAttrs.html">CastHintAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1ObjectPathPair.html">ObjectPathPair</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1SearchTask.html">SearchTask</a> (<a class="el" href="namespacetvm_1_1auto__s [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CastNode.html">CastNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1SimpleObjAllocator_1_1Handler.html">SimpleObjAllocator::Handler</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1ObjectPathPairNode.html">ObjectPathPairNode</a> (<a c [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ClassDoc.html">ClassDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1SEqualReducer_1_1Handler.html">SEqualReducer::Handler</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ObjectPtr.html">ObjectPtr</a> (<a class="el [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ClassDocNode.html">ClassDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1SHashReducer_1_1Handler.html">SHashReducer::Handler</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectPtrEqual.html">ObjectPtrEqual< [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Clause.html">Clause</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structdmlc_1_1serializer_1_1Handler_3_01DLDataType_01_4.html">Handler< DLDataType ></a> (<a class="el" href="namespacedmlc_1_1serializer.html">dmlc::serializer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectPtrHash.html">ObjectPt [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ClauseNode.html">ClauseNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structdmlc_1_1serializer_1_1Handler_3_01DLDevice_01_4.html">Handler< DLDevice ></a> (<a class="el" href="namespacedmlc_1_1serializer.html">dmlc::serializer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ObjectRef.html">ObjectRef [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1ObjectPathPairNode.html">ObjectPathPairNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1SearchTaskNode.html">SearchTaskNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1TypeVarNode.html">TypeVarNode</a> (<a class="el" href="n [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CastHintAttrs.html">CastHintAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ObjectPtr.html">ObjectPtr</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Select.html">Select</a> (<a class="el" href="namespacetvm_1_1tir.ht [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CastNode.html">CastNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1SimpleObjAllocator_1_1Handler.html">SimpleObjAllocator::Handler</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectPtrEqual.html">ObjectPtrEqual</a> ( [...]
</td></tr>
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ClipAttrs.html">ClipAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1HardwareParams.html">HardwareParams</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectTypeChecker.html">ObjectTypeChe [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Closure.html">Closure</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1HardwareParamsNode.html">HardwareParamsNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectTypeChecker_3_01Array_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ClosureObj.html">ClosureObj</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1HybridOp.html">HybridOp</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectTypeChecker_3_01Map_3_01K_00_01V_01_4_01_4.html">ObjectTypeChecker< Map [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CmpOpNode.html">CmpOpNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1HybridOpNode.html">HybridOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1OnDeviceAttrs.html">OnDeviceAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">t [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ClassDoc.html">ClassDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1SHashReducer_1_1Handler.html">SHashReducer::Handler</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectPtrHash.html">ObjectPtrHash</a> (<a cl [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1ClassDocNode.html">ClassDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1SEqualReducer_1_1Handler.html">SEqualReducer::Handler</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ObjectRef.html">ObjectRef</a> (<a c [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Clause.html">Clause</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structdmlc_1_1serializer_1_1Handler_3_01DLDataType_01_4.html">Handler< DLDataType ></a> (<a class="el" href="namespacedmlc_1_1serializer.html">dmlc::serializer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectTypeChecker.html">Obje [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ClauseNode.html">ClauseNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structdmlc_1_1serializer_1_1Handler_3_01DLDevice_01_4.html">Handler< DLDevice ></a> (<a class="el" href="namespacedmlc_1_1serializer.html">dmlc::serializer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectTypeChecker_3_01Ar [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ClipAttrs.html">ClipAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1HardwareParams.html">HardwareParams</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1ObjectTypeChecker_3_01Map_3_01K_00_01 [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Closure.html">Closure</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1HardwareParamsNode.html">HardwareParamsNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1OnDeviceAttrs.html">OnDeviceAt [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1ClosureObj.html">ClosureObj</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1HybridOp.html">HybridOp</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1OneHotAttrs.html">OneHotAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html" [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CmpOpNode.html">CmpOpNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1HybridOpNode.html">HybridOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1Op.html">Op</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign= [...]
<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1CommentDoc.html">CommentDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_i"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  i  </div></td></tr></table>
-</td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1OneHotAttrs.html">OneHotAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1SeqStmt.html">SeqStmt</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1UnknownAttributeAccessPath.html">UnknownAttributeAccessPath</a> (<a class="el" href="namespac [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1CommentDocNode.html">CommentDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1Op.html">Op</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1SeqStmtNode.html">SeqStmtNode</a> (<a class="el" href="namespacetvm_1_1tir.ht [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CommReducer.html">CommReducer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Id.html">Id</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1OpAttrMap.html">OpAttrMap</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CommReducerNode.html">CommReducerNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IdDoc.html">IdDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1Operation.html">Operation</a> (<a class="el" href=" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1CompilationConfig.html">CompilationConfig</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IdDocNode.html">IdDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1OperationDoc.html">OperationDoc</a> (<a cla [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1OpAttrMap.html">OpAttrMap</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1SEqualHandlerDefault.html">SEqualHandlerDefault</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_v"></a><table border="0" cellspacing="0" cellpadding="0"><tr><td><div class="ah">  v  </div></td> [...]
</td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1CompilationConfigNode.html">CompilationConfigNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1IdNode.html">IdNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1OperationDocNode.html">OperationDocNode</a> (<a class="el" href="namespacet [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1CompileError.html">CompileError</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1If.html">If</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1OperationNode.html">OperationNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CompilerAttrs.html">CompilerAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IfDoc.html">IfDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpImplementation.html">OpImplementation</a> ( [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeAtStep.html">ComputeAtStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IfDocNode.html">IfDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpImpleme [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeAtStepNode.html">ComputeAtStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1IfFrame.html">IfFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" hr [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeDAG.html">ComputeDAG</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1IfFrameNode.html">IfFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="cl [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeDAGNode.html">ComputeDAGNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1IfNode.html">IfNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpSpecialization.html">OpSpecialization</a> (<a [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeInlineStep.html">ComputeInlineStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1IfPattern.html">IfPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpSpecializationNode.html">OpSpecia [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeInlineStepNode.html">ComputeInlineStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1IfPatternNode.html">IfPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpStrategy.html">Op [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1te_1_1ComputeOp.html">ComputeOp</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IfThenElse.html">IfThenElse</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpStrategyNode.html">OpStrategyNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm: [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1te_1_1ComputeOpNode.html">ComputeOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IfThenElseNode.html">IfThenElseNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Optional.html">Optional</a> (<a class="el" href="namespacetvm_1_1runtime.ht [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeRootStep.html">ComputeRootStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplSEqualReduce.html">ImplSEqualReduce</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Or.html">Or</a> (<a cla [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeRootStepNode.html">ComputeRootStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplSEqualReduce_3_01T_00_01true_01_4.html">ImplSEqualReduce< T, true ></a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" hr [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ConcatenateAttrs.html">ConcatenateAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplSHashReduce.html">ImplSHashReduce</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_p"></a><table border="0" cellspacing="0" cellpadding="0"><t [...]
-</td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ShuffleNode.html">ShuffleNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1VirtualDeviceCache.html">VirtualDeviceCache</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Constant.html">Constant</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplSHashReduce_3_01T_00_01true_01_4.html">ImplSHashReduce< T, true ></a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1SignaturePrinter.html">Si [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1ConstantInfo.html">ConstantInfo</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplVisitAttrs.html">ImplVisitAttrs</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1PackedFunc.html">PackedFunc</a> (<a class="el" href="namespacetvm_1_1runtime.html"> [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ConstantInfoMetadata.html">ConstantInfoMetadata</a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplVisitAttrs_3_01T_00_01true_01_4.html">ImplVisitAttrs< T, true ></a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a clas [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ConstantInfoMetadataNode.html">ConstantInfoMetadataNode</a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IncompleteType.html">IncompleteType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1PackedFuncSubObj.html"> [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1ConstantInfoNode.html">ConstantInfoNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IncompleteTypeNode.html">IncompleteTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter.html">PackedFuncValueConverter</a> (<a class="el" href="namespacetvm_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1ConstantMemoryPools.html">ConstantMemoryPools</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IndexDataTypeNormalizer.html">IndexDataTypeNormalizer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_3_01Optional_3_01T_01_4_01_4.html"> [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1ConstantMemoryPoolsNode.html">ConstantMemoryPoolsNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IndexDataTypeRewriter.html">IndexDataTypeRewriter</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_3_01PrimExpr_01_4.html">Packed [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1CommentDocNode.html">CommentDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1Operation.html">Operation</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1SEqualReducer.html">SEqualReducer</a> (<a class="el" h [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CommReducer.html">CommReducer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Id.html">Id</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1OperationDoc.html">OperationDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1p [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1tir_1_1CommReducerNode.html">CommReducerNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IdDoc.html">IdDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1OperationDocNode.html">OperationDocN [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1CompilationConfig.html">CompilationConfig</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IdDocNode.html">IdDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1te_1_1OperationNode.html">OperationNode</a> (<a class="el" href= [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1CompilationConfigNode.html">CompilationConfigNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1IdNode.html">IdNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpImplementation.html">OpImplementation</a> (<a class="el" href="namespacetvm_1_1relay. [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1CompileError.html">CompileError</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1If.html">If</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpImplementationNode.html">OpImplementationNode</a> (<a class="el" href="namespacetvm_1_ [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1CompilerAttrs.html">CompilerAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IfDoc.html">IfDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1OpNode.html">OpNode</a> (<a class="el" href="namespace [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeAtStep.html">ComputeAtStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IfDocNode.html">IfDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1OpRegEntry.html">O [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeAtStepNode.html">ComputeAtStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1IfFrame.html">IfFrame</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" hr [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeDAG.html">ComputeDAG</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1ir__builder_1_1tir_1_1IfFrameNode.html">IfFrameNode</a> (<a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html">tvm::script::ir_builder::tir</a>)   </td><td valign="top"><a class="el" href="cl [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeDAGNode.html">ComputeDAGNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1IfNode.html">IfNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpStrategy.html">OpStrategy</a> (<a class="el" [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeInlineStep.html">ComputeInlineStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1IfPattern.html">IfPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1OpStrategyNode.html">OpStrategyNode [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeInlineStepNode.html">ComputeInlineStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1IfPatternNode.html">IfPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1Optional.html">Op [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1te_1_1ComputeOp.html">ComputeOp</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IfThenElse.html">IfThenElse</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Or.html">Or</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   < [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1te_1_1ComputeOpNode.html">ComputeOpNode</a> (<a class="el" href="namespacetvm_1_1te.html">tvm::te</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IfThenElseNode.html">IfThenElseNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1OrNode.html">OrNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeRootStep.html">ComputeRootStep</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplSEqualReduce.html">ImplSEqualReduce</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td rowspan="2" valign="bottom"><a name="letter_p"></a><table border="0" cellsp [...]
+</td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1ShuffleNode.html">ShuffleNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1vm_1_1VirtualMachine.html">VirtualMachine</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td></tr>
+<tr><td valign="top"><a class="el" href="classtvm_1_1auto__scheduler_1_1ComputeRootStepNode.html">ComputeRootStepNode</a> (<a class="el" href="namespacetvm_1_1auto__scheduler.html">tvm::auto_scheduler</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplSEqualReduce_3_01T_00_01true_01_4.html">ImplSEqualReduce< T, true ></a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" hr [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ConcatenateAttrs.html">ConcatenateAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplSHashReduce.html">ImplSHashReduce</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1PackedFunc.html">PackedFunc</a> (<a class="el" [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Constant.html">Constant</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplSHashReduce_3_01T_00_01true_01_4.html">ImplSHashReduce< T, true ></a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1PackedFuncObj.html">Packed [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1ConstantInfo.html">ConstantInfo</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplVisitAttrs.html">ImplVisitAttrs</a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1PackedFuncSubObj.html">PackedFuncSubObj</a> (<a class="el" href="namespacetvm_1_1ru [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ConstantInfoMetadata.html">ConstantInfoMetadata</a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1detail_1_1ImplVisitAttrs_3_01T_00_01true_01_4.html">ImplVisitAttrs< T, true ></a> (<a class="el" href="namespacetvm_1_1detail.html">tvm::detail</a>)   </td><td valign="top"><a clas [...]
</td></tr>
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ConstantNode.html">ConstantNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IndexDoc.html">IndexDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_3_01tvm_1_1Boo [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ConstantPattern.html">ConstantPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IndexDocNode.html">IndexDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ConstantPatternNode.html">ConstantPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IndexMap.html">IndexMap</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_3_1_1tvm_1_1runtime_1_1String_01_4.html" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1ConstantPoolInfo.html">ConstantPoolInfo</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IndexMapNode.html">IndexMapNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1micro__rpc_1_1PacketFieldSizeBytes.html">PacketFieldSizeBytes</a> (<a class="el" href="na [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1ConstantPoolInfoNode.html">ConstantPoolInfoNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1InitOpAttrs.html">InitOpAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1PadAttrs.html">PadAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ConstIntBound.html">ConstIntBound</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1InplaceArrayBase.html">InplaceArrayBase</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1Pass.html">Pass</a> (<a class="el" href="namespa [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ConstIntBoundAnalyzer.html">ConstIntBoundAnalyzer</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1InstanceNormAttrs.html">InstanceNormAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassContext.html">PassContext</a> ( [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ConstIntBoundNode.html">ConstIntBoundNode</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Instruction.html">Instruction</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassContextNode.html">PassContextNode</a> (<a class="el" href= [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ConstraintContext.html">ConstraintContext</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1vm_1_1Instruction.html">Instruction</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassInfo.html">PassInfo</a> (<a c [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1Constructor.html">Constructor</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1InstructionKind.html">InstructionKind</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassInfoNode.html">PassInfoNode</a> (<a class="el" href="namespacetvm_1_1transform.html">tv [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1ConstructorNode.html">ConstructorNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1InstructionKindNode.html">InstructionKindNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1instrument_1_1PassInstrument.html">PassInstrument</a> (<a class="el" href="namespacetvm_ [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ConstructorValue.html">ConstructorValue</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1InstructionKindRegEntry.html">InstructionKindRegEntry</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1instrument_1_1PassInstrumentNode.html">PassInstrument [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ConstructorValueObj.html">ConstructorValueObj</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1InstructionNode.html">InstructionNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassNode.html">PassNode</a> (<a class="el" href=" [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1NDArray_1_1Container.html">NDArray::Container</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntConstraints.html">IntConstraints</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Pattern.html">Pattern</a> (<a class="el" hre [...]
-<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1NDArray_1_1ContainerBase.html">NDArray::ContainerBase</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntConstraintsNode.html">IntConstraintsNode</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternConstructor.html">Pat [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv1DAttrs.html">Conv1DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntConstraintsTransform.html">IntConstraintsTransform</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternConstructorNode.html">PatternConstructor [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv1DTransposeAttrs.html">Conv1DTransposeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntConstraintsTransformNode.html">IntConstraintsTransformNode</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternFunctor.html"> [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv2DAttrs.html">Conv2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1Integer.html">Integer</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternFunctor_3_01R_07const_01Pattern_01_6n_00_01Args_8_8_8_08_4.html">PatternFunctor< R(const Patte [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv2DTransposeAttrs.html">Conv2DTransposeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1InterpreterClosure.html">InterpreterClosure</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternMutator.html">PatternMutator</a> [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv2DWinogradAttrs.html">Conv2DWinogradAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1InterpreterClosureObj.html">InterpreterClosureObj</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternNode.html">PatternNode</a> ( [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1metadata_1_1ConstantInfoMetadataNode.html">ConstantInfoMetadataNode</a> (<a class="el" href="namespacetvm_1_1runtime_1_1metadata.html">tvm::runtime::metadata</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IncompleteType.html">IncompleteType</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConvert [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1ConstantInfoNode.html">ConstantInfoNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1IncompleteTypeNode.html">IncompleteTypeNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_3_01PrimExpr_01_4.html">PackedFuncValueConverter< PrimExpr ></a [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1ConstantMemoryPools.html">ConstantMemoryPools</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IndexDataTypeNormalizer.html">IndexDataTypeNormalizer</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_3_01tvm_1_1Bool_01_4.html">PackedFu [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1ConstantMemoryPoolsNode.html">ConstantMemoryPoolsNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IndexDataTypeRewriter.html">IndexDataTypeRewriter</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_3_01tvm_1_1Integer_01_4.html"> [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ConstantNode.html">ConstantNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IndexDoc.html">IndexDoc</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1PackedFuncValueConverter_3_1_1tvm_1_1ru [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ConstantPattern.html">ConstantPattern</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1script_1_1printer_1_1IndexDocNode.html">IndexDocNode</a> (<a class="el" href="namespacetvm_1_1script_1_1printer.html">tvm::script::printer</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1micro__rpc_1_1PacketFieldS [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ConstantPatternNode.html">ConstantPatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IndexMap.html">IndexMap</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1PadAttrs.html">PadAttrs</a> (<a class="el" href="namespacetvm_1_1re [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1ConstantPoolInfo.html">ConstantPoolInfo</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1IndexMapNode.html">IndexMapNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1Pass.html">Pass</a> (<a class="el" href="namespacetvm_1_1transform.html">tvm::transform [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1ConstantPoolInfoNode.html">ConstantPoolInfoNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1InitOpAttrs.html">InitOpAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassContext.html">PassContext</a> (<a class="el" href="namespacetvm_1_1tr [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ConstIntBound.html">ConstIntBound</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1InplaceArrayBase.html">InplaceArrayBase</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassContextNode.html">PassContextNode</a> (<a cl [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ConstIntBoundAnalyzer.html">ConstIntBoundAnalyzer</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1relay_1_1InstanceNormAttrs.html">InstanceNormAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassInfo.html">PassInfo</a> (<a cla [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ConstIntBoundNode.html">ConstIntBoundNode</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1Instruction.html">Instruction</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassInfoNode.html">PassInfoNode</a> (<a class="el" href="names [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1arith_1_1ConstraintContext.html">ConstraintContext</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="structtvm_1_1runtime_1_1vm_1_1Instruction.html">Instruction</a> (<a class="el" href="namespacetvm_1_1runtime_1_1vm.html">tvm::runtime::vm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1instrument_1_1PassInstrument.html">PassInstrum [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1Constructor.html">Constructor</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1InstructionKind.html">InstructionKind</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1instrument_1_1PassInstrumentNode.html">PassInstrumentNode</a> (<a class="el" href="namespacetvm_1_1instr [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1ConstructorNode.html">ConstructorNode</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1InstructionKindNode.html">InstructionKindNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1transform_1_1PassNode.html">PassNode</a> (<a class="el" href="namespacetvm_1_1transform. [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1relay_1_1ConstructorValue.html">ConstructorValue</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1InstructionKindRegEntry.html">InstructionKindRegEntry</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Pattern.html">Pattern</a> (<a class="el" hre [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1ConstructorValueObj.html">ConstructorValueObj</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1tir_1_1InstructionNode.html">InstructionNode</a> (<a class="el" href="namespacetvm_1_1tir.html">tvm::tir</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternConstructor.html">PatternConstructor</a> (<a c [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1NDArray_1_1Container.html">NDArray::Container</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntConstraints.html">IntConstraints</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternConstructorNode.html">PatternConstruc [...]
+<tr><td valign="top"><a class="el" href="classtvm_1_1runtime_1_1NDArray_1_1ContainerBase.html">NDArray::ContainerBase</a> (<a class="el" href="namespacetvm_1_1runtime.html">tvm::runtime</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntConstraintsNode.html">IntConstraintsNode</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternFunctor.html">Pattern [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv1DAttrs.html">Conv1DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntConstraintsTransform.html">IntConstraintsTransform</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternFunctor_3_01R_07const_01Pattern_01_6n_00 [...]
</td></tr>
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv2DWinogradNNPACKWeightTransformAttrs.html">Conv2DWinogradNNPACKWeightTransformAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntGroupBounds.html">IntGroupBounds</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Pattern [...]
-<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv3DAttrs.html">Conv3DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntGroupBoundsNode.html">IntGroupBoundsNode</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternTupleNode.html">PatternTupleNode</a> (<a class="el [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv1DTransposeAttrs.html">Conv1DTransposeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntConstraintsTransformNode.html">IntConstraintsTransformNode</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternMutator.html"> [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv2DAttrs.html">Conv2DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1Integer.html">Integer</a> (<a class="el" href="namespacetvm.html">tvm</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternNode.html">PatternNode</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)  [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv2DTransposeAttrs.html">Conv2DTransposeAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1InterpreterClosure.html">InterpreterClosure</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternTuple.html">PatternTuple</a> (<a [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv2DWinogradAttrs.html">Conv2DWinogradAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1InterpreterClosureObj.html">InterpreterClosureObj</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternTupleNode.html">PatternTuple [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv2DWinogradNNPACKWeightTransformAttrs.html">Conv2DWinogradNNPACKWeightTransformAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntGroupBounds.html">IntGroupBounds</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1Pattern [...]
+<tr><td valign="top"><a class="el" href="structtvm_1_1relay_1_1Conv3DAttrs.html">Conv3DAttrs</a> (<a class="el" href="namespacetvm_1_1relay.html">tvm::relay</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1arith_1_1IntGroupBoundsNode.html">IntGroupBoundsNode</a> (<a class="el" href="namespacetvm_1_1arith.html">tvm::arith</a>)   </td><td valign="top"><a class="el" href="classtvm_1_1relay_1_1PatternVarNode.html">PatternVarNode</a> (<a class="el" hr [...]
