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 2023/01/11 10:14:24 UTC
[tvm-site] branch asf-site updated: deploying docs (apache/tvm@15e185d922b4de567aa2f74c71aedbc0b56952df)
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 b097f4ae0a deploying docs (apache/tvm@15e185d922b4de567aa2f74c71aedbc0b56952df)
b097f4ae0a is described below
commit b097f4ae0afa9e8ab1b13c51338c8c5488f79864
Author: tvm-bot <95...@users.noreply.github.com>
AuthorDate: Wed Jan 11 10:14:18 2023 +0000
deploying docs (apache/tvm@15e185d922b4de567aa2f74c71aedbc0b56952df)
---
docs/_images/sphx_glr_micro_train_001.png | Bin 332672 -> 302170 bytes
docs/_images/sphx_glr_micro_train_thumb.png | Bin 24174 -> 22480 bytes
.../how_to/compile_models/from_darknet.rst.txt | 2 +-
.../how_to/compile_models/from_keras.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 | 7 +-
.../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 | 20 +-
.../extend_tvm/bring_your_own_datatypes.rst.txt | 2 +-
.../how_to/extend_tvm/sg_execution_times.rst.txt | 8 +-
.../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 | 1912 +++++++++++++++-----
.../tune_network_cuda.rst.txt | 4 +-
.../tune_network_x86.rst.txt | 4 +-
.../tune_sparse_x86.rst.txt | 31 +-
.../tune_with_autotvm/sg_execution_times.rst.txt | 6 +-
.../tune_with_autotvm/tune_conv2d_cuda.rst.txt | 828 ++++++++-
.../work_with_microtvm/micro_autotune.rst.txt | 16 +-
.../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 | 12 +-
.../work_with_relay/sg_execution_times.rst.txt | 8 +-
.../how_to/work_with_schedules/intrin_math.rst.txt | 2 +-
.../work_with_schedules/sg_execution_times.rst.txt | 10 +-
.../how_to/work_with_schedules/tensorize.rst.txt | 2 +-
.../tutorials/autotvm/sg_execution_times.rst.txt | 4 +-
.../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 | 55 +-
.../tutorial/cross_compilation_and_rpc.rst.txt | 2 +-
docs/_sources/tutorial/intro_topi.rst.txt | 2 +-
docs/_sources/tutorial/sg_execution_times.rst.txt | 24 +-
.../tutorial/tensor_expr_get_started.rst.txt | 42 +-
docs/commit_hash | 2 +-
docs/genindex.html | 4 +-
docs/how_to/compile_models/from_darknet.html | 2 +-
docs/how_to/compile_models/from_keras.html | 2 +-
docs/how_to/compile_models/from_mxnet.html | 2 +-
docs/how_to/compile_models/from_oneflow.html | 14 +-
docs/how_to/compile_models/from_pytorch.html | 10 +-
docs/how_to/compile_models/from_tensorflow.html | 2 +-
docs/how_to/compile_models/sg_execution_times.html | 26 +-
.../deploy_models/deploy_model_on_adreno.html | 3 +-
.../deploy_models/deploy_model_on_android.html | 2 +-
.../deploy_object_detection_pytorch.html | 43 +-
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 | 37 +-
docs/how_to/deploy_models/sg_execution_times.html | 20 +-
.../extend_tvm/bring_your_own_datatypes.html | 2 +-
docs/how_to/extend_tvm/sg_execution_times.html | 8 +-
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 | 1912 +++++++++++++++-----
.../tune_with_autoscheduler/tune_network_cuda.html | 4 +-
.../tune_with_autoscheduler/tune_network_x86.html | 4 +-
.../tune_with_autoscheduler/tune_sparse_x86.html | 31 +-
.../tune_with_autotvm/sg_execution_times.html | 6 +-
.../how_to/tune_with_autotvm/tune_conv2d_cuda.html | 828 ++++++++-
docs/how_to/work_with_microtvm/micro_autotune.html | 16 +-
docs/how_to/work_with_microtvm/micro_pytorch.html | 5 +-
docs/how_to/work_with_microtvm/micro_train.html | 16 +-
.../work_with_microtvm/sg_execution_times.html | 12 +-
.../how_to/work_with_relay/sg_execution_times.html | 8 +-
docs/how_to/work_with_schedules/intrin_math.html | 2 +-
.../work_with_schedules/sg_execution_times.html | 10 +-
docs/how_to/work_with_schedules/tensorize.html | 2 +-
docs/objects.inv | Bin 24146 -> 24151 bytes
.../api/doxygen/namespacemembers_func_l.html | 13 +-
docs/reference/api/doxygen/namespacemembers_l.html | 13 +-
docs/reference/api/doxygen/namespacemembers_m.html | 13 +-
.../api/doxygen/namespacemembers_vars.html | 3 +
docs/reference/api/doxygen/namespaces.html | 73 +-
.../api/doxygen/namespacetvm_1_1relay.html | 2 +
.../doxygen/namespacetvm_1_1relay_1_1legalize.html | 114 ++
.../api/doxygen/namespacetvm_1_1tir_1_1attr.html | 19 +
.../reference/api/doxygen/relay_2transform_8h.html | 4 +
.../api/doxygen/relay_2transform_8h_source.html | 2 +-
docs/reference/api/doxygen/search/all_15.js | 1 +
docs/reference/api/doxygen/search/all_d.js | 2 +-
docs/reference/api/doxygen/search/all_e.js | 1 +
docs/reference/api/doxygen/search/functions_c.js | 2 +-
docs/reference/api/doxygen/search/namespaces_1.js | 1 +
docs/reference/api/doxygen/search/variables_c.js | 1 +
docs/reference/api/doxygen/stmt_8h.html | 3 +
docs/reference/api/doxygen/stmt_8h_source.html | 7 +-
docs/reference/api/python/auto_scheduler.html | 4 +-
docs/reference/api/python/topi.html | 74 +-
.../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 +-
.../api/typedoc/classes/graphexecutor.html | 16 +-
docs/reference/api/typedoc/classes/instance.html | 40 +-
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 +-
docs/reference/api/typedoc/classes/scalar.html | 6 +-
.../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 | 112 +-
.../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 | 4 +-
.../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 | 275 +--
docs/tutorial/cross_compilation_and_rpc.html | 2 +-
docs/tutorial/intro_topi.html | 2 +-
docs/tutorial/sg_execution_times.html | 26 +-
docs/tutorial/tensor_expr_get_started.html | 42 +-
150 files changed, 5635 insertions(+), 1863 deletions(-)
diff --git a/docs/_images/sphx_glr_micro_train_001.png b/docs/_images/sphx_glr_micro_train_001.png
index 3f59a37dab..2deab86823 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 6746d44e45..9d8a85810f 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 6d5c174619..17c98c7063 100644
--- a/docs/_sources/how_to/compile_models/from_darknet.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_darknet.rst.txt
@@ -319,7 +319,7 @@ The process is no different from other examples.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 9.382 seconds)
+ **Total running time of the script:** ( 1 minutes 9.878 seconds)
.. _sphx_glr_download_how_to_compile_models_from_darknet.py:
diff --git a/docs/_sources/how_to/compile_models/from_keras.rst.txt b/docs/_sources/how_to/compile_models/from_keras.rst.txt
index 088c9794bf..49c8d14759 100644
--- a/docs/_sources/how_to/compile_models/from_keras.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_keras.rst.txt
@@ -232,7 +232,7 @@ Look up prediction top 1 index in 1000 class synset.
.. code-block:: none
Relay top-1 id: 285, class name: Egyptian cat
-
1/1 [==============================] - ETA: 0s
1/1 [==============================] - 1s 902ms/step
+
1/1 [==============================] - ETA: 0s
1/1 [==============================] - 1s 926ms/step
Keras top-1 id: 285, class name: Egyptian cat
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 5a3486c695..52d3562608 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.zip92ab5ae2-94bf-44fe-a1cc-83745f42673d from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/resnet18_v1-a0666292.zip...
+ Downloading /workspace/.mxnet/models/resnet18_v1-a0666292.zipa59764d6-5791-488f-843d-46863756023f 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 8cacc98f90..98ae1b2aa7 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]
19%|#9 | 7.99M/41.5M [00:00<00:00, 51.8MB/s]
35%|###4 | 14.3M/41.5M [00:00<00:00, 55.7MB/s]
48%|####7 | 19.7M/41.5M [00:00<00:00, 47.6MB/s]
59%|#####8 | 24.3M/41.5M [00:00<00:00, 37.4MB/s]
77%|#######7 | 32.0M/41.5M [00:00<00:00, 45.9MB/s]
92%|#########2| 38.3M/41.5M [00:00<00:00, 46.0MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 47.1MB/s]
+
0%| | 0.00/41.5M [00:00<?, ?B/s]
19%|#9 | 7.99M/41.5M [00:00<00:00, 45.2MB/s]
35%|###5 | 14.6M/41.5M [00:00<00:00, 55.9MB/s]
49%|####9 | 20.3M/41.5M [00:00<00:00, 53.3MB/s]
62%|######1 | 25.6M/41.5M [00:00<00:00, 45.2MB/s]
82%|########2 | 34.1M/41.5M [00:00<00:00, 48.4MB/s]
96%|#########6| 40.0M/41.5M [00:00<00:00, 48.8MB/s]
100%|##########| 41.5M/41.5M [00:00<00:00, 50.4MB/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 333f05c9b0..3c7195a052 100644
--- a/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_pytorch.rst.txt
@@ -102,7 +102,7 @@ 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]
32%|###1 | 14.2M/44.7M [00:00<00:00, 149MB/s]
64%|######3 | 28.4M/44.7M [00:00<00:00, 115MB/s]
89%|########9 | 39.9M/44.7M [00:00<00:00, 109MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 112MB/s]
+
0%| | 0.00/44.7M [00:00<?, ?B/s]
18%|#7 | 7.99M/44.7M [00:00<00:00, 72.4MB/s]
36%|###5 | 16.0M/44.7M [00:00<00:00, 68.9MB/s]
54%|#####3 | 24.0M/44.7M [00:00<00:00, 71.6MB/s]
72%|#######1 | 32.0M/44.7M [00:00<00:00, 67.2MB/s]
90%|########9 | 40.0M/44.7M [00:00<00:00, 62.6MB/s]
100%|##########| 44.7M/44.7M [00:00<00:00, 70.2MB/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 db8953bae8..f8e4f1e609 100644
--- a/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
+++ b/docs/_sources/how_to/compile_models/from_tensorflow.rst.txt
@@ -425,7 +425,7 @@ Run the corresponding model on tensorflow
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 10.094 seconds)
+ **Total running time of the script:** ( 1 minutes 9.853 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 c61f400a03..80349d0207 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
=================
-**05:34.227** total execution time for **how_to_compile_models** files:
+**05:38.333** total execution time for **how_to_compile_models** files:
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:10.094 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:09.878 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_darknet.py` (``from_darknet.py``) | 01:09.382 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tensorflow.py` (``from_tensorflow.py``) | 01:09.853 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:45.021 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_paddle.py` (``from_paddle.py``) | 00:46.139 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:30.859 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_oneflow.py` (``from_oneflow.py``) | 00:30.729 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:27.563 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_mxnet.py` (``from_mxnet.py``) | 00:28.029 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:25.941 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_coreml.py` (``from_coreml.py``) | 00:25.966 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:24.930 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_tflite.py` (``from_tflite.py``) | 00:25.881 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:21.729 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_pytorch.py` (``from_pytorch.py``) | 00:22.036 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:16.284 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_keras.py` (``from_keras.py``) | 00:17.415 | 0.0 MB |
+-----------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.424 | 0.0 MB |
+| :ref:`sphx_glr_how_to_compile_models_from_onnx.py` (``from_onnx.py``) | 00:02.408 | 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 5488731829..853a1a5003 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
@@ -728,18 +728,13 @@ 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)
- 3342.9183 3342.3090 3350.8926 3339.9926 3.0670
+ 2541.4099 2539.7964 2553.2709 2539.0355 4.0725
-.. rst-class:: sphx-glr-timing
-
- **Total running time of the script:** ( 1 minutes 0.711 seconds)
-
-
.. _sphx_glr_download_how_to_deploy_models_deploy_model_on_adreno.py:
.. only:: html
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 1eec79d7f2..4120c3d1e6 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)
- 15.6403 15.5040 16.0189 15.4519 0.2030
+ 15.7695 15.6253 16.5552 15.4373 0.3572
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 0b87218991..b49f427858 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
@@ -131,7 +131,7 @@ 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]
6%|5 | 10.1M/170M [00:00<00:01, 105MB/s]
12%|#1 | 20.0M/170M [00:00<00:02, 63.0MB/s]
19%|#8 | 32.0M/170M [00:00<00:02, 71.2MB/s]
28%|##8 | 48.0M/170M [00:00<00:01, 94.2MB/s]
36%|###6 | 61.9M/170M [00:00<00:01, 109MB/s]
43%|####3 | 73.1M/170M [00:00<00:01, 97.0MB/s]
52%|#####1 | 87.8M/170M [00:00<00:00, 112MB/s]
58%|#####8 | 99.3M/170M [00:01<00:00, 94.4MB/s]
64%|######4 | 109M/170M [00:01<00:00, 88.0MB/s]
70%|####### | 119M/170M [00:01<00:00, 92.1MB/s]
76%|#######5 | 129M/170M [00:01<00:00, 86.4MB/s]
83%|########3 | 141M/170M [00:01<00:00, 97.6MB/s]
89%|########8 | 151M/170M [00:01<00:00, 98.6MB/s]
95%|#########4| 161M/170M [00:01<00:00, 91.1MB/s]
100%|##########| 170M/170M [00:01<00:00, 95.2MB/s]
+
0%| | 0.00/170M [00:00<?, ?B/s]
4%|3 | 6.30M/170M [00:00<00:03, 46.7MB/s]
7%|7 | 12.7M/170M [00:00<00:02, 56.9MB/s]
11%|# | 18.3M/170M [00:00<00:03, 42.2MB/s]
14%|#4 | 24.0M/170M [00:00<00:03, 41.8MB/s]
19%|#8 | 32.0M/170M [00:00<00:02, 51.4MB/s]
24%|##3 | 40.0M/170M [00:00<00:02, 60.1MB/s]
28%|##8 | 48.0M/170M [00:00<00:02, 59.9MB/s]
35%|###4 | 58.9M/170M [00:01<00:01, 74.4MB/s]
39%|###9 | 66.5M/170M [00:01<00:01, 64.2MB/s]
43%|####3 | 73.1M/170M [00:01<00:01, 58.9MB/s]
47%|####7 | 80.0M/170M [00:01<00:01, 55.3MB/s]
51%|##### | 86.3M/170M [00:01<00:01, 54.8MB/s]
54%|#####4 | 91.7M/170M [00:01<00:01, 50.1MB/s]
57%|#####6 | 96.7M/170M [00:01<00:01, 49.9MB/s]
60%|###### | 102M/170M [00:02<00:01, 47.7MB/s]
63%|######2 | 107M/170M [00:02<00:01, 40.7MB/s]
66%|######6 | 112M/170M [00:02<00:01, 44.4MB/s]
71%|####### | 120M/170M [00:02<00:01, 47.5MB/s]
75%|#######5 | 128M/170M [00:02<00:00, 47.4MB/s]
80%|######## | 136M/170M [00:02<00:00, 48.3MB/s]
85%|########4 | 144M/170M [00:02<00:00, 55.5MB/s]
88%|########8 | 150M/170M [00:03<00:00, 45.9MB/s]
91%|#########1| 155M/170M [00:03<00:00, 45.4MB/s]
94%|#########4| 160M/170M [00:03<00:00, 41.6MB/s]
98%|#########7| 166M/170M [00:03<00:00, 39.4MB/s]
100%|##########| 170M/170M [00:03<00:00, 49.8MB/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').