<tr><td></td><td></td><td></td><td></td><td></td></tr>
</table>
<div class="qindex"><a class="qindex" href="#letter_a">a</a> | <a class="qindex" href="#letter_b">b</a> | <a class="qindex" href="#letter_c">c</a> | <a class="qindex" href="#letter_d">d</a> | <a class="qindex" href="#letter_e">e</a> | <a class="qindex" href="#letter_f">f</a> | <a class="qindex" href="#letter_g">g</a> | <a class="qindex" href="#letter_h">h</a> | <a class="qindex" href="#letter_i">i</a> |& [...]
diff --git a/docs/reference/api/doxygen/classtvm_1_1BaseExprNode__coll__graph.svg b/docs/reference/api/doxygen/classtvm_1_1BaseExprNode__coll__graph.svg
index d46f7c9a0a..5897cd113d 100644
--- a/docs/reference/api/doxygen/classtvm_1_1BaseExprNode__coll__graph.svg
+++ b/docs/reference/api/doxygen/classtvm_1_1BaseExprNode__coll__graph.svg
@@ -9,9 +9,9 @@
<g id="graph0" class="graph" transform="scale(1 1) rotate(0) translate(4 1038)">
<title>tvm::BaseExprNode</title>
<polygon fill="#ffffff" stroke="transparent" points="-4,4 -4,-1038 429.5,-1038 429.5,4 -4,4"/>
-<!-- Node77 -->
+<!-- Node76 -->
<g id="node1" class="node">
-<title>Node77</title>
+<title>Node76</title>
<polygon fill="#bfbfbf" stroke="#000000" points="116,-.5 116,-112.5 323,-112.5 323,-.5 116,-.5"/>
<text text-anchor="middle" x="219.5" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::BaseExprNode</text>
<polyline fill="none" stroke="#000000" points="116,-93.5 323,-93.5 "/>
@@ -24,9 +24,9 @@
<polyline fill="none" stroke="#000000" points="116,-19.5 323,-19.5 "/>
<text text-anchor="start" x="124" y="-7.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
</g>
-<!-- Node78 -->
+<!-- Node77 -->
<g id="node2" class="node">
-<title>Node78</title>
+<title>Node77</title>
<g id="a_node2"><a xlink:href="classtvm_1_1runtime_1_1Object.html" target="_top" xlink:title="base class of all object containers. ">
<polygon fill="#ffffff" stroke="#000000" points="0,-160.5 0,-547.5 183,-547.5 183,-160.5 0,-160.5"/>
<text text-anchor="middle" x="91.5" y="-535.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::Object</text>
@@ -67,22 +67,22 @@
</a>
</g>
</g>
-<!-- Node78->Node77 -->
+<!-- Node77->Node76 -->
<g id="edge1" class="edge">
-<title>Node78->Node77</title>
+<title>Node77->Node76</title>
<path fill="none" stroke="#191970" d="M178.9918,-150.6499C184.8201,-137.1037 190.3466,-124.2589 195.3366,-112.6609"/>
<polygon fill="none" stroke="#191970" points="175.6487,-149.5643 174.9115,-160.1334 182.0788,-152.3309 175.6487,-149.5643"/>
</g>
-<!-- Node78->Node78 -->
+<!-- Node77->Node77 -->
<g id="edge2" class="edge">
-<title>Node78->Node78</title>
+<title>Node77->Node77</title>
<path fill="none" stroke="#404040" d="M183.3625,-387.9248C194.0482,-381.6637 201,-370.3555 201,-354 201,-343.0112 197.8618,-334.3007 192.5615,-327.8687"/>
<polygon fill="none" stroke="#404040" points="192.5184,-327.8322 185.3548,-327.0056 183.3625,-320.0752 190.5261,-320.9017 192.5184,-327.8322"/>
<text text-anchor="middle" x="227" y="-351.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> #deleter_</text>
</g>
-<!-- Node79 -->
+<!-- Node78 -->
<g id="node3" class="node">
-<title>Node79</title>
+<title>Node78</title>
<g id="a_node3"><a xlink:href="classtvm_1_1Span.html" target="_top" xlink:title="{tvm::Span\n||+ Span()\l+ Merge()\l+ TVM_DEFINE_OBJECT_REF\l_METHODS()\l}">
<polygon fill="#ffffff" stroke="#000000" points="271.5,-309 271.5,-399 425.5,-399 425.5,-309 271.5,-309"/>
<text text-anchor="middle" x="348.5" y="-387" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::Span</text>
@@ -96,16 +96,16 @@
</a>
</g>
</g>
-<!-- Node79->Node77 -->
+<!-- Node78->Node76 -->
<g id="edge3" class="edge">
-<title>Node79->Node77</title>
+<title>Node78->Node76</title>
<path fill="none" stroke="#404040" d="M328.9077,-308.8162C307.576,-259.621 273.2346,-180.4229 248.7175,-123.8815"/>
<polygon fill="none" stroke="#404040" points="248.5716,-123.5448 242.5148,-119.6314 243.7977,-112.5353 249.8545,-116.4487 248.5716,-123.5448"/>
<text text-anchor="middle" x="271" y="-134" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> +span</text>
</g>
-<!-- Node80 -->
+<!-- Node79 -->
<g id="node4" class="node">
-<title>Node80</title>
+<title>Node79</title>
<g id="a_node4"><a xlink:href="classtvm_1_1runtime_1_1ObjectRef.html" target="_top" xlink:title="Base class of all object reference. ">
<polygon fill="#ffffff" stroke="#000000" points="281.5,-585.5 281.5,-807.5 415.5,-807.5 415.5,-585.5 281.5,-585.5"/>
<text text-anchor="middle" x="348.5" y="-795.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::ObjectRef</text>
@@ -131,15 +131,15 @@
</a>
</g>
</g>
-<!-- Node80->Node79 -->
+<!-- Node79->Node78 -->
<g id="edge4" class="edge">
-<title>Node80->Node79</title>
+<title>Node79->Node78</title>
<path fill="none" stroke="#191970" d="M348.5,-575.1895C348.5,-514.0976 348.5,-443.7137 348.5,-399.0565"/>
<polygon fill="none" stroke="#191970" points="345.0001,-575.3 348.5,-585.3001 352.0001,-575.3001 345.0001,-575.3"/>
</g>
-<!-- Node81 -->
+<!-- Node80 -->
<g id="node5" class="node">
-<title>Node81</title>
+<title>Node80</title>
<g id="a_node5"><a xlink:href="classtvm_1_1runtime_1_1ObjectPtr.html" target="_top" xlink:title="{tvm::runtime::ObjectPtr\l\< tvm::runtime::Object \>\n||+ ObjectPtr()\l+ ObjectPtr()\l+ ObjectPtr()\l+ ObjectPtr()\l+ ObjectPtr()\l+ ObjectPtr()\l+ ~ObjectPtr()\l+ swap()\l+ get()\l+ operator-\>()\land 11 more...\l}">
<polygon fill="#ffffff" stroke="#000000" points="278.5,-855.5 278.5,-1033.5 418.5,-1033.5 418.5,-855.5 278.5,-855.5"/>
<text text-anchor="start" x="286.5" y="-1021.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::ObjectPtr</text>
@@ -161,9 +161,9 @@
</a>
</g>
</g>
-<!-- Node81->Node80 -->
+<!-- Node80->Node79 -->
<g id="edge5" class="edge">
-<title>Node81->Node80</title>
+<title>Node80->Node79</title>
<path fill="none" stroke="#404040" d="M348.5,-855.3167C348.5,-843.8765 348.5,-832.0062 348.5,-820.1402"/>
<polygon fill="none" stroke="#404040" points="348.5001,-819.7944 344.5,-813.7944 348.5,-807.7944 352.5,-813.7943 348.5001,-819.7944"/>
<text text-anchor="middle" x="368" y="-829" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> #data_</text>
diff --git a/docs/reference/api/doxygen/classtvm_1_1BaseExprNode__inherit__graph.svg b/docs/reference/api/doxygen/classtvm_1_1BaseExprNode__inherit__graph.svg
index a155811203..2b48fa4857 100644
--- a/docs/reference/api/doxygen/classtvm_1_1BaseExprNode__inherit__graph.svg
+++ b/docs/reference/api/doxygen/classtvm_1_1BaseExprNode__inherit__graph.svg
@@ -12,116 +12,116 @@
<!-- Node0 -->
<g id="node1" class="node">
<title>Node0</title>
-<polygon fill="#bfbfbf" stroke="#000000" points="6580,-386.5 6580,-509.5 6787,-509.5 6787,-386.5 6580,-386.5"/>
-<text text-anchor="middle" x="6683.5" y="-497.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::BaseExprNode</text>
-<polyline fill="none" stroke="#000000" points="6580,-490.5 6787,-490.5 "/>
-<text text-anchor="start" x="6588" y="-478.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ span</text>
-<text text-anchor="start" x="6588" y="-467.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<text text-anchor="start" x="6588" y="-456.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_sequal</text>
-<text text-anchor="start" x="6588" y="-445.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_reduce</text>
-<text text-anchor="start" x="6588" y="-434.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_shash</text>
-<text text-anchor="start" x="6588" y="-423.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_reduce</text>
-<text text-anchor="start" x="6588" y="-412.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
-<polyline fill="none" stroke="#000000" points="6580,-405.5 6787,-405.5 "/>
-<text text-anchor="start" x="6588" y="-393.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
+<polygon fill="#bfbfbf" stroke="#000000" points="6466,-386.5 6466,-509.5 6673,-509.5 6673,-386.5 6466,-386.5"/>
+<text text-anchor="middle" x="6569.5" y="-497.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::BaseExprNode</text>
+<polyline fill="none" stroke="#000000" points="6466,-490.5 6673,-490.5 "/>
+<text text-anchor="start" x="6474" y="-478.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ span</text>
+<text text-anchor="start" x="6474" y="-467.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<text text-anchor="start" x="6474" y="-456.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_sequal</text>
+<text text-anchor="start" x="6474" y="-445.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_reduce</text>
+<text text-anchor="start" x="6474" y="-434.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_shash</text>
+<text text-anchor="start" x="6474" y="-423.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_reduce</text>
+<text text-anchor="start" x="6474" y="-412.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
+<polyline fill="none" stroke="#000000" points="6466,-405.5 6673,-405.5 "/>
+<text text-anchor="start" x="6474" y="-393.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
</g>
<!-- Node2 -->
<g id="node3" class="node">
<title>Node2</title>
<g id="a_node3"><a xlink:href="classtvm_1_1PrimExprNode.html" target="_top" xlink:title="Base node of all primitive expressions. ">
-<polygon fill="#ffffff" stroke="#000000" points="4199,-237.5 4199,-338.5 4406,-338.5 4406,-237.5 4199,-237.5"/>
-<text text-anchor="middle" x="4302.5" y="-326.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::PrimExprNode</text>
-<polyline fill="none" stroke="#000000" points="4199,-319.5 4406,-319.5 "/>
-<text text-anchor="start" x="4207" y="-307.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ dtype</text>
-<text text-anchor="start" x="4207" y="-296.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<text text-anchor="start" x="4207" y="-285.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
-<polyline fill="none" stroke="#000000" points="4199,-278.5 4406,-278.5 "/>
-<text text-anchor="start" x="4207" y="-266.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_OBJECT_ENABLE_SCRIPT</text>
-<text text-anchor="start" x="4207" y="-255.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_PRINTER()</text>
-<text text-anchor="start" x="4207" y="-244.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
+<polygon fill="#ffffff" stroke="#000000" points="4085,-237.5 4085,-338.5 4292,-338.5 4292,-237.5 4085,-237.5"/>
+<text text-anchor="middle" x="4188.5" y="-326.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::PrimExprNode</text>
+<polyline fill="none" stroke="#000000" points="4085,-319.5 4292,-319.5 "/>
+<text text-anchor="start" x="4093" y="-307.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ dtype</text>
+<text text-anchor="start" x="4093" y="-296.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<text text-anchor="start" x="4093" y="-285.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
+<polyline fill="none" stroke="#000000" points="4085,-278.5 4292,-278.5 "/>
+<text text-anchor="start" x="4093" y="-266.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_OBJECT_ENABLE_SCRIPT</text>
+<text text-anchor="start" x="4093" y="-255.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_PRINTER()</text>
+<text text-anchor="start" x="4093" y="-244.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node0->Node2 -->
<g id="edge2" class="edge">
<title>Node0->Node2</title>
-<path fill="none" stroke="#191970" d="M6569.3877,-440.3318C6163.7302,-413.0722 4794.915,-321.0896 4406.2144,-294.9695"/>
-<polygon fill="none" stroke="#191970" points="6569.4482,-443.8437 6579.6604,-441.0221 6569.9176,-436.8594 6569.4482,-443.8437"/>
+<path fill="none" stroke="#191970" d="M6455.3877,-440.3318C6049.7302,-413.0722 4680.915,-321.0896 4292.2144,-294.9695"/>
+<polygon fill="none" stroke="#191970" points="6455.4482,-443.8437 6465.6604,-441.0221 6455.9176,-436.8594 6455.4482,-443.8437"/>
</g>
-<!-- Node58 -->
-<g id="node41" class="node">
-<title>Node58</title>
-<g id="a_node41"><a xlink:href="classtvm_1_1RelayExprNode.html" target="_top" xlink:title="Base node of all non-primitive expressions. ">
-<polygon fill="#ffffff" stroke="#ff0000" points="9075,-226.5 9075,-349.5 9282,-349.5 9282,-226.5 9075,-226.5"/>
-<text text-anchor="middle" x="9178.5" y="-337.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::RelayExprNode</text>
-<polyline fill="none" stroke="#ff0000" points="9075,-330.5 9282,-330.5 "/>
-<text text-anchor="start" x="9083" y="-318.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ checked_type_</text>
-<text text-anchor="start" x="9083" y="-307.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ virtual_device_</text>
-<text text-anchor="start" x="9083" y="-296.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<text text-anchor="start" x="9083" y="-285.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
-<polyline fill="none" stroke="#ff0000" points="9075,-278.5 9282,-278.5 "/>
-<text text-anchor="start" x="9083" y="-266.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ checked_type()</text>
-<text text-anchor="start" x="9083" y="-255.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ type_as()</text>
-<text text-anchor="start" x="9083" y="-244.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ virtual_device()</text>
-<text text-anchor="start" x="9083" y="-233.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
+<!-- Node57 -->
+<g id="node40" class="node">
+<title>Node57</title>
+<g id="a_node40"><a xlink:href="classtvm_1_1RelayExprNode.html" target="_top" xlink:title="Base node of all non-primitive expressions. ">
+<polygon fill="#ffffff" stroke="#ff0000" points="8961,-226.5 8961,-349.5 9168,-349.5 9168,-226.5 8961,-226.5"/>
+<text text-anchor="middle" x="9064.5" y="-337.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::RelayExprNode</text>
+<polyline fill="none" stroke="#ff0000" points="8961,-330.5 9168,-330.5 "/>
+<text text-anchor="start" x="8969" y="-318.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ checked_type_</text>
+<text text-anchor="start" x="8969" y="-307.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ virtual_device_</text>
+<text text-anchor="start" x="8969" y="-296.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<text text-anchor="start" x="8969" y="-285.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
+<polyline fill="none" stroke="#ff0000" points="8961,-278.5 9168,-278.5 "/>
+<text text-anchor="start" x="8969" y="-266.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ checked_type()</text>
+<text text-anchor="start" x="8969" y="-255.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ type_as()</text>
+<text text-anchor="start" x="8969" y="-244.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ virtual_device()</text>
+<text text-anchor="start" x="8969" y="-233.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
</a>
</g>
</g>
-<!-- Node0->Node58 -->
-<g id="edge40" class="edge">
-<title>Node0->Node58</title>
-<path fill="none" stroke="#191970" d="M6797.1435,-440.7122C7216.3003,-413.8324 8672.1316,-320.4725 9074.6621,-294.6589"/>
-<polygon fill="none" stroke="#191970" points="6796.892,-437.2211 6787.1365,-441.354 6797.34,-444.2067 6796.892,-437.2211"/>
+<!-- Node0->Node57 -->
+<g id="edge39" class="edge">
+<title>Node0->Node57</title>
+<path fill="none" stroke="#191970" d="M6683.1435,-440.7122C7102.3003,-413.8324 8558.1316,-320.4725 8960.6621,-294.6589"/>
+<polygon fill="none" stroke="#191970" points="6682.892,-437.2211 6673.1365,-441.354 6683.34,-444.2067 6682.892,-437.2211"/>
</g>
<!-- Node1 -->
<g id="node2" class="node">
<title>Node1</title>
<g id="a_node2"><a xlink:href="classtvm_1_1runtime_1_1Object.html" target="_top" xlink:title="base class of all object containers. ">
-<polygon fill="#ffffff" stroke="#000000" points="6592,-546.5 6592,-944.5 6775,-944.5 6775,-546.5 6592,-546.5"/>
-<text text-anchor="middle" x="6683.5" y="-932.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::Object</text>
-<polyline fill="none" stroke="#000000" points="6592,-925.5 6775,-925.5 "/>
-<text text-anchor="start" x="6600" y="-913.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<text text-anchor="start" x="6600" y="-902.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_final</text>
-<text text-anchor="start" x="6600" y="-891.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
-<text text-anchor="start" x="6600" y="-880.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots_can</text>
-<text text-anchor="start" x="6600" y="-869.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_overflow</text>
-<text text-anchor="start" x="6600" y="-858.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_visit</text>
-<text text-anchor="start" x="6600" y="-847.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_attrs</text>
-<text text-anchor="start" x="6600" y="-836.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_sequal</text>
-<text text-anchor="start" x="6600" y="-825.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_reduce</text>
-<text text-anchor="start" x="6600" y="-814.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_shash</text>
-<text text-anchor="start" x="6600" y="-803.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_reduce</text>
-<text text-anchor="start" x="6600" y="-792.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_index</text>
-<text text-anchor="start" x="6600" y="-781.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># type_index_</text>
-<text text-anchor="start" x="6600" y="-770.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># ref_counter_</text>
-<text text-anchor="start" x="6600" y="-759.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># deleter_</text>
-<polyline fill="none" stroke="#000000" points="6592,-752.5 6775,-752.5 "/>
-<text text-anchor="start" x="6600" y="-740.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ type_index()</text>
-<text text-anchor="start" x="6600" y="-729.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ GetTypeKey()</text>
-<text text-anchor="start" x="6600" y="-718.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ GetTypeKeyHash()</text>
-<text text-anchor="start" x="6600" y="-707.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ IsInstance()</text>
-<text text-anchor="start" x="6600" y="-696.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ unique()</text>
-<text text-anchor="start" x="6600" y="-685.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Object()</text>
-<text text-anchor="start" x="6600" y="-674.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Object()</text>
-<text text-anchor="start" x="6600" y="-663.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Object()</text>
-<text text-anchor="start" x="6600" y="-652.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator=()</text>
-<text text-anchor="start" x="6600" y="-641.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator=()</text>
-<text text-anchor="start" x="6600" y="-630.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TypeIndex2Key()</text>
-<text text-anchor="start" x="6600" y="-619.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TypeIndex2KeyHash()</text>
-<text text-anchor="start" x="6600" y="-608.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TypeKey2Index()</text>
-<text text-anchor="start" x="6600" y="-597.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _GetOrAllocRuntimeTypeIndex()</text>
-<text text-anchor="start" x="6600" y="-586.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ RuntimeTypeIndex()</text>
-<text text-anchor="start" x="6600" y="-575.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># IncRef()</text>
-<text text-anchor="start" x="6600" y="-564.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># DecRef()</text>
-<text text-anchor="start" x="6600" y="-553.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># GetOrAllocRuntimeTypeIndex()</text>
+<polygon fill="#ffffff" stroke="#000000" points="6478,-546.5 6478,-944.5 6661,-944.5 6661,-546.5 6478,-546.5"/>
+<text text-anchor="middle" x="6569.5" y="-932.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::Object</text>
+<polyline fill="none" stroke="#000000" points="6478,-925.5 6661,-925.5 "/>
+<text text-anchor="start" x="6486" y="-913.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<text text-anchor="start" x="6486" y="-902.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_final</text>
+<text text-anchor="start" x="6486" y="-891.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
+<text text-anchor="start" x="6486" y="-880.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots_can</text>
+<text text-anchor="start" x="6486" y="-869.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_overflow</text>
+<text text-anchor="start" x="6486" y="-858.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_visit</text>
+<text text-anchor="start" x="6486" y="-847.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_attrs</text>
+<text text-anchor="start" x="6486" y="-836.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_sequal</text>
+<text text-anchor="start" x="6486" y="-825.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_reduce</text>
+<text text-anchor="start" x="6486" y="-814.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_has_method_shash</text>
+<text text-anchor="start" x="6486" y="-803.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_reduce</text>
+<text text-anchor="start" x="6486" y="-792.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_index</text>
+<text text-anchor="start" x="6486" y="-781.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># type_index_</text>
+<text text-anchor="start" x="6486" y="-770.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># ref_counter_</text>
+<text text-anchor="start" x="6486" y="-759.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># deleter_</text>
+<polyline fill="none" stroke="#000000" points="6478,-752.5 6661,-752.5 "/>
+<text text-anchor="start" x="6486" y="-740.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ type_index()</text>
+<text text-anchor="start" x="6486" y="-729.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ GetTypeKey()</text>
+<text text-anchor="start" x="6486" y="-718.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ GetTypeKeyHash()</text>
+<text text-anchor="start" x="6486" y="-707.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ IsInstance()</text>
+<text text-anchor="start" x="6486" y="-696.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ unique()</text>
+<text text-anchor="start" x="6486" y="-685.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Object()</text>
+<text text-anchor="start" x="6486" y="-674.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Object()</text>
+<text text-anchor="start" x="6486" y="-663.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Object()</text>
+<text text-anchor="start" x="6486" y="-652.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator=()</text>
+<text text-anchor="start" x="6486" y="-641.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator=()</text>