@@ -300,7 +300,7 @@ Get boxes with score larger than 0.9
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 3 minutes 8.729 seconds)
+ **Total running time of the script:** ( 3 minutes 8.162 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 bc271bad0d..722861f11e 100644
--- a/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
+++ b/docs/_sources/how_to/deploy_models/deploy_prequantized.rst.txt
@@ -240,7 +240,7 @@ 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, 49.2MB/s]
99%|#########9| 13.4M/13.6M [00:00<00:00, 52.7MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 52.3MB/s]
+
0%| | 0.00/13.6M [00:00<?, ?B/s]
59%|#####8 | 7.99M/13.6M [00:00<00:00, 51.8MB/s]
100%|##########| 13.6M/13.6M [00:00<00:00, 62.8MB/s]
@@ -422,7 +422,7 @@ 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)
- 90.0495 89.9418 94.4827 89.7411 0.5090
+ 90.1727 90.0634 92.2951 89.8462 0.3375
@@ -471,7 +471,7 @@ TODO
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 4.623 seconds)
+ **Total running time of the script:** ( 1 minutes 4.723 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 fc1bf78d45..eafd59dfaf 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
@@ -436,7 +436,7 @@ 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)
- 117.7106 117.3661 123.0387 115.9515 1.2004
+ 118.2884 118.1312 120.6947 116.4210 1.1491
@@ -473,7 +473,7 @@ 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 27.708 seconds)
+ **Total running time of the script:** ( 2 minutes 20.990 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 1d509fe038..7cd8e5edac 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 31.171 seconds)
+ **Total running time of the script:** ( 1 minutes 34.558 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 d096a2de05..7427d8a6f8 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]
4%|3 | 5232/132723 [00:00<00:02, 52305.96KB/s]
10%|# | 13353/132723 [00:00<00:01, 69300.86KB/s]
16%|#6 | 21478/132723 [00:00<00:01, 74754.60KB/s]
22%|##2 | 29789/132723 [00:00<00:01, 78050.79KB/s]
29%|##8 | 38078/132723 [00:00<00:01, 79794.79KB/s]
35%|###4 | 46433/132723 [00:00<00:01, 81069.42KB/s]
41%|####1 | 54729/132723 [00:00<00:00, 81685.77KB/s]
48%|####7 | 63047/132723 [00:00<00:00, 82159.47KB/s]
54%|#####3 | 71375/132723 [00:00<00:00, 82509.03KB/s]
60%|###### | 79818/132723 [00:01<00:00, 83094.06KB/s]
66%|######6 | 88128/132723 [00:01<00:00, 81084.51KB/s]
73%|#######2 | 96247/132723 [00:01<00:00, 80809.41KB/s]
79%|#######8 | 104335/132723 [00:01<00:00, 80800.48KB/s]
85%|########4 | 112420/132723 [00:01<00:00, 80797.92KB/s]
91%|######### | 120504/132723 [00:01<00:00, 80759.95KB/s]
97%|########
#6| 128583/132723 [00:01<00:00, 80639.96KB/s]
100%|##########| 132723/132723 [00:01<00:00, 79602.00KB/s]
+
0%| | 0/132723 [00:00<?, ?KB/s]
4%|4 | 5468/132723 [00:00<00:02, 54601.03KB/s]
10%|# | 13312/132723 [00:00<00:01, 68609.69KB/s]
15%|#5 | 20174/132723 [00:00<00:02, 45614.49KB/s]
21%|##1 | 28173/132723 [00:00<00:01, 56008.52KB/s]
27%|##6 | 35532/132723 [00:00<00:01, 61311.98KB/s]
33%|###2 | 43500/132723 [00:00<00:01, 66842.14KB/s]
39%|###8 | 51360/132723 [00:00<00:01, 70378.10KB/s]
45%|####4 | 59412/132723 [00:00<00:00, 73424.00KB/s]
51%|##### | 67408/132723 [00:01<00:00, 75386.50KB/s]
57%|#####6 | 75477/132723 [00:01<00:00, 76976.96KB/s]
63%|######2 | 83553/132723 [00:01<00:00, 78110.97KB/s]
69%|######8 | 91530/132723 [00:01<00:00, 78607.72KB/s]
75%|#######4 | 99480/132723 [00:01<00:00, 78874.32KB/s]
81%|######## | 107450/132723 [00:01<00:00, 79115.81KB/s]
87%|########7 | 115488/132723 [00:01<00:00, 79494.38KB/s]
93%|#########
3| 123603/132723 [00:01<00:00, 79988.94KB/s]
99%|#########9| 131921/132723 [00:01<00:00, 80942.53KB/s]
100%|##########| 132723/132723 [00:01<00:00, 72855.81KB/s]
@@ -246,7 +246,7 @@ Display result
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 3 minutes 2.839 seconds)
+ **Total running time of the script:** ( 3 minutes 3.144 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 046778d489..8edc65a4e3 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
@@ -5,26 +5,26 @@
Computation times
=================
-**13:40.032** total execution time for **how_to_deploy_models** files:
+**13:25.645** total execution time for **how_to_deploy_models** files:
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:08.729 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_object_detection_pytorch.py` (``deploy_object_detection_pytorch.py``) | 03:08.162 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 03:02.839 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_ssd_gluoncv.py` (``deploy_ssd_gluoncv.py``) | 03:03.144 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:27.708 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized_tflite.py` (``deploy_prequantized_tflite.py``) | 02:20.990 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:31.171 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_quantized.py` (``deploy_quantized.py``) | 01:34.558 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:04.623 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_prequantized.py` (``deploy_prequantized.py``) | 01:04.723 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno.py` (``deploy_model_on_adreno.py``) | 01:00.711 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_adreno.py` (``deploy_model_on_adreno.py``) | 00:50.923 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:34.561 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_android.py` (``deploy_model_on_android.py``) | 00:34.624 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_nano.py` (``deploy_model_on_nano.py``) | 00:25.021 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_nano.py` (``deploy_model_on_nano.py``) | 00:24.451 | 0.0 MB |
+------------------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:24.662 | 0.0 MB |
+| :ref:`sphx_glr_how_to_deploy_models_deploy_model_on_rasp.py` (``deploy_model_on_rasp.py``) | 00:24.065 | 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 6425f5e659..d4594c86ce 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
@@ -476,7 +476,7 @@ 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.zip9677fdde-9c12-4bd6-aefe-38a5f844cd59 from https://apache-mxnet.s3-accelerate.dualstack.amazonaws.com/gluon/models/mobilenet0.25-9f83e440.zip...
+ Downloading /workspace/.mxnet/models/mobilenet0.25-9f83e440.zip1440c57f-f2d6-4d23-9bae-eed2f8ecf36f 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 745b13dcba..a8dc4aaf3d 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
=================
-**00:46.389** total execution time for **how_to_extend_tvm** files:
+**00:45.776** 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``) | 00:43.064 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_bring_your_own_datatypes.py` (``bring_your_own_datatypes.py``) | 00:42.488 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.331 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_instrument.py` (``use_pass_instrument.py``) | 00:02.290 | 0.0 MB |
+-------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:00.987 | 0.0 MB |
+| :ref:`sphx_glr_how_to_extend_tvm_use_pass_infra.py` (``use_pass_infra.py``) | 00:00.990 | 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 |
+-------------------------------------------------------------------------------------------------+-----------+--------+
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 85b0086abd..7d78d890b7 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: 7133us [7133us] (46.57%; 46.57%)
- FoldScaleAxis: 8185us [6us] (53.43%; 53.43%)
- FoldConstant: 8178us [1644us] (53.39%; 99.92%)
- InferType: 6534us [6534us] (42.66%; 79.89%)
+ InferType: 7311us [7311us] (46.93%; 46.93%)
+ FoldScaleAxis: 8269us [8us] (53.07%; 53.07%)
+ FoldConstant: 8262us [1697us] (53.03%; 99.91%)
+ InferType: 6565us [6565us] (42.13%; 79.46%)
@@ -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: 6580us [6580us] (44.96%; 44.96%)
- FoldScaleAxis: 8057us [4us] (55.04%; 55.04%)
- FoldConstant: 8053us [1616us] (55.02%; 99.95%)
- InferType: 6436us [6436us] (43.97%; 79.93%)
+ InferType: 6603us [6603us] (44.94%; 44.94%)
+ FoldScaleAxis: 8091us [5us] (55.06%; 55.06%)
+ FoldConstant: 8086us [1686us] (55.03%; 99.94%)
+ InferType: 6400us [6400us] (43.55%; 79.15%)
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 5f2af030cd..cca5e9f76e 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
@@ -344,7 +344,7 @@ latency of convolution.
.. code-block:: none
- Convolution: 38.573760 ms
+ Convolution: 54.223007 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 4373670bcf..03bcab1686 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
@@ -661,7 +661,7 @@ be able to run on our build server
.. code-block:: none
- conv2d with tensor core: 13.346890 ms
+ conv2d with tensor core: 11.942332 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 1fce5cd7ad..4744b7ca39 100644
--- a/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
+++ b/docs/_sources/how_to/optimize_operators/opt_gemm.rst.txt
@@ -147,8 +147,8 @@ Then we write a baseline implementation, the simplest way to write a matrix mult
.. code-block:: none
- Numpy running time: 0.018447
- Baseline: 3.199606
+ Numpy running time: 0.017736
+ Baseline: 3.189127
@@ -242,7 +242,7 @@ fill 32 * 32 * sizeof(float) which is 4KB in the cache whose total size is 32KB
.. code-block:: none
- Opt1: 0.291128
+ Opt1: 0.295809
@@ -344,7 +344,7 @@ In this tutorial, we chose to vectorize the inner loop row data since it is cach
.. code-block:: none
- Opt2: 0.330108
+ Opt2: 0.330921
@@ -439,7 +439,7 @@ the access pattern for A matrix is more cache friendly.
.. code-block:: none
- Opt3: 0.115935
+ Opt3: 0.113154
@@ -563,7 +563,7 @@ flattening.
.. code-block:: none
- Opt4: 0.109205
+ Opt4: 0.109051
@@ -684,7 +684,7 @@ write to C when all the block results are ready.
.. code-block:: none
- Opt5: 0.110408
+ Opt5: 0.111148
@@ -808,7 +808,7 @@ Furthermore, we can also utilize multi-core processors to do the thread-level pa
.. code-block:: none
- Opt6: 0.146240
+ Opt6: 0.146006
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 c35efedb38..ebbe950bad 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
=================
-**00:34.144** total execution time for **how_to_optimize_operators** files:
+**00:34.187** total execution time for **how_to_optimize_operators** files:
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:31.424 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_gemm.py` (``opt_gemm.py``) | 00:31.378 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.566 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_tensorcore.py` (``opt_conv_tensorcore.py``) | 00:01.581 | 0.0 MB |
+-----------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.154 | 0.0 MB |
+| :ref:`sphx_glr_how_to_optimize_operators_opt_conv_cuda.py` (``opt_conv_cuda.py``) | 00:01.227 | 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 4d020d20f2..7bb292567b 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
=================
-**08:53.425** total execution time for **how_to_tune_with_autoscheduler** files:
+**08:58.603** 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``) | 05:29.809 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_conv2d_layer_cuda.py` (``tune_conv2d_layer_cuda.py``) | 05:26.025 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:30.541 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_x86.py` (``tune_network_x86.py``) | 01:30.113 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 01:00.857 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_cuda.py` (``tune_network_cuda.py``) | 01:00.903 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:29.525 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_sparse_x86.py` (``tune_sparse_x86.py``) | 00:38.892 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:11.758 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_arm.py` (``tune_network_arm.py``) | 00:11.784 | 0.0 MB |
+----------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:10.936 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autoscheduler_tune_network_mali.py` (``tune_network_mali.py``) | 00:10.885 | 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 503ec7f26d..149257be4a 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
@@ -243,12 +243,12 @@ cooperative fetching, unrolling and operator fusion.
bias: Buffer(bias_2: Pointer(float32), float32, [1, 512, 1, 1], []),
compute: Buffer(compute_2: Pointer(float32), float32, [1, 512, 7, 7], [])}
buffer_map = {data_1: data, kernel_1: kernel, bias_1: bias, compute_1: compute} {
- attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 64;
- allocate(conv2d_nchw: Pointer(local float32), float32, [8]), storage_scope = local;
- allocate(pad_temp.shared: Pointer(shared float32), float32, [4032]), storage_scope = shared;
+ attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 16;
+ allocate(conv2d_nchw: Pointer(local float32), float32, [14]), storage_scope = local;
+ allocate(pad_temp.shared: Pointer(shared float32), float32, [1008]), storage_scope = shared;
allocate(kernel.shared: Pointer(shared float32), float32, [1536]), storage_scope = shared;
- attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- conv2d_nchw_1: Buffer(conv2d_nchw, float32, [8], [], scope="local", align=32)[0] = 0f32
+ attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112 {
+ conv2d_nchw_1: Buffer(conv2d_nchw, float32, [1], [], scope="local", align=4)[0] = 0f32
conv2d_nchw_1[1] = 0f32
conv2d_nchw_1[2] = 0f32
conv2d_nchw_1[3] = 0f32
@@ -256,258 +256,754 @@ cooperative fetching, unrolling and operator fusion.
conv2d_nchw_1[5] = 0f32
conv2d_nchw_1[6] = 0f32
conv2d_nchw_1[7] = 0f32
- for (rc.outer.outer: int32, 0, 8) {
+ conv2d_nchw_1[8] = 0f32
+ conv2d_nchw_1[9] = 0f32
+ conv2d_nchw_1[10] = 0f32
+ conv2d_nchw_1[11] = 0f32
+ conv2d_nchw_1[12] = 0f32
+ conv2d_nchw_1[13] = 0f32
+ for (rc.outer.outer: int32, 0, 32) {
for (rx.outer.outer: int32, 0, 3) {
- let cse_var_2: int32 = (rc.outer.outer*3136)
- let cse_var_1: int32 = (rc.outer.outer*576)
+ let cse_var_2: int32 = (rc.outer.outer*784)
+ let cse_var_1: int32 = (rc.outer.outer*144)
{
- attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1: Buffer(pad_temp.shared, float32, [4032], [], scope="shared")[(threadIdx.x_1*3)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3: Buffer(data_2, float32, [25088], [])[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) - 8)], 0f32 [...]
- pad_temp.shared_1[((threadIdx.x_1*3) + 1)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) - 7)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) - 6)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 49), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 49), 21)*49 [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 49), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + (f [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 49), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + (f [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 98), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 98), 21)*49 [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 98), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + (f [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 98), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + (f [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 441)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 335)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 442)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 336)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 443)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 337)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 196), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 196), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 196), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 196), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 245), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 245), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 245), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 245), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 882)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 678)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 883)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 679)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 884)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 680)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 343), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 343), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 343), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 343), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 392), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 392), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 392), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 392), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 1323)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1021)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 1324)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1022)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 1325)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1023)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 490), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 490), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 490), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 490), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 539), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 539), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 539), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 539), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 1764)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1364)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 1765)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1365)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 1766)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1366)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 637), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 637), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 637), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 637), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 686), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 686), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 686), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 686), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 2205)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1707)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2206)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1708)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2207)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1709)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 784), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 784), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 784), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 833), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 833), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 833), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 833), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 2646)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2050)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2647)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2051)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2648)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2052)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 931), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 931), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 931), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 931), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 980), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 980), 21)* [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 980), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 980), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + ( [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3087)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2393)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 3088)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2394)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 3089)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2395)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1078), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 1078), 21 [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1078), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1078), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1127), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 1127), 21 [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1127), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1127), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3528)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2736)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 3529)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2737)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 3530)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2738)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1225), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 1225), 21 [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1225), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1225), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1274), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 1274), 21 [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1274), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((((cse_var_2 + [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1274), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((((cse_var_2 + [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- if @tir.likely((threadIdx.x_1 < 21), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3969)] = @tir.if_then_else(((((2 < threadIdx.x_1) && (threadIdx.x_1 < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((cse_var_2 + (threadIdx.x_1*3)) + rx.outer.outer) + 3079)], 0f32, dtype=float32)
- }
- if @tir.likely((threadIdx.x_1 < 21), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3970)] = @tir.if_then_else(((((2 <= threadIdx.x_1) && (threadIdx.x_1 < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((cse_var_2 + ((threadIdx.x_1*3) + 1)) + rx.outer.outer) + 3079)], 0f32, dtype=float32)
- }
- if @tir.likely((threadIdx.x_1 < 21), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3971)] = @tir.if_then_else(((((1 < threadIdx.x_1) && (threadIdx.x_1 < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((cse_var_2 + ((threadIdx.x_1*3) + 2)) + rx.outer.outer) + 3079)], 0f32, dtype=float32)
- }
- }
- attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1: Buffer(kernel.shared, float32, [1536], [], scope="shared")[threadIdx.x_2] = kernel_3: Buffer(kernel_2, float32, [2359296], [])[((((blockIdx.x*36864) + cse_var_1) + (threadIdx.x_2*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 49)] = kernel_3[(((((blockIdx.x*36864) + cse_var_1) + (floordiv((threadIdx.x_2 + 49), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 98)] = kernel_3[(((((blockIdx.x*36864) + cse_var_1) + (floordiv((threadIdx.x_2 + 98), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 147)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 147), 192)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 49), 64)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 196)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 196), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 245)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 245), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 53), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 294)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 294), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 34)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 343)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 343), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 151), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 392)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 392), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 441)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 441), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 19)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 490)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 490), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 106), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 539)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 539), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 155), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 588)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 588), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 637)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 637), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 61), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 686)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 686), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 110), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 735)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 735), 192)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 53), 64)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 784)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 784), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 833)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 833), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 65), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 882)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 882), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 38)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 931)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 931), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 163), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 980)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 980), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1029)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1029), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 23)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1078)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1078), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 118), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1127)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1127), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 167), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1176)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1176), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 8)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1225)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1225), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 73), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1274)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1274), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 122), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1323)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1323), 192)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 57), 64)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1372)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1372), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1421)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1421), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 77), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1470)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1470), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 42)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- if @tir.likely((threadIdx.x_2 < 17), dtype=bool) {
- kernel.shared_1[(threadIdx.x_2 + 1519)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1519), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 175), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- }
- for (rc.inner: int32, 0, 64) {
- let cse_var_3: int32 = (rc.inner*3)
- {
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[cse_var_3]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 192)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 384)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 576)]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 768)]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 960)]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 1152)]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 1344)]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 1)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 193)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 385)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 577)]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 769)]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 961)]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 1153)]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 1345)]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 2)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 194)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 386)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 578)]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 770)]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 962)]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 1154)]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 1346)]))
- }
+ attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1: Buffer(pad_temp.shared, float32, [1008], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else(((((7 <= floormod(threadIdx.x_1, 63)) && (floormod(threadIdx.x_1, 63) < 56)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3: Buffer(data_2, float32, [25088], [])[((((cse_var_2 + (floordiv(threadIdx.x_1, 63)*49)) + rx.outer.outer) + floormod(threadIdx.x_1, 63)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 112)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 7), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 7), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 112), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 7), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x_1, 7)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 224)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 5), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 5), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 224), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 5), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x_1, 7)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 336)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 3), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 336), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 3), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x_1, 7)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 448)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 1), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 1), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 448), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 1), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x_1, 7)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 560)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 8), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 8), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 560), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 8), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x_1, 7)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 672)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 6), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 672), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 6), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x_1, 7)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 784)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 4), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 4), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 4), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x_1, 7)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 896)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 2), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 2), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 896), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 2), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x_1, 7)) - 8)], 0f32, dtype=float32)
+ attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1: Buffer(kernel.shared, float32, [1536], [], scope="shared")[threadIdx.x_2] = kernel_3: Buffer(kernel_2, float32, [2359296], [])[(((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 112)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 112), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 224)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 224), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 48), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 336)] = kernel_3[((((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer) + 32256)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 448)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 448), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 560)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 560), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 48), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 672)] = kernel_3[((((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer) + 64512)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 784)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 784), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 896)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 896), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 48), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1008)] = kernel_3[((((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer) + 96768)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1120)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 1120), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1232)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 1232), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 48), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1344)] = kernel_3[((((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer) + 129024)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ if @tir.likely((threadIdx.x_2 < 80), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 1456)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 1456), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
}
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(floormod(threadIdx.x, 7)*7)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 1)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 2)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 3)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 4)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 5)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 6)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(floormod(threadIdx.x, 7)*7)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 1)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 2)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 3)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 4)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 5)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 6)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 7)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 8)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 9)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 10)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 11)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 12)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 13)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 7)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 8)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 9)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 10)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 11)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 12)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 13)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 14)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 15)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 16)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 17)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 18)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 19)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 20)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 14)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 15)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 16)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 17)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 18)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 19)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 20)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 63)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 64)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 65)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 66)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 67)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 68)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 69)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 63)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 64)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 65)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 66)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 67)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 68)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 69)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 70)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 71)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 72)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 73)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 74)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 75)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 76)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 70)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 71)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 72)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 73)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 74)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 75)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 76)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 77)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 78)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 79)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 80)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 81)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 82)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 83)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 77)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 78)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 79)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 80)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 81)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 82)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 83)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 126)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 127)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 128)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 129)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 130)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 131)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 132)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 126)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 127)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 128)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 129)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 130)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 131)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 132)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 133)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 134)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 135)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 136)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 137)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 138)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 139)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 133)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 134)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 135)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 136)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 137)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 138)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 139)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 140)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 141)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 142)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 143)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 144)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 145)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 146)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 140)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 141)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 142)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 143)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 144)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 145)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 146)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 189)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 190)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 191)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 192)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 193)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 194)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 195)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 189)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 190)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 191)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 192)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 193)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 194)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 195)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 196)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 197)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 198)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 199)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 200)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 201)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 202)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 196)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 197)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 198)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 199)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 200)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 201)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 202)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 203)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 204)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 205)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 206)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 207)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 208)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 209)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 203)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 204)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 205)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 206)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 207)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 208)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 209)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 252)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 253)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 254)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 255)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 256)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 257)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 258)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 252)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 253)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 254)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 255)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 256)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 257)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 258)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 259)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 260)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 261)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 262)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 263)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 264)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 265)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 259)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 260)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 261)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 262)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 263)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 264)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 265)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 266)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 267)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 268)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 269)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 270)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 271)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 272)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 266)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 267)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 268)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 269)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 270)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 271)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 272)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 315)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 316)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 317)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 318)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 319)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 320)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 321)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 315)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 316)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 317)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 318)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 319)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 320)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 321)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 322)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 323)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 324)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 325)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 326)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 327)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 328)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 322)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 323)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 324)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 325)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 326)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 327)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 328)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 329)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 330)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 331)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 332)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 333)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 334)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 335)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 329)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 330)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 331)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 332)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 333)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 334)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 335)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 378)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 379)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 380)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 381)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 382)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 383)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 384)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 378)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 379)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 380)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 381)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 382)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 383)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 384)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 385)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 386)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 387)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 388)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 389)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 390)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 391)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 385)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 386)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 387)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 388)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 389)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 390)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 391)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 392)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 393)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 394)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 395)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 396)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 397)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 398)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 392)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 393)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 394)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 395)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 396)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 397)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 398)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 441)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 442)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 443)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 444)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 445)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 446)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 447)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 441)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 442)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 443)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 444)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 445)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 446)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 447)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 448)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 449)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 450)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 451)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 452)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 453)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 454)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 448)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 449)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 450)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 451)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 452)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 453)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 454)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 455)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 456)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 457)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 458)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 459)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 460)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 461)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 455)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 456)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 457)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 458)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 459)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 460)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 461)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 504)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 505)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 506)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 507)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 508)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 509)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 510)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 504)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 505)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 506)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 507)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 508)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 509)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 510)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 511)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 512)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 513)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 514)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 515)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 516)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 517)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 511)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 512)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 513)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 514)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 515)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 516)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 517)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 518)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 519)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 520)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 521)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 522)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 523)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 524)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 518)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 519)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 520)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 521)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 522)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 523)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 524)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 567)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 568)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 569)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 570)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 571)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 572)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 573)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 567)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 568)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 569)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 570)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 571)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 572)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 573)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 574)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 575)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 576)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 577)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 578)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 579)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 580)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 574)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 575)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 576)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 577)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 578)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 579)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 580)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 581)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 582)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 583)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 584)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 585)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 586)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 587)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 581)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 582)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 583)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 584)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 585)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 586)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 587)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 630)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 631)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 632)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 633)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 634)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 635)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 636)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 630)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 631)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 632)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 633)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 634)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 635)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 636)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 637)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 638)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 639)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 640)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 641)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 642)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 643)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 637)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 638)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 639)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 640)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 641)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 642)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 643)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 644)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 645)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 646)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 647)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 648)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 649)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 650)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 644)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 645)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 646)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 647)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 648)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 649)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 650)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 693)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 694)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 695)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 696)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 697)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 698)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 699)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 693)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 694)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 695)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 696)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 697)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 698)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 699)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 700)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 701)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 702)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 703)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 704)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 705)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 706)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 700)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 701)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 702)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 703)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 704)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 705)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 706)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 707)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 708)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 709)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 710)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 711)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 712)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 713)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 707)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 708)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 709)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 710)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 711)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 712)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 713)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 756)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 757)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 758)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 759)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 760)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 761)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 762)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 756)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 757)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 758)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 759)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 760)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 761)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 762)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 763)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 764)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 765)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 766)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 767)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 768)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 769)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 763)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 764)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 765)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 766)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 767)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 768)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 769)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 770)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 771)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 772)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 773)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 774)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 775)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 776)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 770)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 771)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 772)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 773)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 774)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 775)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 776)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 819)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 820)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 821)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 822)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 823)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 824)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 825)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 819)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 820)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 821)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 822)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 823)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 824)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 825)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 826)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 827)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 828)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 829)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 830)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 831)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 832)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 826)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 827)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 828)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 829)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 830)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 831)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 832)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 833)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 834)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 835)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 836)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 837)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 838)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 839)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 833)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 834)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 835)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 836)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 837)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 838)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 839)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 882)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 883)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 884)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 885)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 886)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 887)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 888)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 882)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 883)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 884)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 885)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 886)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 887)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 888)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 889)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 890)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 891)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 892)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 893)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 894)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 895)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 889)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 890)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 891)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 892)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 893)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 894)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 895)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 896)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 897)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 898)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 899)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 900)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 901)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 902)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 896)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 897)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 898)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 899)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 900)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 901)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 902)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 945)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 946)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 947)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 948)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 949)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 950)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 951)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 945)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 946)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 947)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 948)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 949)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 950)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 951)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 952)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 953)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 954)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 955)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 956)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 957)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 958)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 952)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 953)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 954)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 955)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 956)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 957)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 958)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 959)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 960)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 961)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 962)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 963)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 964)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 965)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 959)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 960)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 961)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 962)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 963)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 964)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 965)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
}
}
}
- for (i1.inner: int32, 0, 8) {
- compute_3: Buffer(compute_2, float32, [25088], [])[(((blockIdx.x*392) + (i1.inner*49)) + threadIdx.x)] = max((conv2d_nchw_1[i1.inner] + bias_3: Buffer(bias_2, float32, [512], [])[((blockIdx.x*8) + i1.inner)]), 0f32)
- }
+ compute_3: Buffer(compute_2, float32, [25088], [])[((blockIdx.x*1568) + (threadIdx.x*7))] = max((conv2d_nchw_1[0] + bias_3: Buffer(bias_2, float32, [512], [])[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 1)] = max((conv2d_nchw_1[1] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 2)] = max((conv2d_nchw_1[2] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 3)] = max((conv2d_nchw_1[3] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 4)] = max((conv2d_nchw_1[4] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 5)] = max((conv2d_nchw_1[5] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 6)] = max((conv2d_nchw_1[6] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 784)] = max((conv2d_nchw_1[7] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 785)] = max((conv2d_nchw_1[8] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 786)] = max((conv2d_nchw_1[9] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 787)] = max((conv2d_nchw_1[10] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 788)] = max((conv2d_nchw_1[11] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 789)] = max((conv2d_nchw_1[12] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 790)] = max((conv2d_nchw_1[13] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
}
}
@@ -561,7 +1057,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 0.289 ms
+ Execution time of this operator: 0.388 ms
@@ -609,37 +1105,37 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
conv2d_nchw_nn_o_o_i, conv2d_nchw_nn_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_i, factor=1)
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=8)
+ 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=1)
- 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_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=2)
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=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=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=64)
- conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=1)
- 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_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_o, conv2d_nchw_xx_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_o_i, factor=7)
+ conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=1)
+ 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_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=3)
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)
s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nchw_yy_o_o_o_o, conv2d_nchw_xx_o_o_o_o, conv2d_nchw_nn_o_o_o_i, conv2d_nchw_ff_o_o_o_i, conv2d_nchw_yy_o_o_o_i, conv2d_nchw_xx_o_o_o_i, conv2d_nchw_nn_o_o_i, conv2d_nchw_ff_o_o_i, conv2d_nchw_yy_o_o_i, conv2d_nchw_xx_o_o_i, conv2d_nchw_rc_o_o, conv2d_nchw_ry_o_o, conv2d_nchw_rx_o_o, conv2d_nchw_rc_o_i, conv2d_nchw_ry_o_i, conv2d_nchw_rx_o_i, conv2d_nchw_nn_o_i, conv2d_nchw_ff_o_i, conv2d_nchw_yy_o_i, conv2 [...]