+<text text-anchor="start" x="6486" y="-630.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TypeIndex2Key()</text>
+<text text-anchor="start" x="6486" y="-619.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TypeIndex2KeyHash()</text>
+<text text-anchor="start" x="6486" y="-608.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TypeKey2Index()</text>
+<text text-anchor="start" x="6486" y="-597.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _GetOrAllocRuntimeTypeIndex()</text>
+<text text-anchor="start" x="6486" y="-586.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ RuntimeTypeIndex()</text>
+<text text-anchor="start" x="6486" y="-575.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># IncRef()</text>
+<text text-anchor="start" x="6486" y="-564.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># DecRef()</text>
+<text text-anchor="start" x="6486" y="-553.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># GetOrAllocRuntimeTypeIndex()</text>
</a>
</g>
</g>
<!-- Node1->Node0 -->
<g id="edge1" class="edge">
<title>Node1->Node0</title>
-<path fill="none" stroke="#191970" d="M6683.5,-536.0774C6683.5,-526.8427 6683.5,-517.9846 6683.5,-509.6845"/>
-<polygon fill="none" stroke="#191970" points="6680.0001,-536.2371 6683.5,-546.2371 6687.0001,-536.2372 6680.0001,-536.2371"/>
+<path fill="none" stroke="#191970" d="M6569.5,-536.0774C6569.5,-526.8427 6569.5,-517.9846 6569.5,-509.6845"/>
+<polygon fill="none" stroke="#191970" points="6566.0001,-536.2371 6569.5,-546.2371 6573.0001,-536.2372 6566.0001,-536.2371"/>
</g>
<!-- Node3 -->
<g id="node4" class="node">
@@ -144,8 +144,8 @@
<!-- Node2->Node3 -->
<g id="edge3" class="edge">
<title>Node2->Node3</title>
-<path fill="none" stroke="#191970" d="M4188.5446,-287.3936C3547.0403,-283.7127 408.0701,-262.0643 217.5,-190 195.2051,-181.5692 174.1958,-166.7264 156.4805,-151.243"/>
-<polygon fill="none" stroke="#191970" points="4188.698,-290.8944 4198.7178,-287.4516 4188.738,-283.8945 4188.698,-290.8944"/>
+<path fill="none" stroke="#191970" d="M4074.6679,-287.3125C3444.219,-283.2383 402.4302,-260.0043 217.5,-190 195.208,-181.5615 174.1995,-166.7166 156.484,-151.2337"/>
+<polygon fill="none" stroke="#191970" points="4074.6472,-290.8124 4084.6696,-287.3768 4074.6923,-283.8125 4074.6472,-290.8124"/>
</g>
<!-- Node5 -->
<g id="node5" class="node">
@@ -168,8 +168,8 @@
<!-- Node2->Node5 -->
<g id="edge4" class="edge">
<title>Node2->Node5</title>
-<path fill="none" stroke="#191970" d="M4188.4437,-287.2251C3568.3448,-282.745 623.8103,-257.9509 444.5,-190 422.211,-181.5534 401.2034,-166.7064 383.4877,-151.2239"/>
-<polygon fill="none" stroke="#191970" points="4188.7004,-290.7269 4198.7254,-287.299 4188.7509,-283.7271 4188.7004,-290.7269"/>
+<path fill="none" stroke="#191970" d="M4074.855,-287.1364C3466.7329,-282.2495 618.2065,-255.9041 444.5,-190 422.2143,-181.5448 401.2076,-166.6954 383.4917,-151.2134"/>
+<polygon fill="none" stroke="#191970" points="4074.9141,-290.6369 4084.9419,-287.2172 4074.9702,-283.6371 4074.9141,-290.6369"/>
</g>
<!-- Node7 -->
<g id="node6" class="node">
@@ -192,8 +192,8 @@
<!-- Node2->Node7 -->
<g id="edge5" class="edge">
<title>Node2->Node7</title>
-<path fill="none" stroke="#191970" d="M4188.4532,-287.0364C3590.6547,-281.7186 839.5795,-253.8475 671.5,-190 649.2177,-181.5357 628.2119,-166.6838 610.4959,-151.2024"/>
-<polygon fill="none" stroke="#191970" points="4188.7673,-290.5392 4198.7979,-287.128 4188.8294,-283.5395 4188.7673,-290.5392"/>
+<path fill="none" stroke="#191970" d="M4074.6979,-286.9348C3488.8457,-281.1852 833.9689,-251.7975 671.5,-190 649.2214,-181.526 628.2167,-166.6714 610.5004,-151.1905"/>
+<polygon fill="none" stroke="#191970" points="4074.8065,-290.4359 4084.8403,-287.034 4074.875,-283.4363 4074.8065,-290.4359"/>
</g>
<!-- Node9 -->
<g id="node7" class="node">
@@ -216,8 +216,8 @@
<!-- Node2->Node9 -->
<g id="edge6" class="edge">
<title>Node2->Node9</title>
-<path fill="none" stroke="#191970" d="M4188.1871,-286.8196C3612.8541,-280.6119 1055.3434,-249.7408 898.5,-190 876.2253,-181.5157 855.2216,-166.6583 837.5051,-151.178"/>
-<polygon fill="none" stroke="#191970" points="4188.5241,-290.3233 4198.5612,-286.9312 4188.5994,-283.3237 4188.5241,-290.3233"/>
+<path fill="none" stroke="#191970" d="M4074.7994,-286.7081C3512.4877,-280.0536 1049.7762,-247.7066 898.5,-190 876.2296,-181.5046 855.227,-166.6442 837.5103,-151.1645"/>
+<polygon fill="none" stroke="#191970" points="4074.9023,-290.2094 4084.9429,-286.8278 4074.9849,-283.2099 4074.9023,-290.2094"/>
</g>
<!-- Node11 -->
<g id="node8" class="node">
@@ -240,8 +240,8 @@
<!-- Node2->Node11 -->
<g id="edge7" class="edge">
<title>Node2->Node11</title>
-<path fill="none" stroke="#191970" d="M4188.2463,-286.5755C3636.8206,-279.4341 1271.1595,-245.6523 1125.5,-190 1103.2341,-181.4928 1082.2328,-166.6292 1064.5157,-151.1502"/>
-<polygon fill="none" stroke="#191970" points="4188.5466,-290.0796 4198.591,-286.7091 4188.637,-283.0802 4188.5466,-290.0796"/>
+<path fill="none" stroke="#191970" d="M4074.8143,-286.447C3536.686,-278.8312 1265.5985,-243.6197 1125.5,-190 1103.239,-181.4801 1082.239,-166.6129 1064.5217,-151.1346"/>
+<polygon fill="none" stroke="#191970" points="4074.8657,-289.948 4084.9142,-286.5896 4074.9646,-282.9487 4074.8657,-289.948"/>
</g>
<!-- Node13 -->
<g id="node9" class="node">
@@ -264,8 +264,8 @@
<!-- Node2->Node13 -->
<g id="edge8" class="edge">
<title>Node2->Node13</title>
-<path fill="none" stroke="#191970" d="M4188.7363,-286.3008C3662.8292,-278.1811 1487.0364,-241.5854 1352.5,-190 1330.2442,-181.4664 1309.2456,-166.5956 1291.528,-151.1181"/>
-<polygon fill="none" stroke="#191970" points="4188.9336,-289.8042 4198.9863,-286.4587 4189.0414,-282.805 4188.9336,-289.8042"/>
+<path fill="none" stroke="#191970" d="M4074.9087,-286.1457C3562.01,-277.5089 1481.4532,-239.5433 1352.5,-190 1330.2499,-181.4516 1309.2529,-166.5767 1291.535,-151.1"/>
+<polygon fill="none" stroke="#191970" points="4074.8533,-289.6451 4084.9107,-286.3137 4074.9709,-282.6461 4074.8533,-289.6451"/>
</g>
<!-- Node15 -->
<g id="node10" class="node">
@@ -288,8 +288,8 @@
<!-- Node2->Node15 -->
<g id="edge9" class="edge">
<title>Node2->Node15</title>
-<path fill="none" stroke="#191970" d="M4188.5247,-285.9692C3687.8428,-276.7837 1702.88,-237.5035 1579.5,-190 1557.256,-181.4357 1536.2607,-166.5564 1518.5424,-151.0806"/>
-<polygon fill="none" stroke="#191970" points="4188.5888,-289.4709 4198.6512,-286.1546 4188.717,-282.472 4188.5888,-289.4709"/>
+<path fill="none" stroke="#191970" d="M4074.5009,-285.7824C3587.0395,-276.0357 1697.2958,-235.4597 1579.5,-190 1557.2627,-181.4182 1536.2693,-166.5342 1518.5506,-151.0594"/>
+<polygon fill="none" stroke="#191970" points="4074.6456,-289.2859 4084.7134,-285.9862 4074.7853,-282.2873 4074.6456,-289.2859"/>
</g>
<!-- Node17 -->
<g id="node11" class="node">
@@ -312,8 +312,8 @@
<!-- Node2->Node17 -->
<g id="edge10" class="edge">
<title>Node2->Node17</title>
-<path fill="none" stroke="#191970" d="M4188.3671,-285.578C3714.1348,-275.2502 1918.7602,-233.433 1806.5,-190 1784.27,-181.3993 1763.2785,-166.5101 1745.5595,-151.0364"/>
-<polygon fill="none" stroke="#191970" points="4188.5735,-289.0832 4198.6472,-285.8014 4188.7256,-282.0848 4188.5735,-289.0832"/>
+<path fill="none" stroke="#191970" d="M4074.6404,-287.6234C3629.2381,-285.565 2018.9836,-272.4382 1806.5,-190 1784.2781,-181.3785 1763.2888,-166.4836 1745.5693,-151.0111"/>
+<polygon fill="none" stroke="#191970" points="4074.9281,-291.1247 4084.9438,-287.67 4074.9598,-284.1247 4074.9281,-291.1247"/>
</g>
<!-- Node19 -->
<g id="node12" class="node">
@@ -336,8 +336,8 @@
<!-- Node2->Node19 -->
<g id="edge11" class="edge">
<title>Node2->Node19</title>
-<path fill="none" stroke="#191970" d="M4188.6511,-287.3705C3756.9312,-284.3996 2234.9935,-268.4115 2033.5,-190 2011.4064,-181.4022 1990.5258,-166.6147 1972.8661,-151.2329"/>
-<polygon fill="none" stroke="#191970" points="4188.9299,-290.8724 4198.9534,-287.4404 4188.9774,-283.8726 4188.9299,-290.8724"/>
+<path fill="none" stroke="#191970" d="M4074.4323,-287.0867C3656.569,-283.157 2223.8923,-264.3415 2033.5,-190 2011.4162,-181.377 1990.5383,-166.5825 1972.8781,-151.2021"/>
+<polygon fill="none" stroke="#191970" points="4074.6784,-290.589 4084.7105,-287.1822 4074.7436,-283.5893 4074.6784,-290.589"/>
</g>
<!-- Node21 -->
<g id="node13" class="node">
@@ -360,8 +360,8 @@
<!-- Node2->Node21 -->
<g id="edge12" class="edge">
<title>Node2->Node21</title>
-<path fill="none" stroke="#191970" d="M4188.6557,-286.7786C3785.4684,-281.8719 2439.9779,-260.3397 2260.5,-190 2238.427,-181.3493 2217.5522,-166.5471 2199.8914,-151.1682"/>
-<polygon fill="none" stroke="#191970" points="4188.8275,-290.2808 4198.869,-286.9018 4188.912,-283.2813 4188.8275,-290.2808"/>
+<path fill="none" stroke="#191970" d="M4074.7863,-286.4312C3686.6762,-280.4981 2428.9765,-256.3033 2260.5,-190 2238.4393,-181.3181 2217.5678,-166.5072 2199.9064,-151.13"/>
+<polygon fill="none" stroke="#191970" points="4074.8519,-289.9325 4084.9039,-286.5847 4074.9582,-282.9333 4074.8519,-289.9325"/>
</g>
<!-- Node23 -->
<g id="node14" class="node">
@@ -384,8 +384,8 @@
<!-- Node2->Node23 -->
<g id="edge13" class="edge">
<title>Node2->Node23</title>
-<path fill="none" stroke="#191970" d="M4188.529,-286.0375C3815.367,-279.0339 2645.043,-252.2877 2487.5,-190 2465.453,-181.2833 2444.5854,-166.4628 2426.9232,-151.0874"/>
-<polygon fill="none" stroke="#191970" points="4188.7373,-289.5419 4198.8009,-286.2291 4188.8679,-282.5431 4188.7373,-289.5419"/>
+<path fill="none" stroke="#191970" d="M4074.6718,-285.5955C3717.5269,-277.4762 2634.0996,-248.267 2487.5,-190 2465.4688,-181.2435 2444.6055,-166.412 2426.9425,-151.0387"/>
+<polygon fill="none" stroke="#191970" points="4074.6921,-289.0968 4084.7687,-285.8238 4074.8504,-282.0985 4074.6921,-289.0968"/>
</g>
<!-- Node25 -->
<g id="node15" class="node">
@@ -408,8 +408,8 @@
<!-- Node2->Node25 -->
<g id="edge14" class="edge">
<title>Node2->Node25</title>
-<path fill="none" stroke="#191970" d="M4188.6582,-285.0919C3847.7572,-275.811 2850.2631,-244.2814 2714.5,-190 2692.6057,-181.2461 2671.8539,-166.5148 2654.2512,-151.2329"/>
-<polygon fill="none" stroke="#191970" points="4188.7046,-288.5943 4198.7958,-285.3666 4188.8943,-281.5969 4188.7046,-288.5943"/>
+<path fill="none" stroke="#191970" d="M4074.501,-284.5014C3750.4629,-273.9868 2839.3491,-240.2622 2714.5,-190 2692.6266,-181.1941 2671.8807,-166.4482 2654.2768,-151.1689"/>
+<polygon fill="none" stroke="#191970" points="4074.7398,-288.0109 4084.8476,-284.8358 4074.9659,-281.0145 4074.7398,-288.0109"/>
</g>
<!-- Node27 -->
<g id="node16" class="node">
@@ -432,8 +432,8 @@
<!-- Node2->Node27 -->
<g id="edge15" class="edge">
<title>Node2->Node27</title>
-<path fill="none" stroke="#191970" d="M4188.7205,-283.8328C3882.3847,-272.0477 3055.6195,-236.3073 2941.5,-190 2919.6508,-181.1341 2898.9118,-166.3712 2881.3067,-151.0949"/>
-<polygon fill="none" stroke="#191970" points="4188.8287,-287.3394 4198.9554,-284.2251 4189.0968,-280.3445 4188.8287,-287.3394"/>
+<path fill="none" stroke="#191970" d="M4074.9031,-283.0363C3787.1192,-269.9011 3044.835,-232.3256 2941.5,-190 2919.68,-181.0626 2898.9491,-166.2797 2881.3426,-151.007"/>
+<polygon fill="none" stroke="#191970" points="4074.7981,-286.535 4084.9469,-283.4932 4075.1163,-279.5423 4074.7981,-286.535"/>
</g>
<!-- Node29 -->
<g id="node17" class="node">
@@ -456,8 +456,8 @@
<!-- Node2->Node29 -->
<g id="edge16" class="edge">
<title>Node2->Node29</title>
-<path fill="none" stroke="#191970" d="M4188.6528,-286.8216C3938.9313,-282.8477 3350.826,-265.5093 3168.5,-190 3146.8333,-181.0268 3126.2191,-166.3318 3108.673,-151.1515"/>
-<polygon fill="none" stroke="#191970" points="4188.7872,-290.324 4198.8402,-286.9793 4188.8956,-283.3249 4188.7872,-290.324"/>
+<path fill="none" stroke="#191970" d="M4074.7697,-285.538C3844.1366,-279.1763 3329.6252,-257.635 3168.5,-190 3146.8765,-180.9232 3126.2746,-166.1986 3108.7265,-151.0231"/>
+<polygon fill="none" stroke="#191970" points="4074.7741,-289.0393 4084.8653,-285.8118 4074.9639,-282.0419 4074.7741,-289.0393"/>
</g>
<!-- Node31 -->
<g id="node18" class="node">
@@ -480,8 +480,8 @@
<!-- Node2->Node31 -->
<g id="edge17" class="edge">
<title>Node2->Node31</title>
-<path fill="none" stroke="#191970" d="M4188.5598,-283.9428C3977.7802,-275.0888 3535.6892,-249.8259 3395.5,-190 3374.0485,-180.8456 3353.5692,-166.1953 3336.0816,-151.1132"/>
-<polygon fill="none" stroke="#191970" points="4188.5907,-287.447 4198.7272,-284.3647 4188.881,-280.4531 4188.5907,-287.447"/>
+<path fill="none" stroke="#191970" d="M4074.6817,-281.8976C3885.3787,-270.4305 3514.829,-242.0084 3395.5,-190 3374.2368,-180.7326 3353.8853,-166.1451 3336.4584,-151.1572"/>
+<polygon fill="none" stroke="#191970" points="4074.7521,-285.408 4084.9438,-282.5134 4075.1714,-278.4206 4074.7521,-285.408"/>
</g>
<!-- Node33 -->
<g id="node19" class="node">
@@ -501,8 +501,8 @@
<!-- Node2->Node33 -->
<g id="edge18" class="edge">
<title>Node2->Node33</title>
-<path fill="none" stroke="#191970" d="M4188.5301,-279.3011C4021.5749,-265.2677 3719.5893,-234.618 3620.5,-190 3591.9037,-177.1237 3564.9315,-154.6899 3544.6241,-134.8553"/>
-<polygon fill="none" stroke="#191970" points="4188.5798,-282.8173 4198.8362,-280.1606 4189.1617,-275.8415 4188.5798,-282.8173"/>
+<path fill="none" stroke="#191970" d="M4074.8593,-275.5721C3932.9671,-258.8223 3699.2943,-226.8213 3620.5,-190 3592.0877,-176.7227 3565.1226,-154.2732 3544.773,-134.5306"/>
+<polygon fill="none" stroke="#191970" points="4074.5738,-279.0625 4084.9135,-276.7516 4075.3895,-272.1102 4074.5738,-279.0625"/>
</g>
<!-- Node36 -->
<g id="node20" class="node">
@@ -524,8 +524,8 @@
<!-- Node2->Node36 -->
<g id="edge19" class="edge">
<title>Node2->Node36</title>
-<path fill="none" stroke="#191970" d="M4188.95,-278.9269C4093.744,-267.7929 3956.302,-243.3121 3847.5,-190 3824.5997,-178.779 3802.4303,-161.9293 3783.8832,-145.5215"/>
-<polygon fill="none" stroke="#191970" points="4188.5642,-282.4055 4198.896,-280.0595 4189.3562,-275.4504 4188.5642,-282.4055"/>
+<path fill="none" stroke="#191970" d="M4075.0184,-267.8766C4006.4154,-252.7313 3918.7482,-227.9537 3847.5,-190 3825.445,-178.2514 3803.7986,-161.6718 3785.4487,-145.6452"/>
+<polygon fill="none" stroke="#191970" points="4074.4828,-271.3417 4084.9976,-270.0417 4075.9671,-264.5009 4074.4828,-271.3417"/>
</g>
<!-- Node37 -->
<g id="node21" class="node">
@@ -547,8 +547,8 @@
<!-- Node2->Node37 -->
<g id="edge20" class="edge">
<title>Node2->Node37</title>
-<path fill="none" stroke="#191970" d="M4189.3646,-247.0551C4151.6434,-231.4188 4110.1686,-211.9921 4074.5,-190 4053.8836,-177.2886 4033.0899,-161.0166 4015.0592,-145.5449"/>
-<polygon fill="none" stroke="#191970" points="4188.2875,-250.3961 4198.868,-250.948 4190.941,-243.9185 4188.2875,-250.3961"/>
+<path fill="none" stroke="#191970" d="M4120.93,-230.5506C4089.2274,-203.5964 4051.7739,-171.7527 4021.1369,-145.7045"/>
+<polygon fill="none" stroke="#191970" points="4118.9727,-233.4805 4128.8584,-237.2915 4123.507,-228.1475 4118.9727,-233.4805"/>
</g>
<!-- Node38 -->
<g id="node22" class="node">
@@ -571,8 +571,8 @@
<!-- Node2->Node38 -->
<g id="edge21" class="edge">
<title>Node2->Node38</title>
-<path fill="none" stroke="#191970" d="M4267.2995,-228.4062C4252.7568,-203.7857 4235.9851,-175.3914 4221.6626,-151.1437"/>
-<polygon fill="none" stroke="#191970" points="4264.4484,-230.4614 4272.5478,-237.2915 4270.4756,-226.9013 4264.4484,-230.4614"/>
+<path fill="none" stroke="#191970" d="M4188.5,-227.107C4188.5,-202.7907 4188.5,-174.966 4188.5,-151.1437"/>
+<polygon fill="none" stroke="#191970" points="4185.0001,-227.2915 4188.5,-237.2915 4192.0001,-227.2916 4185.0001,-227.2915"/>
</g>
<!-- Node39 -->
<g id="node23" class="node">
@@ -595,8 +595,8 @@
<!-- Node2->Node39 -->
<g id="edge22" class="edge">
<title>Node2->Node39</title>
-<path fill="none" stroke="#191970" d="M4337.3917,-228.4062C4351.8068,-203.7857 4368.4314,-175.3914 4382.6283,-151.1437"/>
-<polygon fill="none" stroke="#191970" points="4334.2217,-226.8934 4332.1894,-237.2915 4340.2625,-230.4303 4334.2217,-226.8934"/>
+<path fill="none" stroke="#191970" d="M4255.9378,-230.663C4285.4947,-205.5331 4320.0707,-176.1359 4349.4658,-151.1437"/>
+<polygon fill="none" stroke="#191970" points="4253.493,-228.1475 4248.1416,-237.2915 4258.0273,-233.4805 4253.493,-228.1475"/>
</g>
<!-- Node40 -->
<g id="node24" class="node">
@@ -618,8 +618,8 @@
<!-- Node2->Node40 -->
<g id="edge23" class="edge">
<title>Node2->Node40</title>
-<path fill="none" stroke="#191970" d="M4415.3702,-246.5099C4452.5417,-230.9164 4493.3348,-211.6543 4528.5,-190 4549.1211,-177.3018 4569.9688,-161.1203 4588.0964,-145.7256"/>
-<polygon fill="none" stroke="#191970" points="4413.8998,-243.3306 4406.0034,-250.3945 4416.5814,-249.7966 4413.8998,-243.3306"/>
+<path fill="none" stroke="#191970" d="M4301.9626,-267.6298C4370.2872,-252.4174 4457.5069,-227.6559 4528.5,-190 4550.6906,-178.2298 4572.5143,-161.645 4591.0318,-145.6204"/>
+<polygon fill="none" stroke="#191970" points="4301.0435,-264.248 4292.0232,-269.8053 4302.5403,-271.0861 4301.0435,-264.248"/>
</g>
<!-- Node41 -->
<g id="node25" class="node">
@@ -642,8 +642,8 @@
<!-- Node2->Node41 -->
<g id="edge24" class="edge">
<title>Node2->Node41</title>
-<path fill="none" stroke="#191970" d="M4416.2387,-278.6272C4510.9622,-267.3246 4647.3483,-242.7638 4755.5,-190 4775.8758,-180.0593 4795.714,-165.7272 4812.9764,-151.1975"/>
-<polygon fill="none" stroke="#191970" points="4415.5829,-275.1797 4406.0537,-279.8106 4416.3908,-282.1329 4415.5829,-275.1797"/>
+<path fill="none" stroke="#191970" d="M4302.1902,-275.4146C4443.809,-258.5238 4676.7417,-226.4012 4755.5,-190 4776.4304,-180.3262 4796.6597,-165.805 4814.1199,-151.0081"/>
+<polygon fill="none" stroke="#191970" points="4301.6732,-271.9513 4292.1548,-276.6042 4302.4973,-278.9026 4301.6732,-271.9513"/>
</g>
<!-- Node42 -->
<g id="node26" class="node">
@@ -666,8 +666,8 @@
<!-- Node2->Node42 -->
<g id="edge25" class="edge">
<title>Node2->Node42</title>
-<path fill="none" stroke="#191970" d="M4416.4243,-279.1361C4582.8968,-264.9133 4883.5782,-234.0495 4982.5,-190 5003.6816,-180.5679 5024.0598,-166.0262 5041.5816,-151.1332"/>
-<polygon fill="none" stroke="#191970" points="4415.8159,-275.6751 4406.1474,-280.0076 4416.4074,-282.65 4415.8159,-275.6751"/>
+<path fill="none" stroke="#191970" d="M4302.1548,-281.8712C4491.5735,-270.3501 4862.8467,-241.8304 4982.5,-190 5003.8945,-180.7325 5024.4094,-166.145 5041.9928,-151.157"/>
+<polygon fill="none" stroke="#191970" points="4301.9356,-278.378 4292.1643,-282.4734 4302.3568,-285.3653 4301.9356,-278.378"/>
</g>
<!-- Node43 -->
<g id="node27" class="node">
@@ -690,8 +690,8 @@
<!-- Node2->Node43 -->
<g id="edge26" class="edge">
<title>Node2->Node43</title>
-<path fill="none" stroke="#191970" d="M4416.4163,-283.8867C4627.1556,-274.9379 5069.1814,-249.5218 5209.5,-190 5231.0828,-180.8448 5251.7264,-166.1943 5269.3713,-151.1123"/>
-<polygon fill="none" stroke="#191970" points="4416.0953,-280.397 4406.2508,-284.3132 4416.3888,-287.3908 4416.0953,-280.397"/>
+<path fill="none" stroke="#191970" d="M4302.3177,-285.4907C4533.1321,-279.0406 5048.0687,-257.3433 5209.5,-190 5231.256,-180.9242 5252.0234,-166.1999 5269.7295,-151.0243"/>
+<polygon fill="none" stroke="#191970" points="4302.1143,-281.9949 4292.2142,-285.7683 4302.3066,-288.9922 4302.1143,-281.9949"/>
</g>
<!-- Node44 -->
<g id="node28" class="node">
@@ -713,8 +713,8 @@
<!-- Node2->Node44 -->
<g id="edge27" class="edge">
<title>Node2->Node44</title>
-<path fill="none" stroke="#191970" d="M4416.3235,-286.7643C4665.9969,-282.6738 5254.0107,-265.1138 5436.5,-190 5460.9194,-179.9488 5484.0598,-162.7167 5503.0307,-145.6654"/>
-<polygon fill="none" stroke="#191970" points="4416.0809,-283.2676 4406.1381,-286.9267 4416.1926,-290.2667 4416.0809,-283.2676"/>
+<path fill="none" stroke="#191970" d="M4302.1773,-283.0138C4590.1664,-269.8249 5332.9897,-232.1317 5436.5,-190 5461.0779,-179.9961 5484.3327,-162.6635 5503.3503,-145.5191"/>
+<polygon fill="none" stroke="#191970" points="4301.9563,-279.5202 4292.1263,-283.4726 4302.2756,-286.5129 4301.9563,-279.5202"/>
</g>
<!-- Node45 -->
<g id="node29" class="node">
@@ -736,8 +736,8 @@
<!-- Node2->Node45 -->
<g id="edge28" class="edge">
<title>Node2->Node45</title>
-<path fill="none" stroke="#191970" d="M4416.2689,-283.8066C4722.5777,-271.9548 5549.2804,-236.0597 5663.5,-190 5688.1102,-180.0758 5711.3713,-162.7588 5730.3849,-145.6045"/>
-<polygon fill="none" stroke="#191970" points="4415.8926,-280.3184 4406.0349,-284.2011 4416.1623,-287.3132 4415.8926,-280.3184"/>
+<path fill="none" stroke="#191970" d="M4302.5656,-284.4781C4626.7949,-273.9004 5538.4567,-240.0223 5663.5,-190 5688.1376,-180.144 5711.404,-162.8403 5730.4143,-145.6776"/>
+<polygon fill="none" stroke="#191970" points="4302.094,-280.9915 4292.213,-284.8145 4302.3214,-287.9878 4302.094,-280.9915"/>
</g>
<!-- Node46 -->
<g id="node30" class="node">
@@ -761,66 +761,65 @@
<!-- Node2->Node46 -->
<g id="edge29" class="edge">
<title>Node2->Node46</title>
-<path fill="none" stroke="#191970" d="M4416.2604,-282.924C4768.5788,-266.9477 5822.1657,-217.15 5890.5,-190 5909.8934,-182.2948 5928.4273,-169.9548 5944.704,-156.6844"/>
-<polygon fill="none" stroke="#191970" points="4415.8743,-279.4378 4406.043,-283.3867 4416.1911,-286.4306 4415.8743,-279.4378"/>
+<path fill="none" stroke="#191970" d="M4302.3845,-285.5716C4659.7077,-277.3807 5743.6875,-247.9805 5890.5,-190 5909.9092,-182.3347 5928.4489,-170.0094 5944.7262,-156.7404"/>
+<polygon fill="none" stroke="#191970" points="4302.2002,-282.0748 4292.2826,-285.802 4302.3598,-289.073 4302.2002,-282.0748"/>
</g>
<!-- Node47 -->
<g id="node31" class="node">
<title>Node47</title>
-<g id="a_node31"><a xlink:href="classtvm_1_1tir_1_1LoadNode.html" target="_top" xlink:title="Load the value from buffer_var. ">
-<polygon fill="#ffffff" stroke="#000000" points="6127,-33.5 6127,-156.5 6336,-156.5 6336,-33.5 6127,-33.5"/>
-<text text-anchor="middle" x="6231.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::LoadNode</text>
-<polyline fill="none" stroke="#000000" points="6127,-137.5 6336,-137.5 "/>
-<text text-anchor="start" x="6135" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ buffer_var</text>
-<text text-anchor="start" x="6135" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ index</text>
-<text text-anchor="start" x="6135" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ predicate</text>
-<text text-anchor="start" x="6135" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="6127,-85.5 6336,-85.5 "/>
-<text text-anchor="start" x="6135" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="6135" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="6135" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="6135" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node31"><a xlink:href="classtvm_1_1tir_1_1NotNode.html" target="_top" xlink:title="!a ">
+<polygon fill="#ffffff" stroke="#000000" points="6127,-44.5 6127,-145.5 6336,-145.5 6336,-44.5 6127,-44.5"/>
+<text text-anchor="middle" x="6231.5" y="-133.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::NotNode</text>
+<polyline fill="none" stroke="#000000" points="6127,-126.5 6336,-126.5 "/>