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=8)
- compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=1)
- 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=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=2)
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=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)
+ compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=1)
+ compute_i3_o_o_o, compute_i3_o_o_i = s[compute].split(compute_i3_o_o_i, factor=7)
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)
kernel_shared = s.cache_read(kernel, "shared", [conv2d_nchw])
@@ -658,12 +1154,12 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
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=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=49)
+ 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)
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=3)
+ 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)
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=49)
+ 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)
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", 1024)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
@@ -683,9 +1179,9 @@ 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__(49) 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];
+ 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[14];
+ __shared__ float pad_temp_shared[1008];
__shared__ float kernel_shared[1536];
conv2d_nchw[0] = 0.000000e+00f;
conv2d_nchw[1] = 0.000000e+00f;
@@ -695,165 +1191,729 @@ They can be used for debugging and learning the behavior of the auto-scheduler.
conv2d_nchw[5] = 0.000000e+00f;
conv2d_nchw[6] = 0.000000e+00f;
conv2d_nchw[7] = 0.000000e+00f;
- for (int rc_outer_outer = 0; rc_outer_outer < 8; ++rc_outer_outer) {
+ conv2d_nchw[8] = 0.000000e+00f;
+ conv2d_nchw[9] = 0.000000e+00f;
+ conv2d_nchw[10] = 0.000000e+00f;
+ conv2d_nchw[11] = 0.000000e+00f;
+ conv2d_nchw[12] = 0.000000e+00f;
+ conv2d_nchw[13] = 0.000000e+00f;
+ for (int rc_outer_outer = 0; rc_outer_outer < 32; ++rc_outer_outer) {
for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
__syncthreads();
- pad_temp_shared[(((int)threadIdx.x) * 3)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) - 7)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) - 6)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 49) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 49) / 21) * 49)) + (((((((int)threadIdx.x) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 49) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 49) / 21) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 49) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 49) / 21) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 98) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 98) / 21) * 49)) + (((((((int)threadIdx.x) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 98) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 98) / 21) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 98) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 98) / 21) [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 441)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 335)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 442)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 336)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 443)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 337)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 196) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 196) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 196) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 196) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 196) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 196) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 245) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 245) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 245) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 245) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 245) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 245) / 2 [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 882)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 678)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 883)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 679)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 884)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 680)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 343) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 343) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 343) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 343) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 343) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 343) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 392) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 392) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 392) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 392) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 392) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 392) / 2 [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1323)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1021)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1324)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1022)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1325)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1023)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 490) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 490) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 490) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 490) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 490) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 490) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 539) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 539) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 539) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 539) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 539) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 539) / 2 [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1764)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1364)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1765)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1365)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1766)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1366)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 637) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 637) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 637) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 637) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 637) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 637) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 686) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 686) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 686) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 686) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 686) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 686) / 2 [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2205)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1707)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2206)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1708)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2207)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1709)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 784) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 784) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 784) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 784) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 784) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 784) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 833) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 833) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 833) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 833) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 833) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 833) / 2 [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2646)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2050)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2647)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2051)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2648)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2052)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 931) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 931) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 931) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 931) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 931) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 931) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 980) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 980) / 21) * 49)) + (((((((int)threadIdx.x) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 980) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 980) / 2 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 980) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 980) / 2 [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3087)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2393)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3088)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2394)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3089)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2395)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 1078) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1078) / 21) * 49)) + (((((((int)threadIdx.x [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1078) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1078) / [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1078) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1078) / [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1127) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1127) / 21) * 49)) + (((((((int)threadIdx.x [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1127) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1127) / [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1127) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1127) / [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3528)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2736)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3529)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2737)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3530)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2738)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 1225) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1225) / 21) * 49)) + (((((((int)threadIdx.x [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1225) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1225) / [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1225) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1225) / [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1274) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1274) / 21) * 49)) + (((((((int)threadIdx.x [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1274) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1274) / [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1274) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1274) / [...]
- if (((int)threadIdx.x) < 21) {
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3969)] = (((((2 < ((int)threadIdx.x)) && (((int)threadIdx.x) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((rc_outer_outer * 3136) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 3079)] : 0.000000e+00f);
- }
- if (((int)threadIdx.x) < 21) {
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3970)] = (((((2 <= ((int)threadIdx.x)) && (((int)threadIdx.x) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((rc_outer_outer * 3136) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 3080)] : 0.000000e+00f);
- }
- if (((int)threadIdx.x) < 21) {
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3971)] = (((((1 < ((int)threadIdx.x)) && (((int)threadIdx.x) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((rc_outer_outer * 3136) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 3081)] : 0.000000e+00f);
- }
- kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 36864) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 49)] = kernel[(((((((int)blockIdx.x) * 36864) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 49) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 98)] = kernel[(((((((int)blockIdx.x) * 36864) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 98) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 147) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) / 3) + 49) & 63) * 9)) + ((((int)threadIdx.x) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 196)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 196) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 4) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 245)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 245) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 53) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 294)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 294) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 306)];
- kernel_shared[(((int)threadIdx.x) + 343)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 343) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) + 151) % 192) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 392)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 392) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 8) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 441)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 441) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 171)];
- kernel_shared[(((int)threadIdx.x) + 490)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 490) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 106) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 539)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 539) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) + 155) % 192) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 588)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 588) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 36)];
- kernel_shared[(((int)threadIdx.x) + 637)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 637) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 61) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 686)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 686) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 110) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 735)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 735) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) / 3) + 53) & 63) * 9)) + ((((int)threadIdx.x) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 784) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 16) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 833)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 833) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 65) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 882)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 882) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 342)];
- kernel_shared[(((int)threadIdx.x) + 931)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 931) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) + 163) % 192) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 980)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 980) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 20) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1029)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1029) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 207)];
- kernel_shared[(((int)threadIdx.x) + 1078)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1078) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 118) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1127)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1127) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) + 167) % 192) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1176)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1176) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 72)];
- kernel_shared[(((int)threadIdx.x) + 1225)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1225) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 73) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1274)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1274) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 122) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1323)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1323) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) / 3) + 57) & 63) * 9)) + ((((int)threadIdx.x) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1372)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1372) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 28) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1421)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1421) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 77) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1470)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1470) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 378)];
- if (((int)threadIdx.x) < 17) {
- kernel_shared[(((int)threadIdx.x) + 1519)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1519) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 175) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ pad_temp_shared[((int)threadIdx.x)] = (((((7 <= (((int)threadIdx.x) % 63)) && ((((int)threadIdx.x) % 63) < 56)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 63) * 49)) + rx_outer_outer) + (((int)threadIdx.x) % 63)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((1 <= (((((int)threadIdx.x) / 7) + 7) % 9)) && ((((((int)threadIdx.x) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 112) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 7) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((1 <= (((((int)threadIdx.x) / 7) + 5) % 9)) && ((((((int)threadIdx.x) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 224) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 5) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((1 <= (((((int)threadIdx.x) / 7) + 3) % 9)) && ((((((int)threadIdx.x) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 336) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 3) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((1 <= (((((int)threadIdx.x) / 7) + 1) % 9)) && ((((((int)threadIdx.x) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 448) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 1) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((1 <= (((((int)threadIdx.x) / 7) + 8) % 9)) && ((((((int)threadIdx.x) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 560) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 8) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 672)] = (((((1 <= (((((int)threadIdx.x) / 7) + 6) % 9)) && ((((((int)threadIdx.x) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 672) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 6) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((1 <= (((((int)threadIdx.x) / 7) + 4) % 9)) && ((((((int)threadIdx.x) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 4) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 896)] = (((((1 <= (((((int)threadIdx.x) / 7) + 2) % 9)) && ((((((int)threadIdx.x) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 896) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 2) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ kernel_shared[((int)threadIdx.x)] = kernel[(((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 112) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 224) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 336)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer) + 32256)];
+ kernel_shared[(((int)threadIdx.x) + 448)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 448) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 560) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 672)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer) + 64512)];
+ kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 784) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 896)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 896) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer) + 96768)];
+ kernel_shared[(((int)threadIdx.x) + 1120)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1120) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1232)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1232) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1344)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer) + 129024)];
+ if (((int)threadIdx.x) < 80) {
+ kernel_shared[(((int)threadIdx.x) + 1456)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1456) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
}
__syncthreads();
- for (int rc_inner = 0; rc_inner < 64; ++rc_inner) {
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[(rc_inner * 3)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 192)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 384)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 576)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 768)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 960)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 1152)]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 1344)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 1)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 193)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 385)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 577)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 769)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 961)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 1153)]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 1345)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 2)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 194)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 386)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 578)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 770)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 962)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 1154)]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 1346)]));
- }
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) * 7)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 1)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 2)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 3)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 4)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 5)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 6)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((int)threadIdx.x) % 7) * 7)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 1)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 2)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 3)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 4)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 5)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 6)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 7)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 8)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 9)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 10)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 11)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 12)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 13)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 7)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 8)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 9)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 10)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 11)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 12)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 13)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 14)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 15)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 16)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 17)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 18)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 19)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 20)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 14)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 15)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 16)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 17)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 18)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 19)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 20)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 63)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 64)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 65)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 66)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 67)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 68)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 69)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 63)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 64)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 65)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 66)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 67)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 68)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 69)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 70)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 71)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 72)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 73)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 74)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 75)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 76)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 70)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 71)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 72)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 73)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 74)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 75)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 76)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 77)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 78)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 79)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 80)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 81)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 82)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 83)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 77)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 78)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 79)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 80)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 81)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 82)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 83)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 126)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 127)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 128)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 129)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 130)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 131)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 132)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 126)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 127)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 128)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 129)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 130)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 131)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 132)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 133)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 134)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 135)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 136)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 137)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 138)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 139)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 133)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 134)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 135)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 136)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 137)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 138)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 139)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 140)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 141)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 142)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 143)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 144)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 145)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 146)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 140)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 141)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 142)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 143)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 144)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 145)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 146)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 189)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 190)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 191)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 192)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 193)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 194)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 195)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 189)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 190)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 191)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 192)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 193)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 194)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 195)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 196)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 197)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 198)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 199)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 200)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 201)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 202)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 196)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 197)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 198)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 199)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 200)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 201)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 202)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 203)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 204)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 205)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 206)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 207)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 208)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 209)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 203)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 204)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 205)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 206)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 207)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 208)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 209)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 252)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 253)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 254)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 255)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 256)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 257)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 258)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 252)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 253)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 254)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 255)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 256)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 257)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 258)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 259)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 260)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 261)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 262)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 263)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 264)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 265)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 259)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 260)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 261)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 262)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 263)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 264)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 265)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 266)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 267)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 268)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 269)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 270)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 271)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 272)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 266)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 267)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 268)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 269)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 270)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 271)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 272)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 315)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 316)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 317)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 318)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 319)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 320)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 321)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 315)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 316)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 317)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 318)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 319)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 320)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 321)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 322)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 323)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 324)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 325)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 326)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 327)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 328)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 322)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 323)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 324)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 325)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 326)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 327)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 328)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 329)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 330)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 331)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 332)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 333)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 334)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 335)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 329)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 330)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 331)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 332)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 333)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 334)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 335)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 378)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 379)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 380)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 381)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 382)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 383)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 384)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 378)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 379)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 380)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 381)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 382)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 383)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 384)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 385)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 386)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 387)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 388)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 389)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 390)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 391)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 385)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 386)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 387)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 388)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 389)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 390)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 391)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 392)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 393)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 394)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 395)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 396)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 397)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 398)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 392)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 393)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 394)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 395)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 396)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 397)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 398)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 441)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 442)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 443)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 444)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 445)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 446)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 447)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 441)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 442)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 443)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 444)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 445)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 446)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 447)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 448)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 449)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 450)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 451)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 452)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 453)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 454)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 448)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 449)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 450)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 451)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 452)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 453)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 454)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 455)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 456)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 457)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 458)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 459)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 460)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 461)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 455)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 456)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 457)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 458)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 459)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 460)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 461)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 504)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 505)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 506)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 507)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 508)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 509)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 510)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 504)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 505)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 506)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 507)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 508)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 509)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 510)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 511)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 512)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 513)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 514)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 515)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 516)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 517)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 511)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 512)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 513)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 514)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 515)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 516)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 517)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 518)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 519)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 520)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 521)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 522)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 523)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 524)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 518)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 519)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 520)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 521)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 522)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 523)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 524)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 567)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 568)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 569)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 570)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 571)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 572)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 573)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 567)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 568)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 569)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 570)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 571)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 572)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 573)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 574)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 575)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 576)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 577)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 578)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 579)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 580)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 574)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 575)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 576)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 577)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 578)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 579)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 580)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 581)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 582)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 583)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 584)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 585)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 586)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 587)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 581)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 582)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 583)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 584)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 585)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 586)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 587)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 630)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 631)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 632)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 633)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 634)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 635)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 636)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 630)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 631)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 632)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 633)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 634)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 635)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 636)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 637)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 638)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 639)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 640)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 641)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 642)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 643)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 637)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 638)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 639)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 640)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 641)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 642)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 643)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 644)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 645)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 646)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 647)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 648)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 649)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 650)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 644)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 645)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 646)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 647)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 648)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 649)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 650)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 693)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 694)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 695)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 696)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 697)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 698)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 699)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 693)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 694)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 695)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 696)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 697)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 698)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 699)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 700)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 701)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 702)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 703)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 704)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 705)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 706)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 700)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 701)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 702)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 703)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 704)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 705)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 706)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 707)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 708)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 709)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 710)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 711)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 712)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 713)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 707)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 708)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 709)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 710)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 711)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 712)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 713)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 756)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 757)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 758)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 759)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 760)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 761)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 762)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 756)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 757)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 758)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 759)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 760)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 761)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 762)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 763)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 764)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 765)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 766)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 767)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 768)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 769)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 763)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 764)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 765)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 766)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 767)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 768)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 769)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 770)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 771)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 772)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 773)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 774)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 775)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 776)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 770)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 771)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 772)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 773)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 774)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 775)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 776)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 819)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 820)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 821)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 822)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 823)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 824)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 825)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 819)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 820)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 821)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 822)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 823)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 824)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 825)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 826)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 827)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 828)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 829)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 830)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 831)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 832)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 826)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 827)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 828)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 829)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 830)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 831)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 832)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 833)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 834)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 835)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 836)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 837)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 838)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 839)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 833)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 834)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 835)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 836)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 837)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 838)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 839)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 882)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 883)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 884)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 885)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 886)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 887)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 888)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 882)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 883)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 884)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 885)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 886)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 887)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 888)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 889)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 890)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 891)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 892)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 893)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 894)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 895)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 889)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 890)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 891)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 892)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 893)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 894)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 895)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 896)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 897)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 898)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 899)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 900)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 901)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 902)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 896)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 897)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 898)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 899)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 900)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 901)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 902)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 945)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 946)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 947)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 948)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 949)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 950)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 951)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 945)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 946)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 947)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 948)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 949)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 950)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 951)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 952)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 953)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 954)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 955)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 956)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 957)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 958)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 952)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 953)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 954)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 955)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 956)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 957)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 958)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 959)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 960)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 961)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 962)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 963)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 964)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 965)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 959)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 960)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 961)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 962)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 963)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 964)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 965)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
}
}
- for (int i1_inner = 0; i1_inner < 8; ++i1_inner) {
- compute[(((((int)blockIdx.x) * 392) + (i1_inner * 49)) + ((int)threadIdx.x))] = max((conv2d_nchw[i1_inner] + bias[((((int)blockIdx.x) * 8) + i1_inner)]), 0.000000e+00f);
- }
+ compute[((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7))] = max((conv2d_nchw[0] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 1)] = max((conv2d_nchw[1] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 2)] = max((conv2d_nchw[2] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 3)] = max((conv2d_nchw[3] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 4)] = max((conv2d_nchw[4] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 5)] = max((conv2d_nchw[5] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 6)] = max((conv2d_nchw[6] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 784)] = max((conv2d_nchw[7] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 785)] = max((conv2d_nchw[8] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 786)] = max((conv2d_nchw[9] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 787)] = max((conv2d_nchw[10] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 788)] = max((conv2d_nchw[11] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 789)] = max((conv2d_nchw[12] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 790)] = max((conv2d_nchw[13] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
}
@@ -914,7 +1974,7 @@ In the example below we resume the status and do more 5 trials.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 5 minutes 29.809 seconds)
+ **Total running time of the script:** ( 5 minutes 26.025 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 8a7c0acf3b..3151a3803f 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)
- 7.9005 7.9059 7.9113 7.8842 0.0117
+ 7.9106 7.9078 7.9204 7.9036 0.0071
@@ -675,7 +675,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 0.857 seconds)
+ **Total running time of the script:** ( 1 minutes 0.903 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 1759e5174e..25c928e064 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)
- 746.5433 746.6337 747.1823 745.8140 0.5622
+ 743.2264 743.2485 743.7502 742.6803 0.4371
@@ -694,7 +694,7 @@ Other Tips
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 30.541 seconds)
+ **Total running time of the script:** ( 1 minutes 30.113 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 9920a6ea38..70f94cc3df 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
@@ -390,28 +390,29 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
placeholder_4: Buffer(placeholder_14: Pointer(float32), float32, [128, 512], []),
compute: Buffer(compute_2: Pointer(float32), float32, [128, 512], [])}
buffer_map = {placeholder_5: placeholder, placeholder_6: placeholder_1, placeholder_7: placeholder_2, placeholder_8: placeholder_3, placeholder_9: placeholder_4, compute_1: compute} {
- for (i0.outer.i1.outer.fused: int32, 0, 32) "parallel" {
- allocate(compute_3: Pointer(global float32), float32, [2048]), storage_scope = global {
+ for (i0.outer.i1.outer.fused: int32, 0, 16) "parallel" {
+ allocate(compute_3: Pointer(global float32), float32, [4096]), storage_scope = global {
for (i.outer.inner: int32, 0, 8) {
- for (i.inner.init: int32, 0, 16) {
- for (j.init: int32, 0, 16) {
- compute_4: Buffer(compute_3, float32, [2048], [])[(((i.outer.inner*256) + (i.inner.init*16)) + j.init)] = 0f32
+ for (nb_j.inner: int32, 0, 2) {
+ for (i.inner.init: int32, 0, 16) {
+ for (j.init: int32, 0, 16) {
+ compute_4: Buffer(compute_3, float32, [4096], [])[((((i.outer.inner*512) + (i.inner.init*32)) + (nb_j.inner*16)) + j.init)] = 0f32
+ }
}
- }
- for (elem_idx: int32, 0, (placeholder_15: Buffer(placeholder_13, int32, [33], [])[(i0.outer.i1.outer.fused + 1)] - placeholder_15[i0.outer.i1.outer.fused])) {
- for (i.inner: int32, 0, 16) {
- for (j: int32, 0, 16) {
- if @tir.likely((elem_idx < (placeholder_15[(i0.outer.i1.outer.fused + 1)] - placeholder_15[i0.outer.i1.outer.fused])), dtype=bool) {
- let cse_var_1: int32 = (((i.outer.inner*256) + (i.inner*16)) + j)
- compute_4[cse_var_1] = (compute_4[cse_var_1] + (placeholder_16: Buffer(placeholder_11, float32, [78656], [])[(((placeholder_15[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + j)]*max(placeholder_17: Buffer(placeholder_10, float32, [32768], [])[(((i.outer.inner*4096) + (i.inner*256)) + placeholder_18: Buffer(placeholder_12, int32, [4916], [])[(placeholder_15[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ for (elem_idx: int32, 0, let cse_var_1: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner) in (placeholder_15: Buffer(placeholder_13, int32, [33], [])[(cse_var_1 + 1)] - placeholder_15[cse_var_1])) {
+ for (i.inner: int32, 0, 16) {
+ for (j: int32, 0, 16) {
+ let cse_var_3: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner)
+ let cse_var_2: int32 = ((((i.outer.inner*512) + (i.inner*32)) + (nb_j.inner*16)) + j)
+ compute_4[cse_var_2] = (compute_4[cse_var_2] + (placeholder_16: Buffer(placeholder_11, float32, [78656], [])[(((placeholder_15[cse_var_3]*16) + (elem_idx*16)) + j)]*max(placeholder_17: Buffer(placeholder_10, float32, [32768], [])[(((i.outer.inner*4096) + (i.inner*256)) + placeholder_18: Buffer(placeholder_12, int32, [4916], [])[(placeholder_15[cse_var_3] + elem_idx)])], 0f32)))
}
}
}
}
}
for (i0.inner: int32, 0, 128) {
- let cse_var_2: int32 = ((i0.inner*512) + (i0.outer.i1.outer.fused*16))
- compute_5: Buffer(compute_2, float32, [65536], [])[ramp(cse_var_2, 1, 16)] = max((compute_4[ramp((i0.inner*16), 1, 16)] + placeholder_19: Buffer(placeholder_14, float32, [65536], [])[ramp(cse_var_2, 1, 16)]), broadcast(0f32, 16))
+ let cse_var_4: int32 = ((i0.inner*512) + (i0.outer.i1.outer.fused*32))
+ compute_5: Buffer(compute_2, float32, [65536], [])[ramp(cse_var_4, 1, 32)] = max((compute_4[ramp((i0.inner*32), 1, 32)] + placeholder_19: Buffer(placeholder_14, float32, [65536], [])[ramp(cse_var_4, 1, 32)]), broadcast(0f32, 32))
}
}
}
@@ -467,7 +468,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 1.526 ms
+ Execution time of this operator: 1.508 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 c6357056af..eb30a3e582 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
@@ -5,12 +5,12 @@
Computation times
=================
-**00:48.546** total execution time for **how_to_tune_with_autotvm** files:
+**00:37.776** 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``) | 00:48.511 | 0.0 MB |
+| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_conv2d_cuda.py` (``tune_conv2d_cuda.py``) | 00:37.741 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_tune_with_autotvm_tune_relay_x86.py` (``tune_relay_x86.py``) | 00:00.020 | 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 |
+--------------------------------------------------------------------------------------------------+-----------+--------+
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 59adc93133..fb154c3f37 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
@@ -391,8 +391,9 @@ 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, 8, 16, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10062228
- No: 2 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 64, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3174427
+ No: 2 GFLOPS: 32.47/32.47 result: MeasureResult(costs=(0.007130298227272727,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.338937997817993, timestamp=1673430625.906357) [('tile_f', [-1, 16, 16, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1754369
+ No: 3 GFLOPS: 0.00/32.47 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 +515,8 @@ 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, 4, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2510569
- No: 3 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 16, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6948515
+ No: 4 GFLOPS: 0.00/32.47 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
@@ -637,27 +638,747 @@ 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, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5644349
- No: 4 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 142, in build
- res = future.result()
- File "/usr/lib/python3.7/concurrent/futures/_base.py", line 435, in result
- return self.__get_result()
- File "/usr/lib/python3.7/concurrent/futures/_base.py", line 384, in __get_result
- raise self._exception
- File "/usr/lib/python3.7/concurrent/futures/thread.py", line 57, in run
- result = self.fn(*self.args, **self.kwargs)
- File "/workspace/python/tvm/contrib/popen_pool.py", line 432, in <lambda>
- worker = lambda *args: self._worker_run(*args)
- File "/workspace/python/tvm/contrib/popen_pool.py", line 401, in _worker_run
- return proc.recv()
- File "/workspace/python/tvm/contrib/popen_pool.py", line 309, in recv
- raise TimeoutError()
- TimeoutError
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 64, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9888828
+ No: 5 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 4, 32, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1190022
+ No: 6 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 8, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1098144
+ No: 7 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 32, 1, 4]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 64]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10035625
+ No: 8 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 1, 256, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9941412
+ No: 9 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 1, 8, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9710167
+ No: 10 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 32, 1, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7173461
- No: 5 GFLOPS: 84.05/84.05 result: MeasureResult(costs=(0.002754245216216216,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5222408771514893, timestamp=1673414749.1550088) [('tile_f', [-1, 1, 4, 16]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2392895
- No: 6 GFLOPS: 0.00/84.05 result: Traceback (most recent call last):
+ 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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 32, 1, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 128, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1031145
+ No: 11 GFLOPS: 1.40/32.47 result: MeasureResult(costs=(0.1651578825,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.256547689437866, timestamp=1673430632.4948845) [('tile_f', [-1, 1, 1, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,8887835
+ No: 12 GFLOPS: 0.00/32.47 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
@@ -779,8 +1500,8 @@ 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, 4, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5949874
- No: 7 GFLOPS: 0.00/84.05 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 8, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 16, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10343820
+ No: 13 GFLOPS: 0.00/32.47 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
@@ -902,8 +1623,8 @@ 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, 4, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8103235
- No: 8 GFLOPS: 0.00/84.05 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9632850
+ No: 14 GFLOPS: 0.00/32.47 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
@@ -1025,8 +1746,8 @@ 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, 32, 2, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,962789
- No: 9 GFLOPS: 0.00/84.05 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 32, 1, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,288480
+ No: 15 GFLOPS: 0.00/32.47 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
@@ -1148,10 +1869,8 @@ 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, 32, 1, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 32, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2991400
- No: 10 GFLOPS: 87.66/87.66 result: MeasureResult(costs=(0.002640982239130435,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.2351536750793457, timestamp=1673414751.758597) [('tile_f', [-1, 8, 32, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1744203
- No: 11 GFLOPS: 134.20/134.20 result: MeasureResult(costs=(0.001725022275862069,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.235008955001831, timestamp=1673414752.4624658) [('tile_f', [-1, 2, 32, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,489981
- No: 12 GFLOPS: 0.00/134.20 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4498535
+ No: 16 GFLOPS: 0.00/32.47 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
@@ -1273,9 +1992,8 @@ 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, 16, 2, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3079354
- No: 13 GFLOPS: 94.54/134.20 result: MeasureResult(costs=(0.002448800608695652,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.6984493732452393, timestamp=1673414756.3338737) [('tile_f', [-1, 2, 8, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7478615
- No: 14 GFLOPS: 0.00/134.20 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 128, 1, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1212307
+ No: 17 GFLOPS: 0.00/32.47 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
@@ -1397,8 +2115,8 @@ 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, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 64]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8289826
- No: 15 GFLOPS: 0.00/134.20 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 2, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 8, 16]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7296933
+ No: 18 GFLOPS: 0.00/32.47 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
@@ -1520,12 +2238,26 @@ 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, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 128, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2735327
- No: 16 GFLOPS: 8.07/134.20 result: MeasureResult(costs=(0.02869964775,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5496540069580078, timestamp=1673414757.147663) [('tile_f', [-1, 2, 4, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6008570
- No: 17 GFLOPS: 4.92/134.20 result: MeasureResult(costs=(0.047058536500000005,), error_no=MeasureErrorNo.NO_ERROR, all_cost=7.601166486740112, timestamp=1673414764.9139814) [('tile_f', [-1, 2, 4, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3537090
- No: 18 GFLOPS: 46.92/134.20 result: MeasureResult(costs=(0.004934022047619048,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.2691385746002197, timestamp=1673414765.66909) [('tile_f', [-1, 2, 32, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5229342
- No: 19 GFLOPS: 1.63/134.20 result: MeasureResult(costs=(0.142306076,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.1770129203796387, timestamp=1673414767.9414265) [('tile_f', [-1, 8, 2, 32]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,391353
- No: 20 GFLOPS: 0.00/134.20 result: Traceback (most recent call last):
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 4, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10311738
+ No: 19 GFLOPS: 0.00/32.47 result: Traceback (most recent call last):
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 142, in build
+ res = future.result()
+ File "/usr/lib/python3.7/concurrent/futures/_base.py", line 435, in result
+ return self.__get_result()
+ File "/usr/lib/python3.7/concurrent/futures/_base.py", line 384, in __get_result
+ raise self._exception
+ File "/usr/lib/python3.7/concurrent/futures/thread.py", line 57, in run
+ result = self.fn(*self.args, **self.kwargs)
+ File "/workspace/python/tvm/contrib/popen_pool.py", line 432, in <lambda>
+ worker = lambda *args: self._worker_run(*args)
+ File "/workspace/python/tvm/contrib/popen_pool.py", line 401, in _worker_run
+ return proc.recv()
+ File "/workspace/python/tvm/contrib/popen_pool.py", line 309, in recv
+ raise TimeoutError()
+ TimeoutError
+
+ [('tile_f', [-1, 16, 2, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9919067
+ No: 20 GFLOPS: 0.00/32.47 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
@@ -1647,7 +2379,7 @@ 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, 8, 64, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4494868
+ tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 32, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 512]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,192244
@@ -1702,9 +2434,9 @@ and measure running time.