+<text text-anchor="start" x="6135" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ a</text>
+<text text-anchor="start" x="6135" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="6127,-96.5 6336,-96.5 "/>
+<text text-anchor="start" x="6135" y="-84.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="6135" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="6135" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="6135" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node2->Node47 -->
<g id="edge30" class="edge">
<title>Node2->Node47</title>
-<path fill="none" stroke="#191970" d="M4416.4605,-286.0109C4789.5896,-278.9238 5959.8224,-251.946 6117.5,-190 6137.0428,-182.3223 6155.6965,-169.9047 6172.0466,-156.5432"/>
-<polygon fill="none" stroke="#191970" points="4416.1217,-282.5165 4406.1895,-286.2047 4416.2538,-289.5153 4416.1217,-282.5165"/>
+<path fill="none" stroke="#191970" d="M4302.2622,-286.4069C4690.5391,-280.3941 5948.7921,-255.9697 6117.5,-190 6142.3336,-180.2893 6165.7209,-162.9023 6184.7721,-145.6333"/>
+<polygon fill="none" stroke="#191970" points="4302.0852,-282.9091 4292.1402,-286.5625 4302.1929,-289.9082 4302.0852,-282.9091"/>
</g>
<!-- Node48 -->
<g id="node32" class="node">
<title>Node48</title>
-<g id="a_node32"><a xlink:href="classtvm_1_1tir_1_1NotNode.html" target="_top" xlink:title="!a ">
-<polygon fill="#ffffff" stroke="#000000" points="6354,-44.5 6354,-145.5 6563,-145.5 6563,-44.5 6354,-44.5"/>
-<text text-anchor="middle" x="6458.5" y="-133.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::NotNode</text>
-<polyline fill="none" stroke="#000000" points="6354,-126.5 6563,-126.5 "/>
-<text text-anchor="start" x="6362" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ a</text>
-<text text-anchor="start" x="6362" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="6354,-96.5 6563,-96.5 "/>
-<text text-anchor="start" x="6362" y="-84.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="6362" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="6362" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="6362" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node32"><a xlink:href="classtvm_1_1tir_1_1OrNode.html" target="_top" xlink:title="a || b ">
+<polygon fill="#ffffff" stroke="#000000" points="6354,-39 6354,-151 6563,-151 6563,-39 6354,-39"/>
+<text text-anchor="middle" x="6458.5" y="-139" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::OrNode</text>
+<polyline fill="none" stroke="#000000" points="6354,-132 6563,-132 "/>
+<text text-anchor="start" x="6362" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ a</text>
+<text text-anchor="start" x="6362" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ b</text>
+<text text-anchor="start" x="6362" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="6354,-91 6563,-91 "/>
+<text text-anchor="start" x="6362" y="-79" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="6362" y="-68" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="6362" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="6362" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node2->Node48 -->
<g id="edge31" class="edge">
<title>Node2->Node48</title>
-<path fill="none" stroke="#191970" d="M4416.3339,-286.7519C4819.4856,-281.754 6164.87,-259.9503 6344.5,-190 6369.3472,-180.3242 6392.7371,-162.9438 6411.7866,-145.6704"/>
-<polygon fill="none" stroke="#191970" points="4416.0776,-283.2546 4406.1214,-286.8773 4416.1637,-290.2541 4416.0776,-283.2546"/>
+<path fill="none" stroke="#191970" d="M4302.6101,-287.0619C4720.6303,-283.0449 6153.858,-263.9607 6344.5,-190 6366.7174,-181.3806 6387.7628,-166.5871 6405.5826,-151.2065"/>
+<polygon fill="none" stroke="#191970" points="4302.2943,-283.5646 4292.328,-287.1596 4302.3609,-290.5643 4302.2943,-283.5646"/>
</g>
<!-- Node49 -->
<g id="node33" class="node">
<title>Node49</title>
-<g id="a_node33"><a xlink:href="classtvm_1_1tir_1_1OrNode.html" target="_top" xlink:title="a || b ">
+<g id="a_node33"><a xlink:href="classtvm_1_1tir_1_1ProducerLoadNode.html" target="_top" xlink:title="Load value from the result produced by the producer. ">
<polygon fill="#ffffff" stroke="#000000" points="6581,-39 6581,-151 6790,-151 6790,-39 6581,-39"/>
-<text text-anchor="middle" x="6685.5" y="-139" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::OrNode</text>
+<text text-anchor="middle" x="6685.5" y="-139" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::ProducerLoadNode</text>
<polyline fill="none" stroke="#000000" points="6581,-132 6790,-132 "/>
-<text text-anchor="start" x="6589" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ a</text>
-<text text-anchor="start" x="6589" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ b</text>
+<text text-anchor="start" x="6589" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ producer</text>
+<text text-anchor="start" x="6589" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ indices</text>
<text text-anchor="start" x="6589" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
<polyline fill="none" stroke="#000000" points="6581,-91 6790,-91 "/>
<text text-anchor="start" x="6589" y="-79" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
@@ -833,328 +832,329 @@
<!-- Node2->Node49 -->
<g id="edge32" class="edge">
<title>Node2->Node49</title>
-<path fill="none" stroke="#191970" d="M4416.3385,-287.3437C4848.0202,-284.2744 6369.837,-267.9744 6571.5,-190 6593.7271,-181.4057 6614.7752,-166.6192 6632.5945,-151.2372"/>
-<polygon fill="none" stroke="#191970" points="4416.0122,-283.8458 4406.037,-287.4159 4416.0614,-290.8456 4416.0122,-283.8458"/>
+<path fill="none" stroke="#191970" d="M4302.4551,-287.6273C4748.2312,-285.5833 6359.8369,-272.5036 6572.5,-190 6594.722,-181.3789 6615.7114,-166.4841 6633.4309,-151.0116"/>
+<polygon fill="none" stroke="#191970" points="4302.1272,-284.1286 4292.1431,-287.6735 4302.1587,-291.1285 4302.1272,-284.1286"/>
</g>
<!-- Node50 -->
<g id="node34" class="node">
<title>Node50</title>
-<g id="a_node34"><a xlink:href="classtvm_1_1tir_1_1ProducerLoadNode.html" target="_top" xlink:title="Load value from the result produced by the producer. ">
-<polygon fill="#ffffff" stroke="#000000" points="6808,-39 6808,-151 7017,-151 7017,-39 6808,-39"/>
-<text text-anchor="middle" x="6912.5" y="-139" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::ProducerLoadNode</text>
-<polyline fill="none" stroke="#000000" points="6808,-132 7017,-132 "/>
-<text text-anchor="start" x="6816" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ producer</text>
-<text text-anchor="start" x="6816" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ indices</text>
-<text text-anchor="start" x="6816" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="6808,-91 7017,-91 "/>
-<text text-anchor="start" x="6816" y="-79" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="6816" y="-68" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="6816" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="6816" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node34"><a xlink:href="classtvm_1_1tir_1_1RampNode.html" target="_top" xlink:title="Construct a vector with lanes elements where its i-th element equals base + i * stride. This is useful to construct a index for a continuous vector load. ">
+<polygon fill="#ffffff" stroke="#000000" points="6808,-33.5 6808,-156.5 7017,-156.5 7017,-33.5 6808,-33.5"/>
+<text text-anchor="middle" x="6912.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::RampNode</text>
+<polyline fill="none" stroke="#000000" points="6808,-137.5 7017,-137.5 "/>
+<text text-anchor="start" x="6816" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ base</text>
+<text text-anchor="start" x="6816" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ stride</text>
+<text text-anchor="start" x="6816" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ lanes</text>
+<text text-anchor="start" x="6816" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="6808,-85.5 7017,-85.5 "/>
+<text text-anchor="start" x="6816" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="6816" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="6816" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="6816" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node2->Node50 -->
<g id="edge33" class="edge">
<title>Node2->Node50</title>
-<path fill="none" stroke="#191970" d="M4416.2812,-285.5875C4890.097,-275.278 6687.1632,-233.4616 6799.5,-190 6821.73,-181.3995 6842.7215,-166.5103 6860.4406,-151.0367"/>
-<polygon fill="none" stroke="#191970" points="4415.9323,-282.0941 4406.0107,-285.8105 4416.0844,-289.0925 4415.9323,-282.0941"/>
+<path fill="none" stroke="#191970" d="M4302.5865,-285.784C4790.4214,-276.0443 6681.6134,-235.4928 6799.5,-190 6819.1072,-182.4335 6837.7473,-169.9673 6854.0282,-156.5166"/>
+<polygon fill="none" stroke="#191970" points="4302.2944,-282.2891 4292.3661,-285.9877 4302.4339,-289.2877 4302.2944,-282.2891"/>
</g>
<!-- Node51 -->
<g id="node35" class="node">
<title>Node51</title>
-<g id="a_node35"><a xlink:href="classtvm_1_1tir_1_1RampNode.html" target="_top" xlink:title="Construct a vector with lanes elements where its i-th element equals base + i * stride. This is useful to construct a index for a continuous vector load. ">
-<polygon fill="#ffffff" stroke="#000000" points="7035,-33.5 7035,-156.5 7244,-156.5 7244,-33.5 7035,-33.5"/>
-<text text-anchor="middle" x="7139.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::RampNode</text>
-<polyline fill="none" stroke="#000000" points="7035,-137.5 7244,-137.5 "/>
-<text text-anchor="start" x="7043" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ base</text>
-<text text-anchor="start" x="7043" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ stride</text>
-<text text-anchor="start" x="7043" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ lanes</text>
-<text text-anchor="start" x="7043" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="7035,-85.5 7244,-85.5 "/>
-<text text-anchor="start" x="7043" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="7043" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="7043" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="7043" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node35"><a xlink:href="classtvm_1_1tir_1_1ReduceNode.html" target="_top" xlink:title="Reduction operator operator. ">
+<polygon fill="#ffffff" stroke="#000000" points="7035,-17 7035,-173 7244,-173 7244,-17 7035,-17"/>
+<text text-anchor="middle" x="7139.5" y="-161" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::ReduceNode</text>
+<polyline fill="none" stroke="#000000" points="7035,-154 7244,-154 "/>
+<text text-anchor="start" x="7043" y="-142" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ combiner</text>
+<text text-anchor="start" x="7043" y="-131" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ source</text>
+<text text-anchor="start" x="7043" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ init</text>
+<text text-anchor="start" x="7043" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ axis</text>
+<text text-anchor="start" x="7043" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ condition</text>
+<text text-anchor="start" x="7043" y="-87" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ value_index</text>
+<text text-anchor="start" x="7043" y="-76" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="7035,-69 7244,-69 "/>
+<text text-anchor="start" x="7043" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="7043" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="7043" y="-35" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="7043" y="-24" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node2->Node51 -->
<g id="edge34" class="edge">
<title>Node2->Node51</title>
-<path fill="none" stroke="#191970" d="M4416.5172,-285.97C4917.3829,-276.7879 6903.0744,-237.5201 7026.5,-190 7046.113,-182.4488 7064.7553,-169.988 7081.0364,-156.5377"/>
-<polygon fill="none" stroke="#191970" points="4416.3212,-282.4729 4406.3869,-286.1553 4416.4493,-289.4717 4416.3212,-282.4729"/>
+<path fill="none" stroke="#191970" d="M4302.1714,-286.1472C4815.4317,-277.5171 6897.4554,-239.5766 7026.5,-190 7037.6251,-185.7259 7048.4369,-179.8702 7058.6822,-173.1484"/>
+<polygon fill="none" stroke="#191970" points="4302.1022,-282.6478 4292.1624,-286.3151 4302.2197,-289.6468 4302.1022,-282.6478"/>
</g>
<!-- Node52 -->
<g id="node36" class="node">
<title>Node52</title>
-<g id="a_node36"><a xlink:href="classtvm_1_1tir_1_1ReduceNode.html" target="_top" xlink:title="Reduction operator operator. ">
-<polygon fill="#ffffff" stroke="#000000" points="7262,-17 7262,-173 7471,-173 7471,-17 7262,-17"/>
-<text text-anchor="middle" x="7366.5" y="-161" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::ReduceNode</text>
-<polyline fill="none" stroke="#000000" points="7262,-154 7471,-154 "/>
-<text text-anchor="start" x="7270" y="-142" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ combiner</text>
-<text text-anchor="start" x="7270" y="-131" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ source</text>
-<text text-anchor="start" x="7270" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ init</text>
-<text text-anchor="start" x="7270" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ axis</text>
-<text text-anchor="start" x="7270" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ condition</text>
-<text text-anchor="start" x="7270" y="-87" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ value_index</text>
-<text text-anchor="start" x="7270" y="-76" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="7262,-69 7471,-69 "/>
-<text text-anchor="start" x="7270" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="7270" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="7270" y="-35" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="7270" y="-24" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node36"><a xlink:href="classtvm_1_1tir_1_1SelectNode.html" target="_top" xlink:title="return true_value if condition is true, otherwise return false_value. ">
+<polygon fill="#ffffff" stroke="#000000" points="7262,-33.5 7262,-156.5 7471,-156.5 7471,-33.5 7262,-33.5"/>
+<text text-anchor="middle" x="7366.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::SelectNode</text>
+<polyline fill="none" stroke="#000000" points="7262,-137.5 7471,-137.5 "/>
+<text text-anchor="start" x="7270" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ condition</text>
+<text text-anchor="start" x="7270" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ true_value</text>
+<text text-anchor="start" x="7270" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ false_value</text>
+<text text-anchor="start" x="7270" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="7262,-85.5 7471,-85.5 "/>
+<text text-anchor="start" x="7270" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="7270" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="7270" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="7270" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node2->Node52 -->
<g id="edge35" class="edge">
<title>Node2->Node52</title>
-<path fill="none" stroke="#191970" d="M4416.3023,-286.3015C4942.3876,-278.1852 7118.9178,-241.6021 7253.5,-190 7264.6279,-185.7333 7275.4415,-179.8822 7285.6879,-173.1631"/>
-<polygon fill="none" stroke="#191970" points="4415.9937,-282.8058 4406.0488,-286.4593 4416.1015,-289.8049 4415.9937,-282.8058"/>
+<path fill="none" stroke="#191970" d="M4302.2599,-286.4484C4840.7395,-278.8391 7113.3095,-243.6532 7253.5,-190 7273.1281,-182.488 7291.7757,-170.0414 7308.0572,-156.5921"/>
+<polygon fill="none" stroke="#191970" points="4302.1031,-282.9502 4292.1535,-286.5909 4302.2018,-289.9495 4302.1031,-282.9502"/>
</g>
<!-- Node53 -->
<g id="node37" class="node">
<title>Node53</title>
-<g id="a_node37"><a xlink:href="classtvm_1_1tir_1_1SelectNode.html" target="_top" xlink:title="return true_value if condition is true, otherwise return false_value. ">
-<polygon fill="#ffffff" stroke="#000000" points="7489,-33.5 7489,-156.5 7698,-156.5 7698,-33.5 7489,-33.5"/>
-<text text-anchor="middle" x="7593.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::SelectNode</text>
-<polyline fill="none" stroke="#000000" points="7489,-137.5 7698,-137.5 "/>
-<text text-anchor="start" x="7497" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ condition</text>
-<text text-anchor="start" x="7497" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ true_value</text>
-<text text-anchor="start" x="7497" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ false_value</text>
-<text text-anchor="start" x="7497" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="7489,-85.5 7698,-85.5 "/>
-<text text-anchor="start" x="7497" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="7497" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="7497" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="7497" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node37"><a xlink:href="classtvm_1_1tir_1_1ShuffleNode.html" target="_top" xlink:title="Shuffle instruction. vec = concat(vectors) result = (vec[indices[0]], vec[indices[1]] ...">
+<polygon fill="#ffffff" stroke="#000000" points="7489,-39 7489,-151 7698,-151 7698,-39 7489,-39"/>
+<text text-anchor="middle" x="7593.5" y="-139" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::ShuffleNode</text>
+<polyline fill="none" stroke="#000000" points="7489,-132 7698,-132 "/>
+<text text-anchor="start" x="7497" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ vectors</text>
+<text text-anchor="start" x="7497" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ indices</text>
+<text text-anchor="start" x="7497" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="7489,-91 7698,-91 "/>
+<text text-anchor="start" x="7497" y="-79" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="7497" y="-68" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="7497" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="7497" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node2->Node53 -->
<g id="edge36" class="edge">
<title>Node2->Node53</title>
-<path fill="none" stroke="#191970" d="M4416.3355,-286.5821C4967.0885,-279.4566 7334.7544,-245.6844 7480.5,-190 7500.1324,-182.4991 7518.7815,-170.0565 7535.0631,-156.6075"/>
-<polygon fill="none" stroke="#191970" points="4415.958,-283.0865 4406.004,-286.7154 4416.0484,-290.086 4415.958,-283.0865"/>
+<path fill="none" stroke="#191970" d="M4302.2697,-286.7094C4864.9231,-280.0614 7329.1314,-247.7403 7480.5,-190 7502.7705,-181.5048 7523.7731,-166.6445 7541.4898,-151.1647"/>
+<polygon fill="none" stroke="#191970" points="4302.0781,-283.2113 4292.12,-286.829 4302.1606,-290.2108 4302.0781,-283.2113"/>
</g>
<!-- Node54 -->
<g id="node38" class="node">
<title>Node54</title>
-<g id="a_node38"><a xlink:href="classtvm_1_1tir_1_1ShuffleNode.html" target="_top" xlink:title="Shuffle instruction. vec = concat(vectors) result = (vec[indices[0]], vec[indices[1]] ...">
-<polygon fill="#ffffff" stroke="#000000" points="7716,-39 7716,-151 7925,-151 7925,-39 7716,-39"/>
-<text text-anchor="middle" x="7820.5" y="-139" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::ShuffleNode</text>
-<polyline fill="none" stroke="#000000" points="7716,-132 7925,-132 "/>
-<text text-anchor="start" x="7724" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ vectors</text>
-<text text-anchor="start" x="7724" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ indices</text>
-<text text-anchor="start" x="7724" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="7716,-91 7925,-91 "/>
-<text text-anchor="start" x="7724" y="-79" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="7724" y="-68" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="7724" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="7724" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node38"><a xlink:href="classtvm_1_1tir_1_1StringImmNode.html" target="_top" xlink:title="String constants, only used in asserts. ">
+<polygon fill="#ffffff" stroke="#000000" points="7716,-44.5 7716,-145.5 7925,-145.5 7925,-44.5 7716,-44.5"/>
+<text text-anchor="middle" x="7820.5" y="-133.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::StringImmNode</text>
+<polyline fill="none" stroke="#000000" points="7716,-126.5 7925,-126.5 "/>
+<text text-anchor="start" x="7724" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ value</text>
+<text text-anchor="start" x="7724" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="7716,-96.5 7925,-96.5 "/>
+<text text-anchor="start" x="7724" y="-84.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="7724" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="7724" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="7724" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node2->Node54 -->
<g id="edge37" class="edge">
<title>Node2->Node54</title>
-<path fill="none" stroke="#191970" d="M4416.3747,-286.8253C4990.9488,-280.6325 7550.5674,-249.774 7707.5,-190 7729.7747,-181.5158 7750.7784,-166.6584 7768.4949,-151.1781"/>
-<polygon fill="none" stroke="#191970" points="4415.9767,-283.3293 4406.0149,-286.9366 4416.052,-290.3289 4415.9767,-283.3293"/>
+<path fill="none" stroke="#191970" d="M4302.3668,-286.936C4888.5521,-281.1927 7544.9383,-251.8313 7707.5,-190 7732.4137,-180.5239 7755.7345,-163.0684 7774.6448,-145.6807"/>
+<polygon fill="none" stroke="#191970" points="4302.184,-283.4376 4292.2187,-287.0351 4302.2524,-290.4372 4302.184,-283.4376"/>
</g>
<!-- Node55 -->
<g id="node39" class="node">
<title>Node55</title>
-<g id="a_node39"><a xlink:href="classtvm_1_1tir_1_1StringImmNode.html" target="_top" xlink:title="String constants, only used in asserts. ">
-<polygon fill="#ffffff" stroke="#000000" points="7943,-44.5 7943,-145.5 8152,-145.5 8152,-44.5 7943,-44.5"/>
-<text text-anchor="middle" x="8047.5" y="-133.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::StringImmNode</text>
-<polyline fill="none" stroke="#000000" points="7943,-126.5 8152,-126.5 "/>
-<text text-anchor="start" x="7951" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ value</text>
+<g id="a_node39"><a xlink:href="classtvm_1_1tir_1_1VarNode.html" target="_top" xlink:title="A variable node in the IR. ">
+<polygon fill="#ffffff" stroke="#ff0000" points="7943,-33.5 7943,-156.5 8150,-156.5 8150,-33.5 7943,-33.5"/>
+<text text-anchor="middle" x="8046.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::VarNode</text>
+<polyline fill="none" stroke="#ff0000" points="7943,-137.5 8150,-137.5 "/>
+<text text-anchor="start" x="7951" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ name_hint</text>
+<text text-anchor="start" x="7951" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ type_annotation</text>
<text text-anchor="start" x="7951" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="7943,-96.5 8152,-96.5 "/>
-<text text-anchor="start" x="7951" y="-84.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="7951" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="7951" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="7951" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<text text-anchor="start" x="7951" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
+<polyline fill="none" stroke="#ff0000" points="7943,-85.5 8150,-85.5 "/>
+<text text-anchor="start" x="7951" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="7951" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="7951" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="7951" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
</a>
</g>
</g>
<!-- Node2->Node55 -->
<g id="edge38" class="edge">
<title>Node2->Node55</title>
-<path fill="none" stroke="#191970" d="M4416.5782,-287.037C5014.5413,-281.7223 7766.374,-253.8644 7934.5,-190 7959.4178,-180.5347 7982.7394,-163.0812 8001.6491,-145.6921"/>
-<polygon fill="none" stroke="#191970" points="4416.1992,-283.5401 4406.2306,-287.1286 4416.2612,-290.5398 4416.1992,-283.5401"/>
+<path fill="none" stroke="#191970" d="M4302.2107,-287.1507C4910.6836,-282.3383 7760.8452,-256.3187 7934.5,-190 7954.0318,-182.5408 7972.5423,-170.1131 7988.6817,-156.6653"/>
+<polygon fill="none" stroke="#191970" points="4302.0901,-283.6514 4292.118,-287.2302 4302.1453,-290.6512 4302.0901,-283.6514"/>
</g>
-<!-- Node56 -->
-<g id="node40" class="node">
-<title>Node56</title>
-<g id="a_node40"><a xlink:href="classtvm_1_1tir_1_1VarNode.html" target="_top" xlink:title="A variable node in the IR. ">
-<polygon fill="#ffffff" stroke="#ff0000" points="8170,-33.5 8170,-156.5 8377,-156.5 8377,-33.5 8170,-33.5"/>
-<text text-anchor="middle" x="8273.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::VarNode</text>
-<polyline fill="none" stroke="#ff0000" points="8170,-137.5 8377,-137.5 "/>
-<text text-anchor="start" x="8178" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ name_hint</text>
-<text text-anchor="start" x="8178" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ type_annotation</text>
-<text text-anchor="start" x="8178" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<text text-anchor="start" x="8178" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
-<polyline fill="none" stroke="#ff0000" points="8170,-85.5 8377,-85.5 "/>
-<text text-anchor="start" x="8178" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="8178" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="8178" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="8178" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
+<!-- Node58 -->
+<g id="node41" class="node">
+<title>Node58</title>
+<g id="a_node41"><a xlink:href="classtvm_1_1BaseFuncNode.html" target="_top" xlink:title="Base node of all functions. ">
+<polygon fill="#ffffff" stroke="#ff0000" points="8168,-39 8168,-151 8375,-151 8375,-39 8168,-39"/>
+<text text-anchor="middle" x="8271.5" y="-139" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::BaseFuncNode</text>
+<polyline fill="none" stroke="#ff0000" points="8168,-132 8375,-132 "/>
+<text text-anchor="start" x="8176" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ attrs</text>
+<text text-anchor="start" x="8176" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<text text-anchor="start" x="8176" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
+<polyline fill="none" stroke="#ff0000" points="8168,-91 8375,-91 "/>
+<text text-anchor="start" x="8176" y="-79" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ GetAttr()</text>
+<text text-anchor="start" x="8176" y="-68" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ GetAttr()</text>
+<text text-anchor="start" x="8176" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ HasNonzeroAttr()</text>
+<text text-anchor="start" x="8176" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
</a>
</g>
</g>
-<!-- Node2->Node56 -->
-<g id="edge39" class="edge">
-<title>Node2->Node56</title>
-<path fill="none" stroke="#191970" d="M4416.5909,-287.2388C5036.877,-282.8316 7982.2924,-258.3607 8161.5,-190 8181.0347,-182.5483 8199.5462,-170.1233 8215.6857,-156.6756"/>
-<polygon fill="none" stroke="#191970" points="4416.281,-283.7408 4406.3061,-287.3116 4416.3306,-290.7406 4416.281,-283.7408"/>
+<!-- Node57->Node58 -->
+<g id="edge40" class="edge">
+<title>Node57->Node58</title>
+<path fill="none" stroke="#191970" d="M8950.5515,-279.1905C8784.0476,-265.038 8483.326,-234.2638 8384.5,-190 8363.4483,-180.571 8343.2322,-166.0301 8325.8661,-151.1371"/>
+<polygon fill="none" stroke="#191970" points="8950.5714,-282.7045 8960.8302,-280.0576 8951.1598,-275.7293 8950.5714,-282.7045"/>
</g>
-<!-- Node59 -->
+<!-- Node61 -->
<g id="node42" class="node">
-<title>Node59</title>
-<g id="a_node42"><a xlink:href="classtvm_1_1BaseFuncNode.html" target="_top" xlink:title="Base node of all functions. ">
-<polygon fill="#ffffff" stroke="#ff0000" points="8395,-39 8395,-151 8602,-151 8602,-39 8395,-39"/>
-<text text-anchor="middle" x="8498.5" y="-139" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::BaseFuncNode</text>
-<polyline fill="none" stroke="#ff0000" points="8395,-132 8602,-132 "/>
-<text text-anchor="start" x="8403" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ attrs</text>
-<text text-anchor="start" x="8403" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<text text-anchor="start" x="8403" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_child_slots</text>
-<polyline fill="none" stroke="#ff0000" points="8395,-91 8602,-91 "/>
-<text text-anchor="start" x="8403" y="-79" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ GetAttr()</text>
-<text text-anchor="start" x="8403" y="-68" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ GetAttr()</text>
-<text text-anchor="start" x="8403" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ HasNonzeroAttr()</text>
-<text text-anchor="start" x="8403" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_BASE_OBJECT_INFO()</text>
+<title>Node61</title>
+<g id="a_node42"><a xlink:href="classtvm_1_1ConstructorNode.html" target="_top" xlink:title="ADT constructor. Constructors compare by pointer equality. ">
+<polygon fill="#ffffff" stroke="#000000" points="8393,-22.5 8393,-167.5 8602,-167.5 8602,-22.5 8393,-22.5"/>
+<text text-anchor="middle" x="8497.5" y="-155.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::ConstructorNode</text>
+<polyline fill="none" stroke="#000000" points="8393,-148.5 8602,-148.5 "/>
+<text text-anchor="start" x="8401" y="-136.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ name_hint</text>
+<text text-anchor="start" x="8401" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ inputs</text>
+<text text-anchor="start" x="8401" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ belong_to</text>
+<text text-anchor="start" x="8401" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ tag</text>
+<text text-anchor="start" x="8401" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="8393,-85.5 8602,-85.5 "/>
+<text text-anchor="start" x="8401" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ConstructorNode()</text>
+<text text-anchor="start" x="8401" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="8401" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="8401" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="8401" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
-<!-- Node58->Node59 -->
+<!-- Node57->Node61 -->
<g id="edge41" class="edge">
-<title>Node58->Node59</title>
-<path fill="none" stroke="#191970" d="M9064.7853,-275.4675C8923.1402,-258.6336 8690.1792,-226.5719 8611.5,-190 8590.6986,-180.3311 8570.6304,-165.8113 8553.3251,-151.0143"/>
-<polygon fill="none" stroke="#191970" points="9064.481,-278.9558 9074.8226,-276.6531 9065.3022,-272.0041 9064.481,-278.9558"/>
+<title>Node57->Node61</title>
+<path fill="none" stroke="#191970" d="M8950.7613,-278.6272C8856.0378,-267.3246 8719.6517,-242.7638 8611.5,-190 8598.9966,-183.9 8586.6957,-176.1465 8575.0682,-167.7084"/>
+<polygon fill="none" stroke="#191970" points="8950.6092,-282.1329 8960.9463,-279.8106 8951.4171,-275.1797 8950.6092,-282.1329"/>
</g>
<!-- Node62 -->
<g id="node43" class="node">
<title>Node62</title>
-<g id="a_node43"><a xlink:href="classtvm_1_1ConstructorNode.html" target="_top" xlink:title="ADT constructor. Constructors compare by pointer equality. ">
-<polygon fill="#ffffff" stroke="#000000" points="8620,-22.5 8620,-167.5 8829,-167.5 8829,-22.5 8620,-22.5"/>
-<text text-anchor="middle" x="8724.5" y="-155.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::ConstructorNode</text>
-<polyline fill="none" stroke="#000000" points="8620,-148.5 8829,-148.5 "/>
-<text text-anchor="start" x="8628" y="-136.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ name_hint</text>
-<text text-anchor="start" x="8628" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ inputs</text>
-<text text-anchor="start" x="8628" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ belong_to</text>
-<text text-anchor="start" x="8628" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ tag</text>
-<text text-anchor="start" x="8628" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="8620,-85.5 8829,-85.5 "/>
-<text text-anchor="start" x="8628" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ConstructorNode()</text>
-<text text-anchor="start" x="8628" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="8628" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="8628" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="8628" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node43"><a xlink:href="classtvm_1_1GlobalVarNode.html" target="_top" xlink:title="Global variable that lives in the top-level module. ">
+<polygon fill="#ffffff" stroke="#000000" points="8620,-44.5 8620,-145.5 8829,-145.5 8829,-44.5 8620,-44.5"/>
+<text text-anchor="middle" x="8724.5" y="-133.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::GlobalVarNode</text>
+<polyline fill="none" stroke="#000000" points="8620,-126.5 8829,-126.5 "/>
+<text text-anchor="start" x="8628" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ name_hint</text>
+<text text-anchor="start" x="8628" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="8620,-96.5 8829,-96.5 "/>
+<text text-anchor="start" x="8628" y="-84.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="8628" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="8628" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="8628" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
-<!-- Node58->Node62 -->
+<!-- Node57->Node62 -->
<g id="edge42" class="edge">
-<title>Node58->Node62</title>
-<path fill="none" stroke="#191970" d="M9065.0374,-267.6298C8996.7128,-252.4174 8909.4931,-227.6559 8838.5,-190 8826.5512,-183.6622 8814.7088,-175.9284 8803.4317,-167.6379"/>
-<polygon fill="none" stroke="#191970" points="9064.4597,-271.0861 9074.9768,-269.8053 9065.9565,-264.248 9064.4597,-271.0861"/>
+<title>Node57->Node62</title>
+<path fill="none" stroke="#191970" d="M8951.6298,-246.5099C8914.4583,-230.9164 8873.6652,-211.6543 8838.5,-190 8817.8789,-177.3018 8797.0312,-161.1203 8778.9036,-145.7256"/>
+<polygon fill="none" stroke="#191970" points="8950.4186,-249.7966 8960.9966,-250.3945 8953.1002,-243.3306 8950.4186,-249.7966"/>
</g>
<!-- Node63 -->
<g id="node44" class="node">
<title>Node63</title>
-<g id="a_node44"><a xlink:href="classtvm_1_1GlobalVarNode.html" target="_top" xlink:title="Global variable that lives in the top-level module. ">
-<polygon fill="#ffffff" stroke="#000000" points="8847,-44.5 8847,-145.5 9056,-145.5 9056,-44.5 8847,-44.5"/>
-<text text-anchor="middle" x="8951.5" y="-133.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::GlobalVarNode</text>
-<polyline fill="none" stroke="#000000" points="8847,-126.5 9056,-126.5 "/>
-<text text-anchor="start" x="8855" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ name_hint</text>
-<text text-anchor="start" x="8855" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="8847,-96.5 9056,-96.5 "/>
-<text text-anchor="start" x="8855" y="-84.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="8855" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="8855" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="8855" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node44"><a xlink:href="classtvm_1_1OpNode.html" target="_top" xlink:title="Primitive Op(builtin intrinsics) ">
+<polygon fill="#ffffff" stroke="#000000" points="8847,-.5 8847,-189.5 9056,-189.5 9056,-.5 8847,-.5"/>
+<text text-anchor="middle" x="8951.5" y="-177.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::OpNode</text>
+<polyline fill="none" stroke="#000000" points="8847,-170.5 9056,-170.5 "/>
+<text text-anchor="start" x="8855" y="-158.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ name</text>
+<text text-anchor="start" x="8855" y="-147.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ op_type</text>
+<text text-anchor="start" x="8855" y="-136.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ description</text>
+<text text-anchor="start" x="8855" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ arguments</text>
+<text text-anchor="start" x="8855" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ attrs_type_key</text>
+<text text-anchor="start" x="8855" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ attrs_type_index</text>
+<text text-anchor="start" x="8855" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ num_inputs</text>
+<text text-anchor="start" x="8855" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ support_level</text>
+<text text-anchor="start" x="8855" y="-70.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="8847,-63.5 9056,-63.5 "/>
+<text text-anchor="start" x="8855" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="8855" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="8855" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="8855" y="-18.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ IsPrimitiveOp()</text>
+<text text-anchor="start" x="8855" y="-7.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
-<!-- Node58->Node63 -->
+<!-- Node57->Node63 -->
<g id="edge43" class="edge">
-<title>Node58->Node63</title>
-<path fill="none" stroke="#191970" d="M9098.4288,-219.9219C9069.6918,-195.489 9037.8156,-168.3872 9011.1184,-145.6888"/>
-<polygon fill="none" stroke="#191970" points="9096.1659,-222.592 9106.0517,-226.403 9100.7002,-217.259 9096.1659,-222.592"/>
+<title>Node57->Node63</title>
+<path fill="none" stroke="#191970" d="M9023.2213,-217.4974C9017.9356,-208.4697 9012.4578,-199.1138 9006.9927,-189.7795"/>
+<polygon fill="none" stroke="#191970" points="9020.3624,-219.5417 9028.4354,-226.403 9026.4031,-216.0048 9020.3624,-219.5417"/>
</g>
<!-- Node64 -->
<g id="node45" class="node">
<title>Node64</title>
-<g id="a_node45"><a xlink:href="classtvm_1_1OpNode.html" target="_top" xlink:title="Primitive Op(builtin intrinsics) ">
-<polygon fill="#ffffff" stroke="#000000" points="9074,-.5 9074,-189.5 9283,-189.5 9283,-.5 9074,-.5"/>
-<text text-anchor="middle" x="9178.5" y="-177.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::OpNode</text>
-<polyline fill="none" stroke="#000000" points="9074,-170.5 9283,-170.5 "/>
-<text text-anchor="start" x="9082" y="-158.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ name</text>
-<text text-anchor="start" x="9082" y="-147.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ op_type</text>
-<text text-anchor="start" x="9082" y="-136.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ description</text>
-<text text-anchor="start" x="9082" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ arguments</text>
-<text text-anchor="start" x="9082" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ attrs_type_key</text>
-<text text-anchor="start" x="9082" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ attrs_type_index</text>
-<text text-anchor="start" x="9082" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ num_inputs</text>
-<text text-anchor="start" x="9082" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ support_level</text>
-<text text-anchor="start" x="9082" y="-70.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="9074,-63.5 9283,-63.5 "/>
-<text text-anchor="start" x="9082" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="9082" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="9082" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="9082" y="-18.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ IsPrimitiveOp()</text>
-<text text-anchor="start" x="9082" y="-7.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<g id="a_node45"><a xlink:href="classtvm_1_1relay_1_1CallNode.html" target="_top" xlink:title="Call container. ">
+<polygon fill="#ffffff" stroke="#000000" points="9074,-17 9074,-173 9283,-173 9283,-17 9074,-17"/>
+<text text-anchor="middle" x="9178.5" y="-161" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::CallNode</text>
+<polyline fill="none" stroke="#000000" points="9074,-154 9283,-154 "/>
+<text text-anchor="start" x="9082" y="-142" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ op</text>
+<text text-anchor="start" x="9082" y="-131" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ args</text>
+<text text-anchor="start" x="9082" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ attrs</text>
+<text text-anchor="start" x="9082" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ type_args</text>
+<text text-anchor="start" x="9082" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<text text-anchor="start" x="9082" y="-87" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># saved_deleter_</text>
+<polyline fill="none" stroke="#000000" points="9074,-80 9283,-80 "/>
+<text text-anchor="start" x="9082" y="-68" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="9082" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="9082" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="9082" y="-35" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<text text-anchor="start" x="9082" y="-24" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># Deleter_()</text>
</a>
</g>
</g>
-<!-- Node58->Node64 -->
+<!-- Node57->Node64 -->
<g id="edge44" class="edge">
-<title>Node58->Node64</title>
-<path fill="none" stroke="#191970" d="M9178.5,-216.3673C9178.5,-207.69 9178.5,-198.7249 9178.5,-189.7795"/>
-<polygon fill="none" stroke="#191970" points="9175.0001,-216.4029 9178.5,-226.403 9182.0001,-216.403 9175.0001,-216.4029"/>
+<title>Node57->Node64</title>
+<path fill="none" stroke="#191970" d="M9106.1031,-217.5666C9114.6002,-203.1813 9123.5907,-187.9604 9132.2391,-173.319"/>
+<polygon fill="none" stroke="#191970" points="9102.956,-216.0127 9100.8837,-226.403 9108.9831,-219.5728 9102.956,-216.0127"/>
</g>
<!-- Node65 -->
<g id="node46" class="node">
<title>Node65</title>
-<g id="a_node46"><a xlink:href="classtvm_1_1relay_1_1CallNode.html" target="_top" xlink:title="Call container. ">
-<polygon fill="#ffffff" stroke="#000000" points="9301,-17 9301,-173 9510,-173 9510,-17 9301,-17"/>
-<text text-anchor="middle" x="9405.5" y="-161" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::CallNode</text>
-<polyline fill="none" stroke="#000000" points="9301,-154 9510,-154 "/>
-<text text-anchor="start" x="9309" y="-142" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ op</text>
-<text text-anchor="start" x="9309" y="-131" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ args</text>
-<text text-anchor="start" x="9309" y="-120" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ attrs</text>
-<text text-anchor="start" x="9309" y="-109" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ type_args</text>
-<text text-anchor="start" x="9309" y="-98" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<text text-anchor="start" x="9309" y="-87" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># saved_deleter_</text>
-<polyline fill="none" stroke="#000000" points="9301,-80 9510,-80 "/>
-<text text-anchor="start" x="9309" y="-68" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
-<text text-anchor="start" x="9309" y="-57" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
-<text text-anchor="start" x="9309" y="-46" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
-<text text-anchor="start" x="9309" y="-35" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
-<text text-anchor="start" x="9309" y="-24" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># Deleter_()</text>
+<g id="a_node46"><a xlink:href="classtvm_1_1relay_1_1ConstantNode.html" target="_top" xlink:title="Constant tensor type. ">
+<polygon fill="#ffffff" stroke="#000000" points="9301,-33.5 9301,-156.5 9510,-156.5 9510,-33.5 9301,-33.5"/>
+<text text-anchor="middle" x="9405.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::ConstantNode</text>
+<polyline fill="none" stroke="#000000" points="9301,-137.5 9510,-137.5 "/>
+<text text-anchor="start" x="9309" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ data</text>
+<text text-anchor="start" x="9309" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="9301,-107.5 9510,-107.5 "/>
+<text text-anchor="start" x="9309" y="-95.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ tensor_type()</text>
+<text text-anchor="start" x="9309" y="-84.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ is_scalar()</text>
+<text text-anchor="start" x="9309" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
+<text text-anchor="start" x="9309" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
+<text text-anchor="start" x="9309" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
+<text text-anchor="start" x="9309" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
</a>
</g>
</g>
-<!-- Node58->Node65 -->
+<!-- Node57->Node65 -->
<g id="edge45" class="edge">
-<title>Node58->Node65</title>
-<path fill="none" stroke="#191970" d="M9258.9538,-219.5966C9276.5705,-204.6185 9295.3508,-188.6511 9313.3839,-173.319"/>
-<polygon fill="none" stroke="#191970" points="9256.2998,-217.259 9250.9483,-226.403 9260.834,-222.592 9256.2998,-217.259"/>
+<title>Node57->Node65</title>
+<path fill="none" stroke="#191970" d="M9177.6354,-247.0551C9215.3566,-231.4188 9256.8314,-211.9921 9292.5,-190 9308.2078,-180.3151 9324.0184,-168.5633 9338.6315,-156.6733"/>
+<polygon fill="none" stroke="#191970" points="9176.059,-243.9185 9168.132,-250.948 9178.7125,-250.3961 9176.059,-243.9185"/>
</g>
<!-- Node66 -->
<g id="node47" class="node">
<title>Node66</title>
-<g id="a_node47"><a xlink:href="classtvm_1_1relay_1_1ConstantNode.html" target="_top" xlink:title="Constant tensor type. ">
+<g id="a_node47"><a xlink:href="classtvm_1_1relay_1_1IfNode.html" target="_top" xlink:title="container of If ">
<polygon fill="#ffffff" stroke="#000000" points="9528,-33.5 9528,-156.5 9737,-156.5 9737,-33.5 9528,-33.5"/>
-<text text-anchor="middle" x="9632.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::ConstantNode</text>
+<text text-anchor="middle" x="9632.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::IfNode</text>
<polyline fill="none" stroke="#000000" points="9528,-137.5 9737,-137.5 "/>
-<text text-anchor="start" x="9536" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ data</text>
-<text text-anchor="start" x="9536" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<polyline fill="none" stroke="#000000" points="9528,-107.5 9737,-107.5 "/>