Finish loading 20 records
Best config:
- [('tile_f', [-1, 2, 32, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,489981
+ [('tile_f', [-1, 16, 16, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1754369
Finish loading 20 records
- Time cost of this operator: 0.002178
+ Time cost of this operator: 0.007549
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 4e24c409e4..882ec553a4 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
@@ -368,10 +368,10 @@ 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 309.4 98.717 (1, 2, 10, 10, 3) 2 1 [309.4]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.035 0.968 (1, 6, 10, 10) 1 1 [3.035]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.987 0.315 (1, 1, 10, 10, 3) 1 1 [0.987]
- Total_time - 313.421 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 311.2 98.735 (1, 2, 10, 10, 3) 2 1 [311.2]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.018 0.958 (1, 6, 10, 10) 1 1 [3.018]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.969 0.307 (1, 1, 10, 10, 3) 1 1 [0.969]
+ Total_time - 315.187 - - - - -
@@ -436,10 +436,10 @@ 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 103.4 97.494 (1, 6, 10, 10, 1) 2 1 [103.4]
- tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.811 1.708 (1, 6, 10, 10) 1 1 [1.811]
- tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.847 0.798 (1, 3, 10, 10, 1) 1 1 [0.847]
- Total_time - 106.058 - - - - -
+ tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 103.6 97.529 (1, 6, 10, 10, 1) 2 1 [103.6]
+ tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.774 1.67 (1, 6, 10, 10) 1 1 [1.774]
+ tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.851 0.801 (1, 3, 10, 10, 1) 1 1 [0.851]
+ Total_time - 106.225 - - - - -
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 22c3fe19e2..8b4710db58 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
@@ -117,7 +117,7 @@ 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, 18.4MB/s]
100%|##########| 3.42M/3.42M [00:00<00:00, 29.0MB/s]
+
0%| | 0.00/3.42M [00:00<?, ?B/s]
100%|##########| 3.42M/3.42M [00:00<00:00, 38.8MB/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.
@@ -322,7 +322,7 @@ Look up prediction top 1 index in 1000 class synset.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 1.827 seconds)
+ **Total running time of the script:** ( 1 minutes 1.602 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 c2c7f5588d..613a65a3de 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
@@ -218,7 +218,7 @@ take about **2 minutes** to download the Stanford Cars, while COCO 2017 validati
.. code-block:: none
- '/tmp/tmpy1qmbn_8/images/random'
+ '/tmp/tmpsi_54jvx/images/random'
@@ -309,7 +309,7 @@ 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], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0]
+ :alt: [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [0.0, 1.0]
:srcset: /how_to/work_with_microtvm/images/sphx_glr_micro_train_001.png
:class: sphx-glr-single-img
@@ -318,8 +318,8 @@ objects to other stuff? We can display some examples from our datasets using ``m
.. code-block:: none
- /tmp/tmpy1qmbn_8/images/target contains 8144 images
- /tmp/tmpy1qmbn_8/images/random contains 5000 images
+ /tmp/tmpsi_54jvx/images/target contains 8144 images
+ /tmp/tmpsi_54jvx/images/random contains 5000 images
@@ -494,13 +494,13 @@ the time on our validation set).
.. code-block:: none
Epoch 1/3
- 328/328 - 46s - loss: 0.2135 - accuracy: 0.9272 - val_loss: 0.1840 - val_accuracy: 0.9475 - 46s/epoch - 141ms/step
+ 328/328 - 46s - loss: 0.2090 - accuracy: 0.9301 - val_loss: 0.1511 - val_accuracy: 0.9524 - 46s/epoch - 141ms/step
Epoch 2/3
- 328/328 - 43s - loss: 0.0903 - accuracy: 0.9667 - val_loss: 0.1057 - val_accuracy: 0.9611 - 43s/epoch - 130ms/step
+ 328/328 - 43s - loss: 0.1003 - accuracy: 0.9636 - val_loss: 0.1229 - val_accuracy: 0.9532 - 43s/epoch - 130ms/step
Epoch 3/3
- 328/328 - 43s - loss: 0.0702 - accuracy: 0.9750 - val_loss: 0.1234 - val_accuracy: 0.9581 - 43s/epoch - 130ms/step
+ 328/328 - 43s - loss: 0.0682 - accuracy: 0.9737 - val_loss: 0.1307 - val_accuracy: 0.9615 - 43s/epoch - 130ms/step
- <keras.callbacks.History object at 0x7f21ade11f10>
+ <keras.callbacks.History object at 0x7fa812b525d0>
@@ -857,7 +857,7 @@ 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 23.439 seconds)
+ **Total running time of the script:** ( 4 minutes 45.819 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 7e7efbd744..86d4d3f622 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
@@ -5,18 +5,18 @@
Computation times
=================
-**06:27.123** total execution time for **how_to_work_with_microtvm** files:
+**06:49.719** total execution time for **how_to_work_with_microtvm** files:
+---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:23.439 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_train.py` (``micro_train.py``) | 04:45.819 | 0.0 MB |
+---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_pytorch.py` (``micro_pytorch.py``) | 01:01.827 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_pytorch.py` (``micro_pytorch.py``) | 01:01.602 | 0.0 MB |
+---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 00:50.305 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_autotune.py` (``micro_autotune.py``) | 00:50.647 | 0.0 MB |
+---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_aot.py` (``micro_aot.py``) | 00:07.820 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_aot.py` (``micro_aot.py``) | 00:07.884 | 0.0 MB |
+---------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:03.731 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_microtvm_micro_tflite.py` (``micro_tflite.py``) | 00:03.765 | 0.0 MB |
+---------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_microtvm_micro_reference_vm.py` (``micro_reference_vm.py``) | 00:00.001 | 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 5d5dcb100e..edc26e16c5 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
=================
-**00:43.473** total execution time for **how_to_work_with_relay** files:
+**00:43.599** 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``) | 00:31.860 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_pipeline_executor.py` (``using_pipeline_executor.py``) | 00:32.055 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:10.126 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_using_external_lib.py` (``using_external_lib.py``) | 00:10.096 | 0.0 MB |
+----------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.480 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_relay_build_gcn.py` (``build_gcn.py``) | 00:01.441 | 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 |
+----------------------------------------------------------------------------------------------------+-----------+--------+
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 9aacd18dc9..96e9d5c9a4 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
@@ -265,7 +265,7 @@ The following example customizes CUDA lowering rule for :code:`exp`.
.. code-block:: none
- <function my_cuda_math_rule at 0x7f21ae94db90>
+ <function my_cuda_math_rule at 0x7fa7b5ce87a0>
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 eef37fd9a3..703e0f0804 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
@@ -5,16 +5,16 @@
Computation times
=================
-**00:06.741** total execution time for **how_to_work_with_schedules** files:
+**00:04.921** total execution time for **how_to_work_with_schedules** files:
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:04.280 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_intrin_math.py` (``intrin_math.py``) | 00:02.410 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:01.120 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_tensorize.py` (``tensorize.py``) | 00:01.140 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.575 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_reduction.py` (``reduction.py``) | 00:00.588 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.554 | 0.0 MB |
+| :ref:`sphx_glr_how_to_work_with_schedules_scan.py` (``scan.py``) | 00:00.569 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_how_to_work_with_schedules_extern_op.py` (``extern_op.py``) | 00:00.112 | 0.0 MB |
+------------------------------------------------------------------------------------------------+-----------+--------+
diff --git a/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt b/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
index 747559fd59..96cd8d11b1 100644
--- a/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
+++ b/docs/_sources/how_to/work_with_schedules/tensorize.rst.txt
@@ -347,7 +347,7 @@ The importing needs to happen before the tensorized GEMV being executed.
B: Buffer(B_2: Pointer(float32), float32, [512, 64], []),
C: Buffer(C_2: Pointer(float32), float32, [1024, 512], [])}
buffer_map = {A_1: A, B_1: B, C_1: C} {
- attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp6vzbr1y2/input0.cc'\nsource_filename = \"/tmp/tmp6vzbr1y2/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n %7 = alloca float*, align 8\n %8 = alloca float*, align 8\n %9 = alloca floa [...]
+ attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp6evazljz/input0.cc'\nsource_filename = \"/tmp/tmp6evazljz/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n %7 = alloca float*, align 8\n %8 = alloca float*, align 8\n %9 = alloca floa [...]
for (i, 0, 1024) {
for (j.outer: int32, 0, 32) {
@tir.call_extern("gemv_update", @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), C_2, ((i*512) + (j.outer*16)), 16, 2, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), A_2, (i*64), 64, 1, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), B_2, (j.outer*1024), 1024, 1, dtype=handle), 16, 64, 64, dtype=int32)
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 d5d931c317..3c6945d4e8 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
=================
-**00:25.515** total execution time for **topic_vta_tutorials_autotvm** files:
+**00:25.459** total execution time for **topic_vta_tutorials_autotvm** files:
+---------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:25.509 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_autotvm_tune_relay_vta.py` (``tune_relay_vta.py``) | 00:25.452 | 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 4c92c49de0..81e3835829 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 27.89s!
+ resnet18_v1 inference graph built in 27.85s!
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 d8d2ca7c8c..37269a28a5 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 19.03s!
+ yolov3-tiny inference graph built in 18.99s!
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 20dabc6946..d13bc4baa8 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
=================
-**01:30.537** total execution time for **topic_vta_tutorials_frontend** files:
+**01:30.884** total execution time for **topic_vta_tutorials_frontend** files:
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:45.862 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_detection.py` (``deploy_detection.py``) | 00:46.189 | 0.0 MB |
+------------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:44.675 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_frontend_deploy_classification.py` (``deploy_classification.py``) | 00:44.695 | 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 eb9914d368..ba05d8d1d7 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
=================
-**00:03.113** total execution time for **topic_vta_tutorials_optimize** files:
+**00:03.200** total execution time for **topic_vta_tutorials_optimize** files:
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.653 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_convolution_opt.py` (``convolution_opt.py``) | 00:02.718 | 0.0 MB |
+--------------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.460 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_optimize_matrix_multiply_opt.py` (``matrix_multiply_opt.py``) | 00:00.482 | 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 87eb18d33a..d11b1e5a8a 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
=================
-**00:00.817** total execution time for **topic_vta_tutorials** files:
+**00:00.862** total execution time for **topic_vta_tutorials** files:
+---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.437 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_matrix_multiply.py` (``matrix_multiply.py``) | 00:00.465 | 0.0 MB |
+---------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.380 | 0.0 MB |
+| :ref:`sphx_glr_topic_vta_tutorials_vta_get_started.py` (``vta_get_started.py``) | 00:00.398 | 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 afc90671ea..3ae654e9d8 100644
--- a/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
+++ b/docs/_sources/tutorial/auto_scheduler_matmul_x86.rst.txt
@@ -329,7 +329,7 @@ We build the binary and check its correctness and performance.