-<text text-anchor="start" x="9536" y="-95.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ tensor_type()</text>
-<text text-anchor="start" x="9536" y="-84.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ is_scalar()</text>
+<text text-anchor="start" x="9536" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ cond</text>
+<text text-anchor="start" x="9536" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ true_branch</text>
+<text text-anchor="start" x="9536" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ false_branch</text>
+<text text-anchor="start" x="9536" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<polyline fill="none" stroke="#000000" points="9528,-85.5 9737,-85.5 "/>
<text text-anchor="start" x="9536" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
<text text-anchor="start" x="9536" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
<text text-anchor="start" x="9536" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
@@ -1162,63 +1162,63 @@
</a>
</g>
</g>
-<!-- Node58->Node66 -->
+<!-- Node57->Node66 -->
<g id="edge46" class="edge">
-<title>Node58->Node66</title>
-<path fill="none" stroke="#191970" d="M9291.9816,-267.8766C9360.5846,-252.7313 9448.2518,-227.9537 9519.5,-190 9536.5785,-180.9023 9553.412,-168.9079 9568.6595,-156.5166"/>
-<polygon fill="none" stroke="#191970" points="9291.0329,-264.5009 9282.0024,-270.0417 9292.5172,-271.3417 9291.0329,-264.5009"/>
+<title>Node57->Node66</title>
+<path fill="none" stroke="#191970" d="M9178.05,-278.9269C9273.256,-267.7929 9410.698,-243.3121 9519.5,-190 9537.2219,-181.3164 9554.506,-169.262 9570.0124,-156.65"/>
+<polygon fill="none" stroke="#191970" points="9177.6438,-275.4504 9168.104,-280.0595 9178.4358,-282.4055 9177.6438,-275.4504"/>
</g>
<!-- Node67 -->
<g id="node48" class="node">
<title>Node67</title>
-<g id="a_node48"><a xlink:href="classtvm_1_1relay_1_1IfNode.html" target="_top" xlink:title="container of If ">
-<polygon fill="#ffffff" stroke="#000000" points="9755,-33.5 9755,-156.5 9964,-156.5 9964,-33.5 9755,-33.5"/>
-<text text-anchor="middle" x="9859.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::IfNode</text>
-<polyline fill="none" stroke="#000000" points="9755,-137.5 9964,-137.5 "/>
-<text text-anchor="start" x="9763" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ cond</text>
-<text text-anchor="start" x="9763" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ true_branch</text>
-<text text-anchor="start" x="9763" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ false_branch</text>
-<text text-anchor="start" x="9763" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<g id="a_node48"><a xlink:href="classtvm_1_1relay_1_1LetNode.html" target="_top" xlink:title="A binding of a sub-network. ">
+<polygon fill="#ffffff" stroke="#000000" points="9755,-22.5 9755,-167.5 9964,-167.5 9964,-22.5 9755,-22.5"/>
+<text text-anchor="middle" x="9859.5" y="-155.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::LetNode</text>
+<polyline fill="none" stroke="#000000" points="9755,-148.5 9964,-148.5 "/>
+<text text-anchor="start" x="9763" y="-136.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ var</text>
+<text text-anchor="start" x="9763" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ value</text>
+<text text-anchor="start" x="9763" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ body</text>
+<text text-anchor="start" x="9763" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
+<text text-anchor="start" x="9763" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># saved_deleter_</text>
<polyline fill="none" stroke="#000000" points="9755,-85.5 9964,-85.5 "/>
<text text-anchor="start" x="9763" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
<text text-anchor="start" x="9763" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
<text text-anchor="start" x="9763" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
<text text-anchor="start" x="9763" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
+<text text-anchor="start" x="9763" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># Deleter_()</text>
</a>
</g>
</g>
-<!-- Node58->Node67 -->
+<!-- Node57->Node67 -->
<g id="edge47" class="edge">
-<title>Node58->Node67</title>
-<path fill="none" stroke="#191970" d="M9292.1158,-275.5187C9433.9812,-258.7114 9667.6252,-226.6486 9746.5,-190 9764.6298,-181.5761 9782.202,-169.4663 9797.8734,-156.6986"/>
-<polygon fill="none" stroke="#191970" points="9291.5856,-272.0569 9282.0635,-276.7023 9292.4042,-279.0089 9291.5856,-272.0569"/>
+<title>Node57->Node67</title>
+<path fill="none" stroke="#191970" d="M9178.4452,-279.2462C9345.3685,-265.1416 9647.3132,-234.4008 9746.5,-190 9759.6001,-184.1358 9772.3761,-176.2913 9784.3429,-167.6134"/>
+<polygon fill="none" stroke="#191970" points="9177.8137,-275.7868 9168.1412,-280.1102 9178.3987,-282.7623 9177.8137,-275.7868"/>
</g>
<!-- Node68 -->
<g id="node49" class="node">
<title>Node68</title>
-<g id="a_node49"><a xlink:href="classtvm_1_1relay_1_1LetNode.html" target="_top" xlink:title="A binding of a sub-network. ">
-<polygon fill="#ffffff" stroke="#000000" points="9982,-22.5 9982,-167.5 10191,-167.5 10191,-22.5 9982,-22.5"/>
-<text text-anchor="middle" x="10086.5" y="-155.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::LetNode</text>
-<polyline fill="none" stroke="#000000" points="9982,-148.5 10191,-148.5 "/>
-<text text-anchor="start" x="9990" y="-136.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ var</text>
-<text text-anchor="start" x="9990" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ value</text>
-<text text-anchor="start" x="9990" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ body</text>
-<text text-anchor="start" x="9990" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
-<text text-anchor="start" x="9990" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># saved_deleter_</text>
+<g id="a_node49"><a xlink:href="classtvm_1_1relay_1_1MatchNode.html" target="_top" xlink:title="Match container node. ">
+<polygon fill="#ffffff" stroke="#000000" points="9982,-33.5 9982,-156.5 10191,-156.5 10191,-33.5 9982,-33.5"/>
+<text text-anchor="middle" x="10086.5" y="-144.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::relay::MatchNode</text>
+<polyline fill="none" stroke="#000000" points="9982,-137.5 10191,-137.5 "/>
+<text text-anchor="start" x="9990" y="-125.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ data</text>
+<text text-anchor="start" x="9990" y="-114.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ clauses</text>
+<text text-anchor="start" x="9990" y="-103.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ complete</text>
+<text text-anchor="start" x="9990" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_key</text>
<polyline fill="none" stroke="#000000" points="9982,-85.5 10191,-85.5 "/>
<text text-anchor="start" x="9990" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ VisitAttrs()</text>
<text text-anchor="start" x="9990" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SEqualReduce()</text>
<text text-anchor="start" x="9990" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ SHashReduce()</text>
<text text-anchor="start" x="9990" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DECLARE_FINAL_OBJECT_INFO()</text>
-<text text-anchor="start" x="9990" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># Deleter_()</text>
</a>
</g>
</g>
-<!-- Node58->Node68 -->
+<!-- Node57->Node68 -->
<g id="edge48" class="edge">
-<title>Node58->Node68</title>
-<path fill="none" stroke="#191970" d="M9292.3216,-281.9383C9482.0144,-270.5179 9853.8071,-242.1452 9973.5,-190 9986.7757,-184.2163 9999.6957,-176.3582 10011.7702,-167.6229"/>
-<polygon fill="none" stroke="#191970" points="9292.0905,-278.4458 9282.3167,-282.5351 9292.5074,-285.4334 9292.0905,-278.4458"/>
+<title>Node57->Node68</title>
+<path fill="none" stroke="#191970" d="M9178.2862,-283.9808C9389.3678,-275.1775 9832.9084,-249.9781 9973.5,-190 9992.3594,-181.9544 10010.4668,-169.6599 10026.4505,-156.5578"/>
+<polygon fill="none" stroke="#191970" points="9177.9523,-280.4915 9168.1048,-284.4002 9178.2404,-287.4855 9177.9523,-280.4915"/>
</g>
</g>
</svg>
diff --git a/docs/reference/api/doxygen/classtvm_1_1BaseExpr__coll__graph.svg b/docs/reference/api/doxygen/classtvm_1_1BaseExpr__coll__graph.svg
index 56a74eba62..018cfdfeb8 100644
--- a/docs/reference/api/doxygen/classtvm_1_1BaseExpr__coll__graph.svg
+++ b/docs/reference/api/doxygen/classtvm_1_1BaseExpr__coll__graph.svg
@@ -9,9 +9,9 @@
<g id="graph0" class="graph" transform="scale(1 1) rotate(0) translate(4 559)">
<title>tvm::BaseExpr</title>
<polygon fill="#ffffff" stroke="transparent" points="-4,4 -4,-559 158,-559 158,4 -4,4"/>
-<!-- Node62 -->
+<!-- Node61 -->
<g id="node1" class="node">
-<title>Node62</title>
+<title>Node61</title>
<polygon fill="#bfbfbf" stroke="#000000" points="0,-.5 0,-68.5 154,-68.5 154,-.5 0,-.5"/>
<text text-anchor="middle" x="77" y="-56.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::BaseExpr</text>
<polyline fill="none" stroke="#000000" points="0,-49.5 154,-49.5 "/>
@@ -20,9 +20,9 @@
<text text-anchor="start" x="8" y="-18.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="8" y="-7.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
</g>
-<!-- Node63 -->
+<!-- Node62 -->
<g id="node2" class="node">
-<title>Node63</title>
+<title>Node62</title>
<g id="a_node2"><a xlink:href="classtvm_1_1runtime_1_1ObjectRef.html" target="_top" xlink:title="Base class of all object reference. ">
<polygon fill="#ffffff" stroke="#000000" points="10,-106.5 10,-328.5 144,-328.5 144,-106.5 10,-106.5"/>
<text text-anchor="middle" x="77" y="-316.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::ObjectRef</text>
@@ -48,15 +48,15 @@
</a>
</g>
</g>
-<!-- Node63->Node62 -->
+<!-- Node62->Node61 -->
<g id="edge1" class="edge">
-<title>Node63->Node62</title>
+<title>Node62->Node61</title>
<path fill="none" stroke="#191970" d="M77,-96.2198C77,-86.2931 77,-76.8914 77,-68.5572"/>
<polygon fill="none" stroke="#191970" points="73.5001,-96.3012 77,-106.3012 80.5001,-96.3012 73.5001,-96.3012"/>
</g>
-<!-- Node64 -->
+<!-- Node63 -->
<g id="node3" class="node">
-<title>Node64</title>
+<title>Node63</title>
<g id="a_node3"><a xlink:href="classtvm_1_1runtime_1_1ObjectPtr.html" target="_top" xlink:title="{tvm::runtime::ObjectPtr\l\< tvm::runtime::Object \>\n||+ ObjectPtr()\l+ ObjectPtr()\l+ ObjectPtr()\l+ ObjectPtr()\l+ ObjectPtr()\l+ ObjectPtr()\l+ ~ObjectPtr()\l+ swap()\l+ get()\l+ operator-\>()\land 11 more...\l}">
<polygon fill="#ffffff" stroke="#000000" points="7,-376.5 7,-554.5 147,-554.5 147,-376.5 7,-376.5"/>
<text text-anchor="start" x="15" y="-542.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::ObjectPtr</text>
@@ -78,9 +78,9 @@
</a>
</g>
</g>
-<!-- Node64->Node63 -->
+<!-- Node63->Node62 -->
<g id="edge2" class="edge">
-<title>Node64->Node63</title>
+<title>Node63->Node62</title>
<path fill="none" stroke="#404040" d="M77,-376.3167C77,-364.8765 77,-353.0062 77,-341.1402"/>
<polygon fill="none" stroke="#404040" points="77.0001,-340.7944 73,-334.7944 77,-328.7944 81,-334.7943 77.0001,-340.7944"/>
<text text-anchor="middle" x="96.5" y="-350" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> #data_</text>
diff --git a/docs/reference/api/doxygen/classtvm_1_1BaseExpr__inherit__graph.svg b/docs/reference/api/doxygen/classtvm_1_1BaseExpr__inherit__graph.svg
index c1d50326d8..dfd90bb23a 100644
--- a/docs/reference/api/doxygen/classtvm_1_1BaseExpr__inherit__graph.svg
+++ b/docs/reference/api/doxygen/classtvm_1_1BaseExpr__inherit__graph.svg
@@ -12,91 +12,91 @@
<!-- Node0 -->
<g id="node1" class="node">
<title>Node0</title>
-<polygon fill="#bfbfbf" stroke="#000000" points="4779,-298.5 4779,-366.5 4933,-366.5 4933,-298.5 4779,-298.5"/>
-<text text-anchor="middle" x="4856" y="-354.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::BaseExpr</text>
-<polyline fill="none" stroke="#000000" points="4779,-347.5 4933,-347.5 "/>
-<text text-anchor="middle" x="4856" y="-335.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
-<polyline fill="none" stroke="#000000" points="4779,-328.5 4933,-328.5 "/>
-<text text-anchor="start" x="4787" y="-316.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
-<text text-anchor="start" x="4787" y="-305.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
+<polygon fill="#bfbfbf" stroke="#000000" points="4693,-298.5 4693,-366.5 4847,-366.5 4847,-298.5 4693,-298.5"/>
+<text text-anchor="middle" x="4770" y="-354.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::BaseExpr</text>
+<polyline fill="none" stroke="#000000" points="4693,-347.5 4847,-347.5 "/>
+<text text-anchor="middle" x="4770" y="-335.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
+<polyline fill="none" stroke="#000000" points="4693,-328.5 4847,-328.5 "/>
+<text text-anchor="start" x="4701" y="-316.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
+<text text-anchor="start" x="4701" y="-305.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
</g>
<!-- Node2 -->
<g id="node3" class="node">
<title>Node2</title>
<g id="a_node3"><a xlink:href="classtvm_1_1PrimExpr.html" target="_top" xlink:title="Reference to PrimExprNode. ">
-<polygon fill="#ffffff" stroke="#000000" points="3022,-160.5 3022,-261.5 3176,-261.5 3176,-160.5 3022,-160.5"/>
-<text text-anchor="middle" x="3099" y="-249.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::PrimExpr</text>
-<polyline fill="none" stroke="#000000" points="3022,-242.5 3176,-242.5 "/>
-<text text-anchor="middle" x="3099" y="-230.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
-<polyline fill="none" stroke="#000000" points="3022,-223.5 3176,-223.5 "/>
-<text text-anchor="start" x="3030" y="-211.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ PrimExpr()</text>
-<text text-anchor="start" x="3030" y="-200.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ PrimExpr()</text>
-<text text-anchor="start" x="3030" y="-189.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ dtype()</text>
-<text text-anchor="start" x="3030" y="-178.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
-<text text-anchor="start" x="3030" y="-167.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
+<polygon fill="#ffffff" stroke="#000000" points="2936,-160.5 2936,-261.5 3090,-261.5 3090,-160.5 2936,-160.5"/>
+<text text-anchor="middle" x="3013" y="-249.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::PrimExpr</text>
+<polyline fill="none" stroke="#000000" points="2936,-242.5 3090,-242.5 "/>
+<text text-anchor="middle" x="3013" y="-230.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
+<polyline fill="none" stroke="#000000" points="2936,-223.5 3090,-223.5 "/>
+<text text-anchor="start" x="2944" y="-211.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ PrimExpr()</text>
+<text text-anchor="start" x="2944" y="-200.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ PrimExpr()</text>
+<text text-anchor="start" x="2944" y="-189.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ dtype()</text>
+<text text-anchor="start" x="2944" y="-178.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
+<text text-anchor="start" x="2944" y="-167.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
</a>
</g>
</g>
<!-- Node0->Node2 -->
<g id="edge2" class="edge">
<title>Node0->Node2</title>
-<path fill="none" stroke="#191970" d="M4768.7634,-326.4674C4466.0147,-305.5318 3463.5312,-236.208 3176.4439,-216.3554"/>
-<polygon fill="none" stroke="#191970" points="4768.5888,-329.9636 4778.8064,-327.1619 4769.0717,-322.9803 4768.5888,-329.9636"/>
+<path fill="none" stroke="#191970" d="M4682.7634,-326.4674C4380.0147,-305.5318 3377.5312,-236.208 3090.4439,-216.3554"/>
+<polygon fill="none" stroke="#191970" points="4682.5888,-329.9636 4692.8064,-327.1619 4683.0717,-322.9803 4682.5888,-329.9636"/>
</g>
-<!-- Node43 -->
-<g id="node39" class="node">
-<title>Node43</title>
-<g id="a_node39"><a xlink:href="classtvm_1_1RelayExpr.html" target="_top" xlink:title="Managed reference to RelayExprNode. ">
-<polygon fill="#ffffff" stroke="#ff0000" points="6622,-177 6622,-245 6776,-245 6776,-177 6622,-177"/>
-<text text-anchor="middle" x="6699" y="-233" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::RelayExpr</text>
-<polyline fill="none" stroke="#ff0000" points="6622,-226 6776,-226 "/>
-<text text-anchor="middle" x="6699" y="-214" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
-<polyline fill="none" stroke="#ff0000" points="6622,-207 6776,-207 "/>
-<text text-anchor="start" x="6630" y="-195" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
-<text text-anchor="start" x="6630" y="-184" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
+<!-- Node42 -->
+<g id="node38" class="node">
+<title>Node42</title>
+<g id="a_node38"><a xlink:href="classtvm_1_1RelayExpr.html" target="_top" xlink:title="Managed reference to RelayExprNode. ">
+<polygon fill="#ffffff" stroke="#ff0000" points="6536,-177 6536,-245 6690,-245 6690,-177 6536,-177"/>
+<text text-anchor="middle" x="6613" y="-233" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::RelayExpr</text>
+<polyline fill="none" stroke="#ff0000" points="6536,-226 6690,-226 "/>
+<text text-anchor="middle" x="6613" y="-214" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
+<polyline fill="none" stroke="#ff0000" points="6536,-207 6690,-207 "/>
+<text text-anchor="start" x="6544" y="-195" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
+<text text-anchor="start" x="6544" y="-184" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
</a>
</g>
</g>
-<!-- Node0->Node43 -->
-<g id="edge38" class="edge">
-<title>Node0->Node43</title>
-<path fill="none" stroke="#191970" d="M4943.546,-326.7285C5257.6623,-306.0204 6325.068,-235.6515 6621.868,-216.0849"/>
-<polygon fill="none" stroke="#191970" points="4942.8846,-323.2644 4933.1366,-327.4148 4943.3452,-330.2493 4942.8846,-323.2644"/>
+<!-- Node0->Node42 -->
+<g id="edge37" class="edge">
+<title>Node0->Node42</title>
+<path fill="none" stroke="#191970" d="M4857.546,-326.7285C5171.6623,-306.0204 6239.068,-235.6515 6535.868,-216.0849"/>
+<polygon fill="none" stroke="#191970" points="4856.8846,-323.2644 4847.1366,-327.4148 4857.3452,-330.2493 4856.8846,-323.2644"/>
</g>
<!-- Node1 -->
<g id="node2" class="node">
<title>Node1</title>
<g id="a_node2"><a xlink:href="classtvm_1_1runtime_1_1ObjectRef.html" target="_top" xlink:title="Base class of all object reference. ">
-<polygon fill="#ffffff" stroke="#000000" points="4789,-403.5 4789,-636.5 4923,-636.5 4923,-403.5 4789,-403.5"/>
-<text text-anchor="middle" x="4856" y="-624.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::ObjectRef</text>
-<polyline fill="none" stroke="#000000" points="4789,-617.5 4923,-617.5 "/>
-<text text-anchor="start" x="4797" y="-605.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_is_nullable</text>
-<text text-anchor="start" x="4797" y="-594.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># data_</text>
-<polyline fill="none" stroke="#000000" points="4789,-587.5 4923,-587.5 "/>
-<text text-anchor="start" x="4797" y="-575.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ObjectRef()</text>
-<text text-anchor="start" x="4797" y="-564.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ObjectRef()</text>
-<text text-anchor="start" x="4797" y="-553.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ same_as()</text>
-<text text-anchor="start" x="4797" y="-542.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator==()</text>
-<text text-anchor="start" x="4797" y="-531.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator!=()</text>
-<text text-anchor="start" x="4797" y="-520.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator<()</text>
-<text text-anchor="start" x="4797" y="-509.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ defined()</text>
-<text text-anchor="start" x="4797" y="-498.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ get()</text>
-<text text-anchor="start" x="4797" y="-487.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator->()</text>
-<text text-anchor="start" x="4797" y="-476.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ unique()</text>
-<text text-anchor="start" x="4797" y="-465.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ use_count()</text>
-<text text-anchor="start" x="4797" y="-454.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ as()</text>
-<text text-anchor="start" x="4797" y="-443.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># get_mutable()</text>
-<text text-anchor="start" x="4797" y="-432.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># DowncastNoCheck()</text>
-<text text-anchor="start" x="4797" y="-421.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># FFIClearAfterMove()</text>
-<text text-anchor="start" x="4797" y="-410.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># GetDataPtr()</text>
+<polygon fill="#ffffff" stroke="#000000" points="4703,-403.5 4703,-636.5 4837,-636.5 4837,-403.5 4703,-403.5"/>
+<text text-anchor="middle" x="4770" y="-624.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::runtime::ObjectRef</text>
+<polyline fill="none" stroke="#000000" points="4703,-617.5 4837,-617.5 "/>
+<text text-anchor="start" x="4711" y="-605.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ _type_is_nullable</text>
+<text text-anchor="start" x="4711" y="-594.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># data_</text>
+<polyline fill="none" stroke="#000000" points="4703,-587.5 4837,-587.5 "/>
+<text text-anchor="start" x="4711" y="-575.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ObjectRef()</text>
+<text text-anchor="start" x="4711" y="-564.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ObjectRef()</text>
+<text text-anchor="start" x="4711" y="-553.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ same_as()</text>
+<text text-anchor="start" x="4711" y="-542.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator==()</text>
+<text text-anchor="start" x="4711" y="-531.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator!=()</text>
+<text text-anchor="start" x="4711" y="-520.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator<()</text>
+<text text-anchor="start" x="4711" y="-509.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ defined()</text>
+<text text-anchor="start" x="4711" y="-498.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ get()</text>
+<text text-anchor="start" x="4711" y="-487.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ operator->()</text>
+<text text-anchor="start" x="4711" y="-476.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ unique()</text>
+<text text-anchor="start" x="4711" y="-465.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ use_count()</text>
+<text text-anchor="start" x="4711" y="-454.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ as()</text>
+<text text-anchor="start" x="4711" y="-443.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># get_mutable()</text>
+<text text-anchor="start" x="4711" y="-432.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># DowncastNoCheck()</text>
+<text text-anchor="start" x="4711" y="-421.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># FFIClearAfterMove()</text>