.. code-block:: none
- Execution time of this operator: 95.448 ms
+ Execution time of this operator: 93.261 ms
@@ -447,7 +447,7 @@ operations.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 1 minutes 18.257 seconds)
+ **Total running time of the script:** ( 1 minutes 17.980 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 1c37b7ad8c..0f37929477 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: 1.92/1.92 result: MeasureResult(costs=(0.1397195844,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.4619967937469482, timestamp=1673413345.2397616) [('tile_y', [-1, 2]), ('tile_x', [-1, 2])],None,11
- No: 2 GFLOPS: 12.34/12.34 result: MeasureResult(costs=(0.021760005000000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6083128452301025, timestamp=1673413346.5937173) [('tile_y', [-1, 8]), ('tile_x', [-1, 256])],None,83
- No: 3 GFLOPS: 2.93/12.34 result: MeasureResult(costs=(0.09150838959999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.689319372177124, timestamp=1673413348.3066304) [('tile_y', [-1, 2]), ('tile_x', [-1, 8])],None,31
- No: 4 GFLOPS: 13.09/13.09 result: MeasureResult(costs=(0.020503555399999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5703060626983643, timestamp=1673413349.6385214) [('tile_y', [-1, 64]), ('tile_x', [-1, 512])],None,96
- No: 5 GFLOPS: 2.69/13.09 result: MeasureResult(costs=(0.09965446580000001,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8444793224334717, timestamp=1673413351.6700304) [('tile_y', [-1, 2]), ('tile_x', [-1, 16])],None,41
- No: 6 GFLOPS: 1.18/13.09 result: MeasureResult(costs=(0.2267873122,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.8997581005096436, timestamp=1673413356.3305795) [('tile_y', [-1, 16]), ('tile_x', [-1, 1])],None,4
- No: 7 GFLOPS: 12.87/13.09 result: MeasureResult(costs=(0.0208499138,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6586570739746094, timestamp=1673413356.9184117) [('tile_y', [-1, 256]), ('tile_x', [-1, 128])],None,78
- No: 8 GFLOPS: 12.29/13.09 result: MeasureResult(costs=(0.0218348038,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5916509628295898, timestamp=1673413357.5249972) [('tile_y', [-1, 256]), ('tile_x', [-1, 256])],None,88
- No: 9 GFLOPS: 3.68/13.09 result: MeasureResult(costs=(0.0728537726,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.3817014694213867, timestamp=1673413359.023297) [('tile_y', [-1, 256]), ('tile_x', [-1, 16])],None,48
- No: 10 GFLOPS: 12.58/13.09 result: MeasureResult(costs=(0.0213389654,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5646340847015381, timestamp=1673413359.6193628) [('tile_y', [-1, 64]), ('tile_x', [-1, 256])],None,86
+ No: 1 GFLOPS: 14.58/14.58 result: MeasureResult(costs=(0.018415960999999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6023857593536377, timestamp=1673429246.066462) [('tile_y', [-1, 32]), ('tile_x', [-1, 64])],None,65
+ No: 2 GFLOPS: 11.69/14.58 result: MeasureResult(costs=(0.022962678,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.636038064956665, timestamp=1673429246.685786) [('tile_y', [-1, 32]), ('tile_x', [-1, 32])],None,55
+ No: 3 GFLOPS: 1.61/14.58 result: MeasureResult(costs=(0.16642992380000002,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.8850674629211426, timestamp=1673429250.340677) [('tile_y', [-1, 4]), ('tile_x', [-1, 1])],None,2
+ No: 4 GFLOPS: 13.15/14.58 result: MeasureResult(costs=(0.020408874599999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6245789527893066, timestamp=1673429251.6660557) [('tile_y', [-1, 128]), ('tile_x', [-1, 128])],None,77
+ No: 5 GFLOPS: 11.90/14.58 result: MeasureResult(costs=(0.0225592394,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6617028713226318, timestamp=1673429253.1838212) [('tile_y', [-1, 64]), ('tile_x', [-1, 32])],None,56
+ No: 6 GFLOPS: 13.10/14.58 result: MeasureResult(costs=(0.020497674599999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5695252418518066, timestamp=1673429253.7658253) [('tile_y', [-1, 4]), ('tile_x', [-1, 512])],None,92
+ No: 7 GFLOPS: 2.71/14.58 result: MeasureResult(costs=(0.09888162959999999,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.8308913707733154, timestamp=1673429255.594606) [('tile_y', [-1, 2]), ('tile_x', [-1, 16])],None,41
+ No: 8 GFLOPS: 12.46/14.58 result: MeasureResult(costs=(0.0215481452,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.5928652286529541, timestamp=1673429256.1908658) [('tile_y', [-1, 64]), ('tile_x', [-1, 256])],None,86
+ No: 9 GFLOPS: 10.00/14.58 result: MeasureResult(costs=(0.026847188799999998,), error_no=MeasureErrorNo.NO_ERROR, all_cost=0.6478948593139648, timestamp=1673429256.9508803) [('tile_y', [-1, 8]), ('tile_x', [-1, 32])],None,53
+ No: 10 GFLOPS: 2.10/14.58 result: MeasureResult(costs=(0.1280598344,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.2577879428863525, timestamp=1673429259.251074) [('tile_y', [-1, 128]), ('tile_x', [-1, 4])],None,27
diff --git a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
index 66eec5b3fe..eef1789753 100644
--- a/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
+++ b/docs/_sources/tutorial/autotvm_relay_x86.rst.txt
@@ -324,7 +324,7 @@ standard deviation.
.. code-block:: none
- {'mean': 510.46957689999545, 'median': 509.98423024998374, 'std': 1.8196836642888992}
+ {'mean': 509.2969556299795, 'median': 509.13396689993533, 'std': 2.566962968850706}
@@ -558,30 +558,31 @@ 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: 19.08/ 19.20 GFLOPS | Progress: (4/20) | 10.40 s
[Task 1/25] Current/Best: 19.44/ 19.44 GFLOPS | Progress: (8/20) | 13.68 s
[Task 1/25] Current/Best: 18.36/ 23.18 GFLOPS | Progress: (12/20) | 15.50 s
[Task 1/25] Current/Best: 23.17/ 23.18 GFLOPS | Progress: (16/20) | 18.51 s
[Task 1/25] Current/Best: 14.40/ 23.18 GFLOPS | Progress: (20/20) | 20.73 s Done.
-
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 13.38/ 18.04 GFLOPS | Progress: (4/20) | 3.16 s
[Task 2/25] Current/Best: 16.51/ 18.04 GFLOPS | Progress: (8/20) | 5.08 s
[Task 2/25] Current/Best: 14.08/ 18.04 GFLOPS | Progress: (12/20) | 8.18 s
[Task 2/25] Current/Best: 15.84/ 18.04 GFLOPS | Progress: (16/20) | 9.75 s
[Task 2/25] Current/Best: 21.40/ 21.40 GFLOPS | Progress: (20/20) | 11.43 s Done.
-
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 12.81/ 16.43 GFLOPS | Progress: (4/20) | 4.14 s
[Task 3/25] Current/Best: 7.57/ 21.85 GFLOPS | Progress: (8/20) | 7.61 s
[Task 3/25] Current/Best: 23.12/ 23.12 GFLOPS | Progress: (12/20) | 9.65 s
[Task 3/25] Current/Best: 15.53/ 23.12 GFLOPS | Progress: (16/20) | 12.12 s
[Task 3/25] Current/Best: 20.97/ 23.12 GFLOPS | Progress: (20/20) | 14.47 s Done.
-
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 6.91/ 16.95 GFLOPS | Progress: (4/20) | 3.87 s
[Task 4/25] Current/Best: 9.03/ 16.95 GFLOPS | Progress: (8/20) | 9.03 s
[Task 4/25] Current/Best: 15.95/ 21.33 GFLOPS | Progress: (12/20) | 11.39 s
[Task 4/25] Current/Best: 8.79/ 21.33 GFLOPS | Progress: (16/20) | 14.81 s
[Task 4/25] Current/Best: 18.09/ 21.33 GFLOPS | Progress: (20/20) | 16.95 s Done.
-
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 4.11/ 7.08 GFLOPS | Progress: (4/20) | 4.50 s
[Task 5/25] Current/Best: 5.13/ 9.20 GFLOPS | Progress: (8/20) | 6.86 s
[Task 5/25] Current/Best: 8.93/ 17.46 GFLOPS | Progress: (12/20) | 9.37 s
[Task 5/25] Current/Best: 10.70/ 17.67 GFLOPS | Progress: (16/20) | 11.25 s
[Task 5/25] Current/Best: 17.99/ 18.08 GFLOPS | Progress: (20/20) | 13.35 s Done.
-
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 11.88/ 18.52 GFLOPS | Progress: (4/20) | 4.81 s
[Task 6/25] Current/Best: 5.39/ 18.52 GFLOPS | Progress: (8/20) | 8.95 s
[Task 6/25] Current/Best: 10.07/ 18.52 GFLOPS | Progress: (12/20) | 16.06 s
[Task 6/25] Current/Best: 13.51/ 18.52 GFLOPS | Progress: (16/20) | 19.53 s
[Task 6/25] Current/Best: 13.72/ 18.52 GFLOPS | Progress: (20/20) | 22.18 s Done.
-
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 18.51/ 18.63 GFLOPS | Progress: (4/20) | 3.81 s
[Task 7/25] Current/Best: 2.89/ 18.63 GFLOPS | Progress: (8/20) | 8.44 s
[Task 7/25] Current/Best: 7.61/ 21.97 GFLOPS | Progress: (12/20) | 11.37 s
[Task 7/25] Current/Best: 12.84/ 21.97 GFLOPS | Progress: (16/20) | 13.73 s
[Task 7/25] Current/Best: 13.62/ 21.97 GFLOPS | Progress: (20/20) | 16.26 s Done.
-
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 13.87/ 19.80 GFLOPS | Progress: (4/20) | 8.45 s
[Task 8/25] Current/Best: 15.39/ 19.80 GFLOPS | Progress: (8/20) | 19.93 s
[Task 8/25] Current/Best: 8.44/ 19.80 GFLOPS | Progress: (12/20) | 32.65 s
[Task 8/25] Current/Best: 15.91/ 19.80 GFLOPS | Progress: (16/20) | 35.58 s
[Task 8/25] Current/Best: 18.14/ 19.80 GFLOPS | Progress: (20/20) | 37.66 s
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 14.38/ 18.85 GFLOPS | Progress: (4/20) | 3.99 s
[Task 9/25] Current/Best: 10.96/ 19.47 GFLOPS | Progress: (8/20) | 10.49 s
[Task 9/25] Current/Best: 8.02/ 22.89 GFLOPS | Progress: (12/20) | 12.20 s
[Task 9/25] Current/Best: 17.43/ 22.89 GFLOPS | Progress: (16/20) | 23.36 s
[Task 9/25] Current/Best: 15.20/ 22.89 GFLOPS | Progress: (20/
20) | 26.23 s
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 15.41/ 15.41 GFLOPS | Progress: (4/20) | 5.15 s
[Task 10/25] Current/Best: 13.99/ 15.41 GFLOPS | Progress: (8/20) | 6.98 s
[Task 10/25] Current/Best: 5.98/ 19.03 GFLOPS | Progress: (12/20) | 8.62 s
[Task 10/25] Current/Best: 5.52/ 19.03 GFLOPS | Progress: (16/20) | 10.54 s
[Task 10/25] Current/Best: 6.84/ 19.03 GFLOPS | Progress: (20/20) | 13.62 s Done.
-
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 14.69/ 15.67 GFLOPS | Progress: (4/20) | 4.56 s
[Task 11/25] Current/Best: 11.08/ 19.49 GFLOPS | Progress: (8/20) | 7.28 s
[Task 11/25] Current/Best: 16.42/ 19.49 GFLOPS | Progress: (12/20) | 9.77 s
[Task 11/25] Current/Best: 22.74/ 22.74 GFLOPS | Progress: (16/20) | 11.82 s
[Task 11/25] Current/Best: 16.53/ 22.74 GFLOPS | Progress: (20/20) | 15.15 s Done.
-
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 14.93/ 21.23 GFLOPS | Progress: (4/20) | 3.85 s
[Task 12/25] Current/Best: 18.35/ 21.23 GFLOPS | Progress: (8/20) | 8.01 s
[Task 12/25] Current/Best: 4.93/ 21.23 GFLOPS | Progress: (12/20) | 10.79 s
[Task 12/25] Current/Best: 13.56/ 21.23 GFLOPS | Progress: (16/20) | 14.36 s
[Task 12/25] Current/Best: 14.15/ 21.23 GFLOPS | Progress: (20/20) | 18.07 s Done.
-
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 18.41/ 18.41 GFLOPS | Progress: (4/20) | 4.51 s
[Task 13/25] Current/Best: 17.14/ 18.41 GFLOPS | Progress: (8/20) | 7.96 s Done.
+
[Task 1/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 1/25] Current/Best: 9.56/ 23.57 GFLOPS | Progress: (4/20) | 7.00 s
[Task 1/25] Current/Best: 12.50/ 23.57 GFLOPS | Progress: (8/20) | 10.92 s
[Task 1/25] Current/Best: 22.45/ 23.57 GFLOPS | Progress: (12/20) | 13.14 s
[Task 1/25] Current/Best: 11.95/ 23.57 GFLOPS | Progress: (16/20) | 16.74 s
[Task 1/25] Current/Best: 13.59/ 23.57 GFLOPS | Progress: (20/20) | 19.96 s Done.
+
[Task 2/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 2/25] Current/Best: 5.73/ 14.76 GFLOPS | Progress: (4/20) | 3.74 s
[Task 2/25] Current/Best: 6.15/ 18.43 GFLOPS | Progress: (8/20) | 5.34 s
[Task 2/25] Current/Best: 8.63/ 18.43 GFLOPS | Progress: (12/20) | 6.73 s
[Task 2/25] Current/Best: 8.32/ 18.43 GFLOPS | Progress: (16/20) | 8.97 s
[Task 2/25] Current/Best: 11.74/ 18.43 GFLOPS | Progress: (20/20) | 11.04 s Done.
+
[Task 3/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 3/25] Current/Best: 8.64/ 17.15 GFLOPS | Progress: (4/20) | 4.13 s
[Task 3/25] Current/Best: 8.60/ 23.49 GFLOPS | Progress: (8/20) | 6.15 s
[Task 3/25] Current/Best: 13.94/ 23.49 GFLOPS | Progress: (12/20) | 8.53 s
[Task 3/25] Current/Best: 18.96/ 23.49 GFLOPS | Progress: (16/20) | 10.71 s
[Task 3/25] Current/Best: 15.77/ 23.49 GFLOPS | Progress: (20/20) | 13.07 s Done.
+
[Task 4/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 4/25] Current/Best: 7.44/ 17.98 GFLOPS | Progress: (4/20) | 3.91 s
[Task 4/25] Current/Best: 17.61/ 17.98 GFLOPS | Progress: (8/20) | 14.97 s
[Task 4/25] Current/Best: 8.52/ 17.98 GFLOPS | Progress: (12/20) | 20.90 s
[Task 4/25] Current/Best: 8.78/ 17.98 GFLOPS | Progress: (16/20) | 26.75 s
[Task 4/25] Current/Best: 8.31/ 18.02 GFLOPS | Progress: (20/20) | 28.38 s Done.
+
[Task 5/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 5/25] Current/Best: 10.72/ 13.76 GFLOPS | Progress: (4/20) | 4.23 s
[Task 5/25] Current/Best: 14.14/ 20.34 GFLOPS | Progress: (8/20) | 6.70 s
[Task 5/25] Current/Best: 12.38/ 20.34 GFLOPS | Progress: (12/20) | 9.40 s
[Task 5/25] Current/Best: 17.52/ 20.34 GFLOPS | Progress: (16/20) | 11.30 s
[Task 5/25] Current/Best: 12.91/ 20.71 GFLOPS | Progress: (20/20) | 13.47 s Done.
+
[Task 6/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 6/25] Current/Best: 11.85/ 21.73 GFLOPS | Progress: (4/20) | 5.23 s
[Task 6/25] Current/Best: 11.63/ 21.73 GFLOPS | Progress: (8/20) | 7.89 s
[Task 6/25] Current/Best: 5.33/ 21.73 GFLOPS | Progress: (12/20) | 11.76 s
[Task 6/25] Current/Best: 11.07/ 21.73 GFLOPS | Progress: (16/20) | 15.69 s
[Task 6/25] Current/Best: 3.13/ 21.73 GFLOPS | Progress: (20/20) | 19.56 s Done.
+
[Task 7/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 7/25] Current/Best: 8.86/ 14.12 GFLOPS | Progress: (4/20) | 4.06 s
[Task 7/25] Current/Best: 15.64/ 18.36 GFLOPS | Progress: (8/20) | 6.14 s
[Task 7/25] Current/Best: 19.25/ 19.25 GFLOPS | Progress: (12/20) | 8.73 s
[Task 7/25] Current/Best: 3.13/ 20.31 GFLOPS | Progress: (16/20) | 11.67 s
[Task 7/25] Current/Best: 17.99/ 20.31 GFLOPS | Progress: (20/20) | 13.85 s Done.