+<text text-anchor="start" x="4711" y="-410.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"># GetDataPtr()</text>
</a>
</g>
</g>
<!-- Node1->Node0 -->
<g id="edge1" class="edge">
<title>Node1->Node0</title>
-<path fill="none" stroke="#191970" d="M4856,-393.2822C4856,-383.749 4856,-374.737 4856,-366.7166"/>
-<polygon fill="none" stroke="#191970" points="4852.5001,-393.3857 4856,-403.3857 4859.5001,-393.3857 4852.5001,-393.3857"/>
+<path fill="none" stroke="#191970" d="M4770,-393.2822C4770,-383.749 4770,-374.737 4770,-366.7166"/>
+<polygon fill="none" stroke="#191970" points="4766.5001,-393.3857 4770,-403.3857 4773.5001,-393.3857 4766.5001,-393.3857"/>
</g>
<!-- Node3 -->
<g id="node4" class="node">
@@ -115,8 +115,8 @@
<!-- Node2->Node3 -->
<g id="edge3" class="edge">
<title>Node2->Node3</title>
-<path fill="none" stroke="#191970" d="M3011.7445,-209.8556C2539.0363,-203.4742 301.1548,-170.8788 163,-124 145.3957,-118.0265 128.296,-107.1159 114.0429,-96.0936"/>
-<polygon fill="none" stroke="#191970" points="3011.8717,-213.3575 3021.9179,-209.9925 3011.966,-206.3582 3011.8717,-213.3575"/>
+<path fill="none" stroke="#191970" d="M2925.6691,-209.7734C2461.1981,-203.0685 296.8368,-169.4803 163,-124 145.3984,-118.0186 128.2993,-107.1064 114.0459,-96.085"/>
+<polygon fill="none" stroke="#191970" points="2925.6216,-213.273 2935.671,-209.9175 2925.7224,-206.2737 2925.6216,-213.273"/>
</g>
<!-- Node6 -->
<g id="node5" class="node">
@@ -138,8 +138,8 @@
<!-- Node2->Node6 -->
<g id="edge4" class="edge">
<title>Node2->Node6</title>
-<path fill="none" stroke="#191970" d="M3011.7525,-209.6886C2555.9803,-202.6567 464.5378,-168.0879 335,-124 326.6362,-121.1534 318.3855,-117.1946 310.4993,-112.6407"/>
-<polygon fill="none" stroke="#191970" points="3011.8401,-213.1902 3021.8928,-209.8446 3011.9478,-206.1911 3011.8401,-213.1902"/>
+<path fill="none" stroke="#191970" d="M2925.5966,-209.5944C2478.1684,-202.2178 460.22,-166.6889 335,-124 326.6377,-121.1492 318.3878,-117.1877 310.5022,-112.6324"/>
+<polygon fill="none" stroke="#191970" points="2925.8157,-213.0984 2935.8719,-209.7634 2925.9309,-206.0993 2925.8157,-213.0984"/>
</g>
<!-- Node7 -->
<g id="node6" class="node">
@@ -161,8 +161,8 @@
<!-- Node2->Node7 -->
<g id="edge5" class="edge">
<title>Node2->Node7</title>
-<path fill="none" stroke="#191970" d="M3011.9075,-209.5021C2573.8936,-201.7882 627.9496,-165.3058 507,-124 498.6392,-121.1447 490.3903,-117.1804 482.5052,-112.6235"/>
-<polygon fill="none" stroke="#191970" points="3011.9122,-213.0026 3021.9721,-209.6789 3012.0352,-206.0037 3011.9122,-213.0026"/>
+<path fill="none" stroke="#191970" d="M2925.6833,-209.3937C2496.1784,-201.3119 623.6341,-163.9069 507,-124 498.6409,-121.1399 490.393,-117.1726 482.5085,-112.6139"/>
+<polygon fill="none" stroke="#191970" points="2925.7974,-212.8964 2935.8614,-209.5848 2925.9289,-205.8976 2925.7974,-212.8964"/>
</g>
<!-- Node10 -->
<g id="node7" class="node">
@@ -184,8 +184,8 @@
<!-- Node2->Node10 -->
<g id="edge6" class="edge">
<title>Node2->Node10</title>
-<path fill="none" stroke="#191970" d="M3011.6244,-209.2807C2591.1139,-200.826 791.3384,-162.5145 679,-124 670.6426,-121.1347 662.3958,-117.1642 654.512,-112.6037"/>
-<polygon fill="none" stroke="#191970" points="3011.8239,-212.7854 3021.8921,-209.4868 3011.9643,-205.7868 3011.8239,-212.7854"/>
+<path fill="none" stroke="#191970" d="M2925.8022,-210.7328C2527.8617,-209.1121 894.2934,-197.9711 679,-124 670.6445,-121.1292 662.3989,-117.1553 654.5157,-112.5928"/>
+<polygon fill="none" stroke="#191970" points="2925.8003,-214.2327 2935.8142,-210.7726 2925.8282,-207.2328 2925.8003,-214.2327"/>
</g>
<!-- Node11 -->
<g id="node8" class="node">
@@ -207,8 +207,8 @@
<!-- Node2->Node11 -->
<g id="edge7" class="edge">
<title>Node2->Node11</title>
-<path fill="none" stroke="#191970" d="M3011.8488,-210.5986C2623.15,-208.4081 1057.7552,-195.2022 851,-124 842.6466,-121.1233 834.4022,-117.1456 826.5198,-112.581"/>
-<polygon fill="none" stroke="#191970" points="3011.89,-214.0988 3021.9092,-210.6543 3011.9288,-207.0989 3011.89,-214.0988"/>
+<path fill="none" stroke="#191970" d="M2925.7724,-210.4531C2546.2869,-207.6737 1049.2061,-192.4285 851,-124 842.6488,-121.1168 834.4058,-117.1352 826.5241,-112.5683"/>
+<polygon fill="none" stroke="#191970" points="2925.8445,-213.9536 2935.8696,-210.526 2925.8952,-206.9538 2925.8445,-213.9536"/>
</g>
<!-- Node12 -->
<g id="node9" class="node">
@@ -230,8 +230,8 @@
<!-- Node2->Node12 -->
<g id="edge8" class="edge">
<title>Node2->Node12</title>
-<path fill="none" stroke="#191970" d="M3011.8502,-210.1196C2643.4931,-206.0202 1224.5069,-186.2887 1035,-124 1026.2171,-121.1132 1017.4935,-117.1292 1009.1156,-112.5611"/>
-<polygon fill="none" stroke="#191970" points="3011.8821,-213.62 3021.9201,-210.2306 3011.9593,-206.6205 3011.8821,-213.62"/>
+<path fill="none" stroke="#191970" d="M2925.6435,-209.9446C2566.7255,-205.2305 1215.9194,-183.6495 1035,-124 1026.2198,-121.1051 1017.4978,-117.1162 1009.1208,-112.5452"/>
+<polygon fill="none" stroke="#191970" points="2925.6727,-213.4452 2935.7175,-210.0759 2925.7639,-206.4458 2925.6727,-213.4452"/>
</g>
<!-- Node13 -->
<g id="node10" class="node">
@@ -253,8 +253,8 @@
<!-- Node2->Node13 -->
<g id="edge9" class="edge">
<title>Node2->Node13</title>
-<path fill="none" stroke="#191970" d="M3011.9393,-209.9147C2663.3038,-205.1724 1378.5141,-183.7641 1207,-124 1198.6571,-121.0929 1190.4193,-117.0963 1182.5406,-112.5209"/>
-<polygon fill="none" stroke="#191970" points="3011.9325,-213.4149 3021.9788,-210.0502 3012.027,-206.4155 3011.9325,-213.4149"/>
+<path fill="none" stroke="#191970" d="M2925.7222,-209.7058C2586.9769,-204.2863 1369.996,-180.9957 1207,-124 1198.6603,-121.0838 1190.4245,-117.0815 1182.5469,-112.5029"/>
+<polygon fill="none" stroke="#191970" points="2925.9113,-213.2091 2935.9656,-209.8685 2926.0225,-206.21 2925.9113,-213.2091"/>
</g>
<!-- Node14 -->
<g id="node11" class="node">
@@ -276,8 +276,8 @@
<!-- Node2->Node14 -->
<g id="edge10" class="edge">
<title>Node2->Node14</title>
-<path fill="none" stroke="#191970" d="M3011.5414,-209.4759C2683.0071,-203.355 1533.5004,-178.2333 1379,-124 1370.8374,-121.1347 1362.7733,-117.2318 1355.0466,-112.7687"/>
-<polygon fill="none" stroke="#191970" points="3011.8881,-212.9829 3021.9512,-209.6686 3012.0177,-205.9841 3011.8881,-212.9829"/>
+<path fill="none" stroke="#191970" d="M2925.9053,-209.2331C2608.5628,-202.4024 1525.1058,-175.5046 1379,-124 1370.8412,-121.1239 1362.7795,-117.2141 1355.0542,-112.747"/>
+<polygon fill="none" stroke="#191970" points="2925.8974,-212.7337 2935.97,-209.4485 2926.0472,-205.7353 2925.8974,-212.7337"/>
</g>
<!-- Node15 -->
<g id="node12" class="node">
@@ -299,8 +299,8 @@
<!-- Node2->Node15 -->
<g id="edge11" class="edge">
<title>Node2->Node15</title>
-<path fill="none" stroke="#191970" d="M3011.6356,-208.9493C2704.9586,-201.3583 1688.6316,-172.7456 1551,-124 1542.8455,-121.1119 1534.7865,-117.1944 1527.0627,-112.723"/>
-<polygon fill="none" stroke="#191970" points="3011.7252,-212.4525 3021.8084,-209.1998 3011.8976,-205.4546 3011.7252,-212.4525"/>
+<path fill="none" stroke="#191970" d="M2925.2867,-208.629C2629.5311,-200.2418 1680.1672,-169.9874 1551,-124 1542.8503,-121.0985 1534.7942,-117.1725 1527.0722,-112.6961"/>
+<polygon fill="none" stroke="#191970" points="2925.4346,-212.1344 2935.5295,-208.9181 2925.6322,-205.1372 2925.4346,-212.1344"/>
</g>
<!-- Node16 -->
<g id="node13" class="node">
@@ -322,8 +322,8 @@
<!-- Node2->Node16 -->
<g id="edge12" class="edge">
<title>Node2->Node16</title>
-<path fill="none" stroke="#191970" d="M3011.6686,-208.2909C2728.0057,-199.1028 1843.8302,-167.2716 1723,-124 1714.8557,-121.0834 1706.803,-117.1479 1699.083,-112.6659"/>
-<polygon fill="none" stroke="#191970" points="3011.7996,-211.7969 3021.9073,-208.6212 3012.0254,-204.8005 3011.7996,-211.7969"/>
+<path fill="none" stroke="#191970" d="M2925.5998,-207.8933C2653.8951,-197.8488 1835.4514,-164.5374 1723,-124 1714.8618,-121.0663 1706.8131,-117.12 1699.0953,-112.6318"/>
+<polygon fill="none" stroke="#191970" points="2925.6851,-211.3988 2935.8072,-208.2692 2925.9428,-204.4035 2925.6851,-211.3988"/>
</g>
<!-- Node17 -->
<g id="node14" class="node">
@@ -345,8 +345,8 @@
<!-- Node2->Node17 -->
<g id="edge13" class="edge">
<title>Node2->Node17</title>
-<path fill="none" stroke="#191970" d="M3011.6742,-207.4451C2752.4806,-196.5097 1999.1179,-161.8155 1895,-124 1886.8689,-121.0468 1878.8246,-117.0881 1871.1094,-112.5928"/>
-<polygon fill="none" stroke="#191970" points="3011.6569,-210.9473 3021.7952,-207.8706 3011.951,-203.9535 3011.6569,-210.9473"/>
+<path fill="none" stroke="#191970" d="M2925.5959,-210.2633C2696.6139,-207.3744 2083.9264,-193.2094 1895,-124 1886.877,-121.0243 1878.838,-117.0515 1871.1258,-112.5479"/>
+<polygon fill="none" stroke="#191970" points="2925.837,-213.7664 2935.879,-210.3887 2925.9224,-206.767 2925.837,-213.7664"/>
</g>
<!-- Node18 -->
<g id="node15" class="node">
@@ -368,8 +368,8 @@
<!-- Node2->Node18 -->
<g id="edge14" class="edge">
<title>Node2->Node18</title>
-<path fill="none" stroke="#191970" d="M3011.8454,-209.6075C2795.6102,-205.2061 2239.5315,-187.8346 2067,-124 2059.0593,-121.062 2051.1955,-117.1781 2043.6375,-112.7829"/>
-<polygon fill="none" stroke="#191970" points="3011.8148,-213.1075 3021.8826,-209.8074 3011.9543,-206.1089 3011.8148,-213.1075"/>
+<path fill="none" stroke="#191970" d="M2925.9439,-208.8338C2722.8577,-202.8465 2223.1705,-182.4573 2067,-124 2059.0705,-121.0319 2051.214,-117.1286 2043.6602,-112.722"/>
+<polygon fill="none" stroke="#191970" points="2925.8752,-212.3331 2935.9726,-209.1248 2926.0784,-205.3361 2925.8752,-212.3331"/>
</g>
<!-- Node19 -->
<g id="node16" class="node">
@@ -391,8 +391,8 @@
<!-- Node2->Node19 -->
<g id="edge15" class="edge">
<title>Node2->Node19</title>
-<path fill="none" stroke="#191970" d="M3011.7934,-207.8997C2822.3115,-200.2417 2378.8348,-177.0701 2239,-124 2231.0841,-120.9958 2223.2363,-117.0693 2215.6876,-112.6492"/>
-<polygon fill="none" stroke="#191970" points="3011.859,-211.405 3021.9907,-208.3067 3012.1382,-204.4105 3011.859,-211.405"/>
+<path fill="none" stroke="#191970" d="M2925.6752,-206.7609C2750.5426,-197.3534 2362.6129,-171.7011 2239,-124 2231.1009,-120.9518 2223.2639,-116.9973 2215.7216,-112.5605"/>
+<polygon fill="none" stroke="#191970" points="2925.7124,-210.2678 2935.8842,-207.3038 2926.0842,-203.2776 2925.7124,-210.2678"/>
</g>
<!-- Node20 -->
<g id="node17" class="node">
@@ -414,8 +414,8 @@
<!-- Node2->Node20 -->
<g id="edge16" class="edge">
<title>Node2->Node20</title>
-<path fill="none" stroke="#191970" d="M3011.442,-205.3325C2851.4619,-194.0901 2518.4893,-166.3368 2411,-124 2403.2935,-120.9646 2395.639,-117.0807 2388.2563,-112.7402"/>
-<polygon fill="none" stroke="#191970" points="3011.5061,-208.8454 3021.7252,-206.0491 3011.9928,-201.8623 3011.5061,-208.8454"/>
+<path fill="none" stroke="#191970" d="M2925.6336,-203.5375C2781.9642,-190.4075 2502.6076,-161.0234 2411,-124 2403.3207,-120.8964 2395.6839,-116.9682 2388.3118,-112.6011"/>
+<polygon fill="none" stroke="#191970" points="2925.4634,-207.0363 2935.7389,-204.4546 2926.0962,-200.065 2925.4634,-207.0363"/>
</g>
<!-- Node21 -->
<g id="node18" class="node">
@@ -437,8 +437,8 @@
<!-- Node2->Node21 -->
<g id="edge17" class="edge">
<title>Node2->Node21</title>
-<path fill="none" stroke="#191970" d="M3011.8179,-201.1602C2885.6336,-186.0977 2658.8999,-155.7179 2583,-124 2575.5276,-120.8773 2568.0833,-116.9965 2560.8781,-112.7105"/>
-<polygon fill="none" stroke="#191970" points="3011.5005,-204.6471 3021.8434,-202.35 3012.3255,-197.6958 3011.5005,-204.6471"/>
+<path fill="none" stroke="#191970" d="M2925.7936,-204.2501C2836.5423,-194.8783 2695.389,-173.133 2583,-124 2575.7481,-120.8297 2568.506,-116.9752 2561.4751,-112.7557"/>
+<polygon fill="none" stroke="#191970" points="2925.5731,-207.7457 2935.8768,-205.2794 2926.284,-200.7819 2925.5731,-207.7457"/>
</g>
<!-- Node22 -->
<g id="node19" class="node">
@@ -460,8 +460,8 @@
<!-- Node2->Node22 -->
<g id="edge18" class="edge">
<title>Node2->Node22</title>
-<path fill="none" stroke="#191970" d="M3011.8505,-198.3094C2940.4087,-185.7375 2837.9232,-162.6518 2755,-124 2747.9932,-120.734 2740.9707,-116.8709 2734.1263,-112.6944"/>
-<polygon fill="none" stroke="#191970" points="3011.2701,-201.7609 3021.7199,-200.0142 3012.4616,-194.8631 3011.2701,-201.7609"/>
+<path fill="none" stroke="#191970" d="M2926.0382,-188.7385C2874.7307,-173.8962 2809.4013,-151.9726 2755,-124 2748.4524,-120.6333 2741.8514,-116.805 2735.3734,-112.7416"/>
+<polygon fill="none" stroke="#191970" points="2925.2002,-192.139 2935.7769,-191.5197 2927.1225,-185.4081 2925.2002,-192.139"/>
</g>
<!-- Node23 -->
<g id="node20" class="node">
@@ -483,8 +483,8 @@
<!-- Node2->Node23 -->
<g id="edge19" class="edge">
<title>Node2->Node23</title>
-<path fill="none" stroke="#191970" d="M3012.5182,-170.5589C2984.7107,-156.6871 2954.0994,-140.4832 2927,-124 2921.1815,-120.4609 2915.2485,-116.6498 2909.3556,-112.719"/>
-<polygon fill="none" stroke="#191970" points="3011.3173,-173.8698 3021.832,-175.1709 3014.4237,-167.5968 3011.3173,-173.8698"/>
+<path fill="none" stroke="#191970" d="M2946.7236,-153.5862C2931.1448,-140.0906 2914.6534,-125.8044 2899.447,-112.6314"/>
+<polygon fill="none" stroke="#191970" points="2944.4384,-156.2372 2954.2884,-160.1394 2949.0217,-150.9463 2944.4384,-156.2372"/>
</g>
<!-- Node24 -->
<g id="node21" class="node">
@@ -506,19 +506,19 @@
<!-- Node2->Node24 -->
<g id="edge20" class="edge">
<title>Node2->Node24</title>
-<path fill="none" stroke="#191970" d="M3064.5833,-151.371C3057.1748,-138.5354 3049.412,-125.0858 3042.2235,-112.6314"/>
-<polygon fill="none" stroke="#191970" points="3061.6139,-153.2281 3069.6442,-160.1394 3067.6766,-149.7289 3061.6139,-153.2281"/>
+<path fill="none" stroke="#191970" d="M3013,-149.8874C3013,-137.5023 3013,-124.6068 3013,-112.6314"/>
+<polygon fill="none" stroke="#191970" points="3009.5001,-150.1394 3013,-160.1394 3016.5001,-150.1394 3009.5001,-150.1394"/>
</g>
<!-- Node25 -->
<g id="node22" class="node">
<title>Node25</title>
-<g id="a_node22"><a xlink:href="classtvm_1_1tir_1_1Load.html" target="_top" xlink:title="Managed reference to LoadNode. ">
+<g id="a_node22"><a xlink:href="classtvm_1_1tir_1_1LT.html" target="_top" xlink:title="Managed reference to LTNode. ">
<polygon fill="#ffffff" stroke="#000000" points="3108,-11.5 3108,-112.5 3262,-112.5 3262,-11.5 3108,-11.5"/>
-<text text-anchor="middle" x="3185" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Load</text>
+<text text-anchor="middle" x="3185" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::LT</text>
<polyline fill="none" stroke="#000000" points="3108,-93.5 3262,-93.5 "/>
<text text-anchor="middle" x="3185" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="3108,-74.5 3262,-74.5 "/>
-<text text-anchor="start" x="3116" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Load()</text>
+<text text-anchor="start" x="3116" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ LT()</text>
<text text-anchor="start" x="3116" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="3116" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="3116" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -529,19 +529,19 @@
<!-- Node2->Node25 -->
<g id="edge21" class="edge">
<title>Node2->Node25</title>
-<path fill="none" stroke="#191970" d="M3133.4167,-151.371C3140.8252,-138.5354 3148.588,-125.0858 3155.7765,-112.6314"/>
-<polygon fill="none" stroke="#191970" points="3130.3234,-149.7289 3128.3558,-160.1394 3136.3861,-153.2281 3130.3234,-149.7289"/>
+<path fill="none" stroke="#191970" d="M3079.2764,-153.5862C3094.8552,-140.0906 3111.3466,-125.8044 3126.553,-112.6314"/>
+<polygon fill="none" stroke="#191970" points="3076.9783,-150.9463 3071.7116,-160.1394 3081.5616,-156.2372 3076.9783,-150.9463"/>
</g>
<!-- Node26 -->
<g id="node23" class="node">
<title>Node26</title>
-<g id="a_node23"><a xlink:href="classtvm_1_1tir_1_1LT.html" target="_top" xlink:title="Managed reference to LTNode. ">
+<g id="a_node23"><a xlink:href="classtvm_1_1tir_1_1Max.html" target="_top" xlink:title="Managed reference to MaxNode. ">
<polygon fill="#ffffff" stroke="#000000" points="3280,-11.5 3280,-112.5 3434,-112.5 3434,-11.5 3280,-11.5"/>
-<text text-anchor="middle" x="3357" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::LT</text>
+<text text-anchor="middle" x="3357" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Max</text>
<polyline fill="none" stroke="#000000" points="3280,-93.5 3434,-93.5 "/>
<text text-anchor="middle" x="3357" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="3280,-74.5 3434,-74.5 "/>
-<text text-anchor="start" x="3288" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ LT()</text>
+<text text-anchor="start" x="3288" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Max()</text>
<text text-anchor="start" x="3288" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="3288" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="3288" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -552,19 +552,19 @@
<!-- Node2->Node26 -->
<g id="edge22" class="edge">
<title>Node2->Node26</title>
-<path fill="none" stroke="#191970" d="M3185.4818,-170.5589C3213.2893,-156.6871 3243.9006,-140.4832 3271,-124 3276.8185,-120.4609 3282.7515,-116.6498 3288.6444,-112.719"/>
-<polygon fill="none" stroke="#191970" points="3183.5763,-167.5968 3176.168,-175.1709 3186.6827,-173.8698 3183.5763,-167.5968"/>
+<path fill="none" stroke="#191970" d="M3099.9618,-188.7385C3151.2693,-173.8962 3216.5987,-151.9726 3271,-124 3277.5476,-120.6333 3284.1486,-116.805 3290.6266,-112.7416"/>
+<polygon fill="none" stroke="#191970" points="3098.8775,-185.4081 3090.2231,-191.5197 3100.7998,-192.139 3098.8775,-185.4081"/>
</g>
<!-- Node27 -->
<g id="node24" class="node">
<title>Node27</title>
-<g id="a_node24"><a xlink:href="classtvm_1_1tir_1_1Max.html" target="_top" xlink:title="Managed reference to MaxNode. ">
+<g id="a_node24"><a xlink:href="classtvm_1_1tir_1_1Min.html" target="_top" xlink:title="Managed reference to MinNode. ">
<polygon fill="#ffffff" stroke="#000000" points="3452,-11.5 3452,-112.5 3606,-112.5 3606,-11.5 3452,-11.5"/>
-<text text-anchor="middle" x="3529" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Max</text>
+<text text-anchor="middle" x="3529" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Min</text>
<polyline fill="none" stroke="#000000" points="3452,-93.5 3606,-93.5 "/>
<text text-anchor="middle" x="3529" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="3452,-74.5 3606,-74.5 "/>
-<text text-anchor="start" x="3460" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Max()</text>
+<text text-anchor="start" x="3460" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Min()</text>
<text text-anchor="start" x="3460" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="3460" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="3460" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -575,19 +575,19 @@
<!-- Node2->Node27 -->
<g id="edge23" class="edge">
<title>Node2->Node27</title>
-<path fill="none" stroke="#191970" d="M3186.1495,-198.3094C3257.5913,-185.7375 3360.0768,-162.6518 3443,-124 3450.0068,-120.734 3457.0293,-116.8709 3463.8737,-112.6944"/>
-<polygon fill="none" stroke="#191970" points="3185.5384,-194.8631 3176.2801,-200.0142 3186.7299,-201.7609 3185.5384,-194.8631"/>
+<path fill="none" stroke="#191970" d="M3100.2064,-204.2501C3189.4577,-194.8783 3330.611,-173.133 3443,-124 3450.2519,-120.8297 3457.494,-116.9752 3464.5249,-112.7557"/>
+<polygon fill="none" stroke="#191970" points="3099.716,-200.7819 3090.1232,-205.2794 3100.4269,-207.7457 3099.716,-200.7819"/>
</g>
<!-- Node28 -->
<g id="node25" class="node">
<title>Node28</title>
-<g id="a_node25"><a xlink:href="classtvm_1_1tir_1_1Min.html" target="_top" xlink:title="Managed reference to MinNode. ">
+<g id="a_node25"><a xlink:href="classtvm_1_1tir_1_1Mod.html" target="_top" xlink:title="Managed reference to ModNode. ">
<polygon fill="#ffffff" stroke="#000000" points="3624,-11.5 3624,-112.5 3778,-112.5 3778,-11.5 3624,-11.5"/>
-<text text-anchor="middle" x="3701" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Min</text>
+<text text-anchor="middle" x="3701" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Mod</text>
<polyline fill="none" stroke="#000000" points="3624,-93.5 3778,-93.5 "/>
<text text-anchor="middle" x="3701" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="3624,-74.5 3778,-74.5 "/>
-<text text-anchor="start" x="3632" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Min()</text>
+<text text-anchor="start" x="3632" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Mod()</text>
<text text-anchor="start" x="3632" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="3632" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="3632" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -598,19 +598,19 @@
<!-- Node2->Node28 -->
<g id="edge24" class="edge">
<title>Node2->Node28</title>
-<path fill="none" stroke="#191970" d="M3186.1821,-201.1602C3312.3664,-186.0977 3539.1001,-155.7179 3615,-124 3622.4724,-120.8773 3629.9167,-116.9965 3637.1219,-112.7105"/>
-<polygon fill="none" stroke="#191970" points="3185.6745,-197.6958 3176.1566,-202.35 3186.4995,-204.6471 3185.6745,-197.6958"/>
+<path fill="none" stroke="#191970" d="M3100.3664,-203.5375C3244.0358,-190.4075 3523.3924,-161.0234 3615,-124 3622.6793,-120.8964 3630.3161,-116.9682 3637.6882,-112.6011"/>
+<polygon fill="none" stroke="#191970" points="3099.9038,-200.065 3090.2611,-204.4546 3100.5366,-207.0363 3099.9038,-200.065"/>
</g>
<!-- Node29 -->
<g id="node26" class="node">
<title>Node29</title>
-<g id="a_node26"><a xlink:href="classtvm_1_1tir_1_1Mod.html" target="_top" xlink:title="Managed reference to ModNode. ">
+<g id="a_node26"><a xlink:href="classtvm_1_1tir_1_1Mul.html" target="_top" xlink:title="Managed reference to MulNode. ">
<polygon fill="#ffffff" stroke="#000000" points="3796,-11.5 3796,-112.5 3950,-112.5 3950,-11.5 3796,-11.5"/>
-<text text-anchor="middle" x="3873" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Mod</text>
+<text text-anchor="middle" x="3873" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Mul</text>
<polyline fill="none" stroke="#000000" points="3796,-93.5 3950,-93.5 "/>
<text text-anchor="middle" x="3873" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="3796,-74.5 3950,-74.5 "/>
-<text text-anchor="start" x="3804" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Mod()</text>
+<text text-anchor="start" x="3804" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Mul()</text>
<text text-anchor="start" x="3804" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="3804" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="3804" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -621,19 +621,19 @@
<!-- Node2->Node29 -->
<g id="edge25" class="edge">
<title>Node2->Node29</title>
-<path fill="none" stroke="#191970" d="M3186.558,-205.3325C3346.5381,-194.0901 3679.5107,-166.3368 3787,-124 3794.7065,-120.9646 3802.361,-117.0807 3809.7437,-112.7402"/>
-<polygon fill="none" stroke="#191970" points="3186.0072,-201.8623 3176.2748,-206.0491 3186.4939,-208.8454 3186.0072,-201.8623"/>
+<path fill="none" stroke="#191970" d="M3100.3248,-206.7609C3275.4574,-197.3534 3663.3871,-171.7011 3787,-124 3794.8991,-120.9518 3802.7361,-116.9973 3810.2784,-112.5605"/>
+<polygon fill="none" stroke="#191970" points="3099.9158,-203.2776 3090.1158,-207.3038 3100.2876,-210.2678 3099.9158,-203.2776"/>
</g>
<!-- Node30 -->
<g id="node27" class="node">
<title>Node30</title>
-<g id="a_node27"><a xlink:href="classtvm_1_1tir_1_1Mul.html" target="_top" xlink:title="Managed reference to MulNode. ">
+<g id="a_node27"><a xlink:href="classtvm_1_1tir_1_1NE.html" target="_top" xlink:title="Managed reference to NENode. ">
<polygon fill="#ffffff" stroke="#000000" points="3968,-11.5 3968,-112.5 4122,-112.5 4122,-11.5 3968,-11.5"/>
-<text text-anchor="middle" x="4045" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Mul</text>
+<text text-anchor="middle" x="4045" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::NE</text>
<polyline fill="none" stroke="#000000" points="3968,-93.5 4122,-93.5 "/>
<text text-anchor="middle" x="4045" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="3968,-74.5 4122,-74.5 "/>
-<text text-anchor="start" x="3976" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Mul()</text>
+<text text-anchor="start" x="3976" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ NE()</text>
<text text-anchor="start" x="3976" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="3976" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="3976" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -644,19 +644,19 @@
<!-- Node2->Node30 -->
<g id="edge26" class="edge">
<title>Node2->Node30</title>
-<path fill="none" stroke="#191970" d="M3186.2066,-207.8997C3375.6885,-200.2417 3819.1652,-177.0701 3959,-124 3966.9159,-120.9958 3974.7637,-117.0693 3982.3124,-112.6492"/>
-<polygon fill="none" stroke="#191970" points="3185.8618,-204.4105 3176.0093,-208.3067 3186.141,-211.405 3185.8618,-204.4105"/>
+<path fill="none" stroke="#191970" d="M3100.0561,-208.8338C3303.1423,-202.8465 3802.8295,-182.4573 3959,-124 3966.9295,-121.0319 3974.786,-117.1286 3982.3398,-112.722"/>
+<polygon fill="none" stroke="#191970" points="3099.9216,-205.3361 3090.0274,-209.1248 3100.1248,-212.3331 3099.9216,-205.3361"/>
</g>
<!-- Node31 -->
<g id="node28" class="node">
<title>Node31</title>
-<g id="a_node28"><a xlink:href="classtvm_1_1tir_1_1NE.html" target="_top" xlink:title="Managed reference to NENode. ">
+<g id="a_node28"><a xlink:href="classtvm_1_1tir_1_1Not.html" target="_top" xlink:title="Managed reference to NotNode. ">
<polygon fill="#ffffff" stroke="#000000" points="4140,-11.5 4140,-112.5 4294,-112.5 4294,-11.5 4140,-11.5"/>
-<text text-anchor="middle" x="4217" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::NE</text>
+<text text-anchor="middle" x="4217" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Not</text>
<polyline fill="none" stroke="#000000" points="4140,-93.5 4294,-93.5 "/>
<text text-anchor="middle" x="4217" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="4140,-74.5 4294,-74.5 "/>
-<text text-anchor="start" x="4148" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ NE()</text>
+<text text-anchor="start" x="4148" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Not()</text>
<text text-anchor="start" x="4148" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="4148" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="4148" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -667,19 +667,19 @@
<!-- Node2->Node31 -->
<g id="edge27" class="edge">
<title>Node2->Node31</title>
-<path fill="none" stroke="#191970" d="M3186.1546,-209.6075C3402.3898,-205.2061 3958.4685,-187.8346 4131,-124 4138.9407,-121.062 4146.8045,-117.1781 4154.3625,-112.7829"/>
-<polygon fill="none" stroke="#191970" points="3186.0457,-206.1089 3176.1174,-209.8074 3186.1852,-213.1075 3186.0457,-206.1089"/>
+<path fill="none" stroke="#191970" d="M3100.4041,-210.2633C3329.3861,-207.3744 3942.0736,-193.2094 4131,-124 4139.123,-121.0243 4147.162,-117.0515 4154.8742,-112.5479"/>
+<polygon fill="none" stroke="#191970" points="3100.0776,-206.767 3090.121,-210.3887 3100.163,-213.7664 3100.0776,-206.767"/>
</g>
<!-- Node32 -->
<g id="node29" class="node">
<title>Node32</title>
-<g id="a_node29"><a xlink:href="classtvm_1_1tir_1_1Not.html" target="_top" xlink:title="Managed reference to NotNode. ">
+<g id="a_node29"><a xlink:href="classtvm_1_1tir_1_1Or.html" target="_top" xlink:title="Managed reference to OrNode. ">
<polygon fill="#ffffff" stroke="#000000" points="4312,-11.5 4312,-112.5 4466,-112.5 4466,-11.5 4312,-11.5"/>
-<text text-anchor="middle" x="4389" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Not</text>
+<text text-anchor="middle" x="4389" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Or</text>
<polyline fill="none" stroke="#000000" points="4312,-93.5 4466,-93.5 "/>
<text text-anchor="middle" x="4389" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="4312,-74.5 4466,-74.5 "/>
-<text text-anchor="start" x="4320" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Not()</text>
+<text text-anchor="start" x="4320" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Or()</text>
<text text-anchor="start" x="4320" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="4320" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="4320" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -690,19 +690,19 @@
<!-- Node2->Node32 -->
<g id="edge28" class="edge">
<title>Node2->Node32</title>
-<path fill="none" stroke="#191970" d="M3186.3258,-207.4451C3445.5194,-196.5097 4198.8821,-161.8155 4303,-124 4311.1311,-121.0468 4319.1754,-117.0881 4326.8906,-112.5928"/>
-<polygon fill="none" stroke="#191970" points="3186.049,-203.9535 3176.2048,-207.8706 3186.3431,-210.9473 3186.049,-203.9535"/>
+<path fill="none" stroke="#191970" d="M3100.4002,-207.8933C3372.1049,-197.8488 4190.5486,-164.5374 4303,-124 4311.1382,-121.0663 4319.1869,-117.12 4326.9047,-112.6318"/>
+<polygon fill="none" stroke="#191970" points="3100.0572,-204.4035 3090.1928,-208.2692 3100.3149,-211.3988 3100.0572,-204.4035"/>
</g>
<!-- Node33 -->
<g id="node30" class="node">
<title>Node33</title>
-<g id="a_node30"><a xlink:href="classtvm_1_1tir_1_1Or.html" target="_top" xlink:title="Managed reference to OrNode. ">
+<g id="a_node30"><a xlink:href="classtvm_1_1tir_1_1ProducerLoad.html" target="_top" xlink:title="Managed reference to ProducerLoadNode. ">
<polygon fill="#ffffff" stroke="#000000" points="4484,-11.5 4484,-112.5 4638,-112.5 4638,-11.5 4484,-11.5"/>
-<text text-anchor="middle" x="4561" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Or</text>
+<text text-anchor="middle" x="4561" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::ProducerLoad</text>
<polyline fill="none" stroke="#000000" points="4484,-93.5 4638,-93.5 "/>
<text text-anchor="middle" x="4561" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="4484,-74.5 4638,-74.5 "/>
-<text text-anchor="start" x="4492" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Or()</text>
+<text text-anchor="start" x="4492" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ProducerLoad()</text>
<text text-anchor="start" x="4492" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="4492" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="4492" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -713,19 +713,19 @@
<!-- Node2->Node33 -->
<g id="edge29" class="edge">
<title>Node2->Node33</title>
-<path fill="none" stroke="#191970" d="M3186.3314,-208.2909C3469.9943,-199.1028 4354.1698,-167.2716 4475,-124 4483.1443,-121.0834 4491.197,-117.1479 4498.917,-112.6659"/>
-<polygon fill="none" stroke="#191970" points="3185.9746,-204.8005 3176.0927,-208.6212 3186.2004,-211.7969 3185.9746,-204.8005"/>
+<path fill="none" stroke="#191970" d="M3100.7133,-208.629C3396.4689,-200.2418 4345.8328,-169.9874 4475,-124 4483.1497,-121.0985 4491.2058,-117.1725 4498.9278,-112.6961"/>
+<polygon fill="none" stroke="#191970" points="3100.3678,-205.1372 3090.4705,-208.9181 3100.5654,-212.1344 3100.3678,-205.1372"/>
</g>
<!-- Node34 -->
<g id="node31" class="node">
<title>Node34</title>
-<g id="a_node31"><a xlink:href="classtvm_1_1tir_1_1ProducerLoad.html" target="_top" xlink:title="Managed reference to ProducerLoadNode. ">
+<g id="a_node31"><a xlink:href="classtvm_1_1tir_1_1Ramp.html" target="_top" xlink:title="Managed reference to RampNode. ">
<polygon fill="#ffffff" stroke="#000000" points="4656,-11.5 4656,-112.5 4810,-112.5 4810,-11.5 4656,-11.5"/>
-<text text-anchor="middle" x="4733" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::ProducerLoad</text>
+<text text-anchor="middle" x="4733" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Ramp</text>
<polyline fill="none" stroke="#000000" points="4656,-93.5 4810,-93.5 "/>
<text text-anchor="middle" x="4733" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="4656,-74.5 4810,-74.5 "/>
-<text text-anchor="start" x="4664" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ProducerLoad()</text>
+<text text-anchor="start" x="4664" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Ramp()</text>
<text text-anchor="start" x="4664" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="4664" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="4664" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -736,19 +736,19 @@
<!-- Node2->Node34 -->
<g id="edge30" class="edge">
<title>Node2->Node34</title>
-<path fill="none" stroke="#191970" d="M3186.3644,-208.9493C3493.0414,-201.3583 4509.3684,-172.7456 4647,-124 4655.1545,-121.1119 4663.2135,-117.1944 4670.9373,-112.723"/>
-<polygon fill="none" stroke="#191970" points="3186.1024,-205.4546 3176.1916,-209.1998 3186.2748,-212.4525 3186.1024,-205.4546"/>
+<path fill="none" stroke="#191970" d="M3100.0947,-209.2331C3417.4372,-202.4024 4500.8942,-175.5046 4647,-124 4655.1588,-121.1239 4663.2205,-117.2141 4670.9458,-112.747"/>
+<polygon fill="none" stroke="#191970" points="3099.9528,-205.7353 3090.03,-209.4485 3100.1026,-212.7337 3099.9528,-205.7353"/>
</g>
<!-- Node35 -->
<g id="node32" class="node">
<title>Node35</title>
-<g id="a_node32"><a xlink:href="classtvm_1_1tir_1_1Ramp.html" target="_top" xlink:title="Managed reference to RampNode. ">
+<g id="a_node32"><a xlink:href="classtvm_1_1tir_1_1Reduce.html" target="_top" xlink:title="Managed reference to ReduceNode. ">
<polygon fill="#ffffff" stroke="#000000" points="4828,-11.5 4828,-112.5 4982,-112.5 4982,-11.5 4828,-11.5"/>
-<text text-anchor="middle" x="4905" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Ramp</text>
+<text text-anchor="middle" x="4905" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Reduce</text>
<polyline fill="none" stroke="#000000" points="4828,-93.5 4982,-93.5 "/>
<text text-anchor="middle" x="4905" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="4828,-74.5 4982,-74.5 "/>
-<text text-anchor="start" x="4836" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Ramp()</text>
+<text text-anchor="start" x="4836" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Reduce()</text>
<text text-anchor="start" x="4836" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="4836" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="4836" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -759,19 +759,19 @@
<!-- Node2->Node35 -->
<g id="edge31" class="edge">
<title>Node2->Node35</title>
-<path fill="none" stroke="#191970" d="M3186.4586,-209.4759C3514.9929,-203.355 4664.4996,-178.2333 4819,-124 4827.1626,-121.1347 4835.2267,-117.2318 4842.9534,-112.7687"/>
-<polygon fill="none" stroke="#191970" points="3185.9823,-205.9841 3176.0488,-209.6686 3186.1119,-212.9829 3185.9823,-205.9841"/>
+<path fill="none" stroke="#191970" d="M3100.2778,-209.7058C3439.0231,-204.2863 4656.004,-180.9957 4819,-124 4827.3397,-121.0838 4835.5755,-117.0815 4843.4531,-112.5029"/>
+<polygon fill="none" stroke="#191970" points="3099.9775,-206.21 3090.0344,-209.8685 3100.0887,-213.2091 3099.9775,-206.21"/>
</g>
<!-- Node36 -->
<g id="node33" class="node">
<title>Node36</title>
-<g id="a_node33"><a xlink:href="classtvm_1_1tir_1_1Reduce.html" target="_top" xlink:title="Managed reference to ReduceNode. ">
+<g id="a_node33"><a xlink:href="classtvm_1_1tir_1_1Select.html" target="_top" xlink:title="Managed reference to SelectNode. ">
<polygon fill="#ffffff" stroke="#000000" points="5000,-11.5 5000,-112.5 5154,-112.5 5154,-11.5 5000,-11.5"/>
-<text text-anchor="middle" x="5077" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Reduce</text>
+<text text-anchor="middle" x="5077" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Select</text>
<polyline fill="none" stroke="#000000" points="5000,-93.5 5154,-93.5 "/>
<text text-anchor="middle" x="5077" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="5000,-74.5 5154,-74.5 "/>
-<text text-anchor="start" x="5008" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Reduce()</text>
+<text text-anchor="start" x="5008" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Select()</text>
<text text-anchor="start" x="5008" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="5008" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="5008" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -782,67 +782,67 @@
<!-- Node2->Node36 -->
<g id="edge32" class="edge">
<title>Node2->Node36</title>
-<path fill="none" stroke="#191970" d="M3186.0607,-209.9147C3534.6962,-205.1724 4819.4859,-183.7641 4991,-124 4999.3429,-121.0929 5007.5807,-117.0963 5015.4594,-112.5209"/>
-<polygon fill="none" stroke="#191970" points="3185.973,-206.4155 3176.0212,-210.0502 3186.0675,-213.4149 3185.973,-206.4155"/>
+<path fill="none" stroke="#191970" d="M3100.108,-210.1019C3458.7786,-206.0065 4810.9964,-186.5209 4991,-124 4999.3458,-121.1012 5007.5854,-117.1098 5015.4652,-112.5374"/>
+<polygon fill="none" stroke="#191970" points="3100.0013,-206.6028 3090.0416,-210.2158 3100.0806,-213.6024 3100.0013,-206.6028"/>
</g>
<!-- Node37 -->
<g id="node34" class="node">
<title>Node37</title>
-<g id="a_node34"><a xlink:href="classtvm_1_1tir_1_1Select.html" target="_top" xlink:title="Managed reference to SelectNode. ">
-<polygon fill="#ffffff" stroke="#000000" points="5172,-11.5 5172,-112.5 5326,-112.5 5326,-11.5 5172,-11.5"/>
-<text text-anchor="middle" x="5249" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Select</text>
-<polyline fill="none" stroke="#000000" points="5172,-93.5 5326,-93.5 "/>
-<text text-anchor="middle" x="5249" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
-<polyline fill="none" stroke="#000000" points="5172,-74.5 5326,-74.5 "/>
-<text text-anchor="start" x="5180" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Select()</text>
-<text text-anchor="start" x="5180" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
-<text text-anchor="start" x="5180" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
-<text text-anchor="start" x="5180" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
-<text text-anchor="start" x="5180" y="-18.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_COW_METHOD()</text>
+<g id="a_node34"><a xlink:href="classtvm_1_1tir_1_1Shuffle.html" target="_top" xlink:title="Managed reference to ShuffleNode. ">
+<polygon fill="#ffffff" stroke="#000000" points="5172,-.5 5172,-123.5 5326,-123.5 5326,-.5 5172,-.5"/>
+<text text-anchor="middle" x="5249" y="-111.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Shuffle</text>
+<polyline fill="none" stroke="#000000" points="5172,-104.5 5326,-104.5 "/>
+<text text-anchor="middle" x="5249" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
+<polyline fill="none" stroke="#000000" points="5172,-85.5 5326,-85.5 "/>
+<text text-anchor="start" x="5180" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Shuffle()</text>
+<text text-anchor="start" x="5180" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
+<text text-anchor="start" x="5180" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
+<text text-anchor="start" x="5180" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
+<text text-anchor="start" x="5180" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_COW_METHOD()</text>
+<text text-anchor="start" x="5180" y="-18.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Concat()</text>
+<text text-anchor="start" x="5180" y="-7.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ExtractElement()</text>
</a>
</g>
</g>
<!-- Node2->Node37 -->
<g id="edge33" class="edge">
<title>Node2->Node37</title>
-<path fill="none" stroke="#191970" d="M3186.2017,-210.2733C3554.7711,-206.8031 4974.5019,-189.2781 5163,-124 5171.3485,-121.1089 5179.5897,-117.1222 5187.4704,-112.5525"/>
-<polygon fill="none" stroke="#191970" points="3186.0929,-206.7741 3176.126,-210.3672 3186.1582,-213.7738 3186.0929,-206.7741"/>
+<path fill="none" stroke="#191970" d="M3100.0581,-210.4332C3477.95,-207.5738 4965.9545,-192.0525 5163,-124 5165.9576,-122.9785 5168.9017,-121.8193 5171.8209,-120.5452"/>
+<polygon fill="none" stroke="#191970" points="3099.9761,-206.9336 3090.0026,-210.5083 3100.0284,-213.9334 3099.9761,-206.9336"/>
</g>
<!-- Node38 -->
<g id="node35" class="node">
<title>Node38</title>
-<g id="a_node35"><a xlink:href="classtvm_1_1tir_1_1Shuffle.html" target="_top" xlink:title="Managed reference to ShuffleNode. ">
-<polygon fill="#ffffff" stroke="#000000" points="5344,-.5 5344,-123.5 5498,-123.5 5498,-.5 5344,-.5"/>
-<text text-anchor="middle" x="5421" y="-111.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Shuffle</text>
-<polyline fill="none" stroke="#000000" points="5344,-104.5 5498,-104.5 "/>
-<text text-anchor="middle" x="5421" y="-92.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
-<polyline fill="none" stroke="#000000" points="5344,-85.5 5498,-85.5 "/>
-<text text-anchor="start" x="5352" y="-73.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Shuffle()</text>
-<text text-anchor="start" x="5352" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
-<text text-anchor="start" x="5352" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
-<text text-anchor="start" x="5352" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
-<text text-anchor="start" x="5352" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_COW_METHOD()</text>
-<text text-anchor="start" x="5352" y="-18.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Concat()</text>
-<text text-anchor="start" x="5352" y="-7.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ ExtractElement()</text>
+<g id="a_node35"><a xlink:href="classtvm_1_1tir_1_1StringImm.html" target="_top" xlink:title="Managed reference to StringImmNode. ">
+<polygon fill="#ffffff" stroke="#000000" points="5344,-11.5 5344,-112.5 5498,-112.5 5498,-11.5 5344,-11.5"/>
+<text text-anchor="middle" x="5421" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::StringImm</text>
+<polyline fill="none" stroke="#000000" points="5344,-93.5 5498,-93.5 "/>
+<text text-anchor="middle" x="5421" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
+<polyline fill="none" stroke="#000000" points="5344,-74.5 5498,-74.5 "/>
+<text text-anchor="start" x="5352" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ StringImm()</text>
+<text text-anchor="start" x="5352" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
+<text text-anchor="start" x="5352" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
+<text text-anchor="start" x="5352" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
+<text text-anchor="start" x="5352" y="-18.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_COW_METHOD()</text>
</a>
</g>
</g>
<!-- Node2->Node38 -->
<g id="edge34" class="edge">
<title>Node2->Node38</title>
-<path fill="none" stroke="#191970" d="M3186.3291,-210.578C3574.0338,-208.3035 5129.468,-194.8048 5335,-124 5337.9584,-122.9808 5340.9031,-121.8236 5343.823,-120.5512"/>
-<polygon fill="none" stroke="#191970" points="3186.2725,-207.0782 3176.2929,-210.6359 3186.313,-214.0781 3186.2725,-207.0782"/>
+<path fill="none" stroke="#191970" d="M3100.3708,-210.7139C3497.3421,-209.0119 5120.9309,-197.5735 5335,-124 5343.3552,-121.1284 5351.6006,-117.154 5359.4837,-112.5912"/>
+<polygon fill="none" stroke="#191970" points="3100.0874,-207.215 3090.1022,-210.7569 3100.1168,-214.2149 3100.0874,-207.215"/>
</g>
<!-- Node39 -->
<g id="node36" class="node">
<title>Node39</title>
-<g id="a_node36"><a xlink:href="classtvm_1_1tir_1_1StringImm.html" target="_top" xlink:title="Managed reference to StringImmNode. ">
+<g id="a_node36"><a xlink:href="classtvm_1_1tir_1_1Sub.html" target="_top" xlink:title="Managed reference to SubNode. ">
<polygon fill="#ffffff" stroke="#000000" points="5516,-11.5 5516,-112.5 5670,-112.5 5670,-11.5 5516,-11.5"/>
-<text text-anchor="middle" x="5593" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::StringImm</text>
+<text text-anchor="middle" x="5593" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Sub</text>
<polyline fill="none" stroke="#000000" points="5516,-93.5 5670,-93.5 "/>
<text text-anchor="middle" x="5593" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
<polyline fill="none" stroke="#000000" points="5516,-74.5 5670,-74.5 "/>
-<text text-anchor="start" x="5524" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ StringImm()</text>
+<text text-anchor="start" x="5524" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Sub()</text>
<text text-anchor="start" x="5524" y="-51.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
<text text-anchor="start" x="5524" y="-40.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">_METHODS()</text>
<text text-anchor="start" x="5524" y="-29.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ TVM_DEFINE_OBJECT_REF</text>
@@ -853,176 +853,176 @@
<!-- Node2->Node39 -->
<g id="edge35" class="edge">
<title>Node2->Node39</title>
-<path fill="none" stroke="#191970" d="M3186.2607,-210.8407C3592.0634,-209.6975 5284.356,-200.3543 5507,-124 5515.3571,-121.134 5523.6038,-117.163 5531.4875,-112.6023"/>
-<polygon fill="none" stroke="#191970" points="3186.0472,-207.3411 3176.0568,-210.8684 3186.0663,-214.3411 3186.0472,-207.3411"/>
+<path fill="none" stroke="#191970" d="M3100.25,-209.3799C3528.3673,-201.2497 5390.9589,-163.7149 5507,-124 5515.3589,-121.1392 5523.6067,-117.1715 5531.4911,-112.6126"/>
+<polygon fill="none" stroke="#191970" points="3100.0358,-205.8832 3090.104,-209.5721 3100.1685,-212.8819 3100.0358,-205.8832"/>
</g>
<!-- Node40 -->
<g id="node37" class="node">
<title>Node40</title>
-<g id="a_node37"><a xlink:href="classtvm_1_1tir_1_1Sub.html" target="_top" xlink:title="Managed reference to SubNode. ">
-<polygon fill="#ffffff" stroke="#000000" points="5688,-11.5 5688,-112.5 5842,-112.5 5842,-11.5 5688,-11.5"/>
-<text text-anchor="middle" x="5765" y="-100.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">tvm::tir::Sub</text>
-<polyline fill="none" stroke="#000000" points="5688,-93.5 5842,-93.5 "/>
-<text text-anchor="middle" x="5765" y="-81.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000"> </text>
-<polyline fill="none" stroke="#000000" points="5688,-74.5 5842,-74.5 "/>
-<text text-anchor="start" x="5696" y="-62.5" font-family="Helvetica,sans-Serif" font-size="10.00" fill="#000000">+ Sub()</text>
... 42350 lines suppressed ...