+
[Task 8/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 8/25] Current/Best: 18.70/ 18.70 GFLOPS | Progress: (4/20) | 5.19 s
[Task 8/25] Current/Best: 5.28/ 18.70 GFLOPS | Progress: (8/20) | 10.89 s
[Task 8/25] Current/Best: 8.89/ 18.70 GFLOPS | Progress: (12/20) | 17.73 s
[Task 8/25] Current/Best: 12.32/ 18.70 GFLOPS | Progress: (16/20) | 20.54 s
[Task 8/25] Current/Best: 7.69/ 19.13 GFLOPS | Progress: (20/20) | 24.65 s Done.
+
[Task 9/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 9/25] Current/Best: 8.89/ 13.03 GFLOPS | Progress: (4/20) | 9.14 s
[Task 9/25] Current/Best: 18.53/ 18.53 GFLOPS | Progress: (8/20) | 11.00 s
[Task 9/25] Current/Best: 18.22/ 18.53 GFLOPS | Progress: (12/20) | 12.70 s
[Task 9/25] Current/Best: 13.81/ 18.53 GFLOPS | Progress: (16/20) | 15.66 s
[Task 9/25] Current/Best: 12.51/ 18.53 GFLOPS | Progress: (20/20) | 22.83 s Done.
+
[Task 10/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 10/25] Current/Best: 15.36/ 15.36 GFLOPS | Progress: (4/20) | 4.36 s
[Task 10/25] Current/Best: 2.06/ 15.36 GFLOPS | Progress: (8/20) | 7.47 s
[Task 10/25] Current/Best: 13.65/ 18.75 GFLOPS | Progress: (12/20) | 9.03 s
[Task 10/25] Current/Best: 15.84/ 18.75 GFLOPS | Progress: (16/20) | 11.12 s
[Task 10/25] Current/Best: 10.67/ 20.94 GFLOPS | Progress: (20/20) | 13.18 s Done.
+
[Task 11/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 11/25] Current/Best: 14.48/ 19.82 GFLOPS | Progress: (4/20) | 4.25 s
[Task 11/25] Current/Best: 9.41/ 19.82 GFLOPS | Progress: (8/20) | 6.77 s
[Task 11/25] Current/Best: 15.42/ 19.82 GFLOPS | Progress: (12/20) | 9.04 s
[Task 11/25] Current/Best: 11.73/ 19.82 GFLOPS | Progress: (16/20) | 12.54 s
[Task 11/25] Current/Best: 10.79/ 19.82 GFLOPS | Progress: (20/20) | 14.51 s Done.
+
[Task 12/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 12/25] Current/Best: 13.40/ 19.24 GFLOPS | Progress: (4/20) | 4.12 s
[Task 12/25] Current/Best: 11.27/ 19.24 GFLOPS | Progress: (8/20) | 6.59 s
[Task 12/25] Current/Best: 13.35/ 19.24 GFLOPS | Progress: (12/20) | 10.29 s
[Task 12/25] Current/Best: 7.68/ 21.32 GFLOPS | Progress: (16/20) | 15.11 s
[Task 12/25] Current/Best: 10.07/ 21.32 GFLOPS | Progress: (20/20) | 18.25 s Done.
+
[Task 13/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 13/25] Current/Best: 10.19/ 17.25 GFLOPS | Progress: (4/20) | 4.85 s
[Task 13/25] Current/Best: 10.32/ 22.03 GFLOPS | Progress: (8/20) | 8.24 s
[Task 13/25] Current/Best: 17.18/ 22.03 GFLOPS | Progress: (12/20) | 11.90 s
[Task 13/25] Current/Best: 13.10/ 22.03 GFLOPS | Progress: (16/20) | 14.74 s
[Task 13/25] Current/Best: 19.42/ 22.03 GFLOPS | Progress: (20/20) | 18.02 s Done.
+
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 7.77/ 14.02 GFLOPS | Progress: (4/20) | 9.54 s
[Task 14/25] Current/Best: 15.21/ 15.21 GFLOPS | Progress: (8/20) | 12.28 s
[Task 14/25] Current/Best: 16.33/ 19.94 GFLOPS | Progress: (12/20) | 14.27 s
[Task 14/25] Current/Best: 17.02/ 19.94 GFLOPS | Progress: (16/20) | 17.30 s
[Task 14/25] Current/Best: 5.09/ 21.98 GFLOPS | Progress: (20/20) | 24.42 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 14.12/ 14.26 GFLOPS | Progress: (4/20) | 3.72 s
[Task 15/25] Current/Best: 15.23/ 18.81 GFLOPS | Progress: (8/20) | 5.24 s
[Task 15/25] Current/Best: 14.45/ 18.81 GFLOPS | Progress: (12/20) | 7.77 s
[Task 15/25] Current/Best: 13.58/ 18.81 GFLOPS | Progress: (16/20) | 10.75 s
[Task 15/25] Current/Best: 19.67/ 19.67 GFLOPS | Progress: (20/20
) | 12.81 s
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 17.44/ 20.49 GFLOPS | Progress: (4/20) | 3.92 s
[Task 16/25] Current/Best: 15.49/ 20.49 GFLOPS | Progress: (8/20) | 6.95 s Done.
Done.
-
[Task 13/25] Current/Best: 18.07/ 18.41 GFLOPS | Progress: (12/20) | 11.08 s
[Task 13/25] Current/Best: 18.37/ 18.41 GFLOPS | Progress: (16/20) | 13.83 s
[Task 13/25] Current/Best: 18.69/ 18.69 GFLOPS | Progress: (20/20) | 17.87 s Done.
-
[Task 14/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 14/25] Current/Best: 12.76/ 15.00 GFLOPS | Progress: (4/20) | 4.40 s
[Task 14/25] Current/Best: 1.62/ 18.90 GFLOPS | Progress: (8/20) | 9.22 s
[Task 14/25] Current/Best: 17.98/ 18.90 GFLOPS | Progress: (12/20) | 11.06 s
[Task 14/25] Current/Best: 8.61/ 18.90 GFLOPS | Progress: (16/20) | 14.32 s
[Task 14/25] Current/Best: 4.94/ 18.90 GFLOPS | Progress: (20/20) | 19.57 s
[Task 15/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 15/25] Current/Best: 6.10/ 18.60 GFLOPS | Progress: (4/20) | 3.63 s
[Task 15/25] Current/Best: 10.24/ 21.25 GFLOPS | Progress: (8/20) | 6.66 s
[Task 15/25] Current/Best: 6.81/ 21.25 GFLOPS | Progress: (12/20) | 12.09 s
[Task 15/25] Current/Best: 18.28/ 22.23 GFLOPS | Progress: (16/20) | 13.55 s
[Task 15/25] Current/Best: 6.04/ 22.23 GFLOPS | Progress: (20/20
) | 15.46 s
[Task 16/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 16/25] Current/Best: 17.08/ 17.08 GFLOPS | Progress: (4/20) | 4.27 s
[Task 16/25] Current/Best: 18.73/ 18.73 GFLOPS | Progress: (8/20) | 7.03 s
[Task 16/25] Current/Best: 13.38/ 18.73 GFLOPS | Progress: (12/20) | 8.70 s
[Task 16/25] Current/Best: 5.85/ 21.95 GFLOPS | Progress: (16/20) | 10.67 s
[Task 16/25] Current/Best: 18.19/ 21.95 GFLOPS | Progress: (20/20) | 13.26 s Done.
-
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 20.64/ 22.33 GFLOPS | Progress: (4/20) | 5.42 s
[Task 17/25] Current/Best: 19.03/ 23.07 GFLOPS | Progress: (8/20) | 7.77 s
[Task 17/25] Current/Best: 9.18/ 23.07 GFLOPS | Progress: (12/20) | 10.03 s
[Task 17/25] Current/Best: 9.65/ 23.07 GFLOPS | Progress: (16/20) | 12.17 s
[Task 17/25] Current/Best: 17.80/ 23.07 GFLOPS | Progress: (20/20) | 15.24 s Done.
-
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 6.57/ 19.81 GFLOPS | Progress: (4/20) | 3.86 s
[Task 18/25] Current/Best: 14.81/ 19.81 GFLOPS | Progress: (8/20) | 6.34 s
[Task 18/25] Current/Best: 14.98/ 20.74 GFLOPS | Progress: (12/20) | 8.12 s
[Task 18/25] Current/Best: 20.47/ 20.91 GFLOPS | Progress: (16/20) | 11.96 s
[Task 18/25] Current/Best: 17.23/ 20.91 GFLOPS | Progress: (20/20) | 18.21 s Done.
-
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 5.03/ 19.41 GFLOPS | Progress: (4/20) | 7.33 s
[Task 19/25] Current/Best: 5.33/ 19.41 GFLOPS | Progress: (8/20) | 11.17 s
[Task 19/25] Current/Best: 4.72/ 19.41 GFLOPS | Progress: (12/20) | 13.99 s
[Task 19/25] Current/Best: 21.84/ 21.84 GFLOPS | Progress: (16/20) | 16.20 s
[Task 19/25] Current/Best: 18.23/ 21.84 GFLOPS | Progress: (20/20) | 19.51 s Done.
-
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 12.48/ 16.94 GFLOPS | Progress: (4/20) | 3.78 s Done.
+
[Task 16/25] Current/Best: 10.59/ 20.49 GFLOPS | Progress: (12/20) | 8.72 s
[Task 16/25] Current/Best: 12.61/ 20.49 GFLOPS | Progress: (16/20) | 10.58 s
[Task 16/25] Current/Best: 11.62/ 20.49 GFLOPS | Progress: (20/20) | 13.08 s Done.
+
[Task 17/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 17/25] Current/Best: 17.77/ 17.77 GFLOPS | Progress: (4/20) | 5.58 s
[Task 17/25] Current/Best: 18.34/ 18.34 GFLOPS | Progress: (8/20) | 8.23 s
[Task 17/25] Current/Best: 6.54/ 18.34 GFLOPS | Progress: (12/20) | 11.70 s
[Task 17/25] Current/Best: 20.53/ 20.53 GFLOPS | Progress: (16/20) | 16.37 s
[Task 17/25] Current/Best: 12.09/ 20.53 GFLOPS | Progress: (20/20) | 19.03 s Done.
+
[Task 18/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 18/25] Current/Best: 4.96/ 8.10 GFLOPS | Progress: (4/20) | 5.81 s
[Task 18/25] Current/Best: 20.26/ 20.26 GFLOPS | Progress: (8/20) | 9.91 s
[Task 18/25] Current/Best: 20.58/ 20.58 GFLOPS | Progress: (12/20) | 14.94 s
[Task 18/25] Current/Best: 13.60/ 20.58 GFLOPS | Progress: (16/20) | 20.25 s
[Task 18/25] Current/Best: 13.24/ 20.58 GFLOPS | Progress: (20/20) | 24.22 s Done.
+
[Task 19/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 19/25] Current/Best: 19.62/ 19.62 GFLOPS | Progress: (4/20) | 3.81 s
[Task 19/25] Current/Best: 9.76/ 19.62 GFLOPS | Progress: (8/20) | 6.74 s
[Task 19/25] Current/Best: 9.06/ 19.62 GFLOPS | Progress: (12/20) | 10.52 s
[Task 19/25] Current/Best: 18.97/ 19.62 GFLOPS | Progress: (16/20) | 13.02 s
[Task 19/25] Current/Best: 14.03/ 19.62 GFLOPS | Progress: (20/20) | 16.31 s Done.
+
[Task 20/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 20/25] Current/Best: 5.06/ 17.30 GFLOPS | Progress: (4/20) | 3.72 s
[Task 20/25] Current/Best: 11.83/ 17.30 GFLOPS | Progress: (8/20) | 7.19 s
[Task 20/25] Current/Best: 16.47/ 17.30 GFLOPS | Progress: (12/20) | 9.37 s
[Task 20/25] Current/Best: 10.65/ 19.96 GFLOPS | Progress: (16/20) | 10.82 s
[Task 20/25] Current/Best: 5.36/ 19.96 GFLOPS | Progress: (20/20) | 14.75 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 2.74/ 19.65 GFLOPS | Progress: (4/20) | 3.88 s
[Task 21/25] Current/Best: 13.86/ 19.65 GFLOPS | Progress: (8/20) | 5.45 s
[Task 21/25] Current/Best: 11.67/ 19.65 GFLOPS | Progress: (12/20) | 8.84 s
[Task 21/25] Current/Best: 21.86/ 21.86 GFLOPS | Progress: (16/20) | 11.18 s
[Task 21/25] Current/Best: 14.36/ 21.86 GFLOPS | Progress: (20/20)
| 12.54 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 10.48/ 21.89 GFLOPS | Progress: (4/20) | 4.45 s Done.
Done.
-
[Task 20/25] Current/Best: 14.49/ 16.94 GFLOPS | Progress: (8/20) | 8.15 s
[Task 20/25] Current/Best: 1.58/ 16.94 GFLOPS | Progress: (12/20) | 12.17 s
[Task 20/25] Current/Best: 2.71/ 16.94 GFLOPS | Progress: (16/20) | 15.26 s
[Task 20/25] Current/Best: 9.69/ 16.94 GFLOPS | Progress: (20/20) | 17.75 s
[Task 21/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 21/25] Current/Best: 16.42/ 17.17 GFLOPS | Progress: (4/20) | 3.71 s
[Task 21/25] Current/Best: 7.42/ 19.28 GFLOPS | Progress: (8/20) | 7.27 s
[Task 21/25] Current/Best: 8.38/ 19.28 GFLOPS | Progress: (12/20) | 11.66 s
[Task 21/25] Current/Best: 16.97/ 19.28 GFLOPS | Progress: (16/20) | 13.41 s
[Task 21/25] Current/Best: 3.15/ 19.28 GFLOPS | Progress: (20/20) | 15.70 s
[Task 22/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 22/25] Current/Best: 14.92/ 20.97 GFLOPS | Progress: (4/2
0) | 3.72 s
[Task 22/25] Current/Best: 17.37/ 20.97 GFLOPS | Progress: (8/20) | 5.75 s
[Task 22/25] Current/Best: 6.19/ 20.97 GFLOPS | Progress: (12/20) | 10.20 s
[Task 22/25] Current/Best: 7.79/ 20.97 GFLOPS | Progress: (16/20) | 12.12 s
[Task 22/25] Current/Best: 10.78/ 20.97 GFLOPS | Progress: (20/20) | 14.77 s Done.
-
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 10.65/ 19.98 GFLOPS | Progress: (4/20) | 4.00 s
[Task 23/25] Current/Best: 9.15/ 20.08 GFLOPS | Progress: (8/20) | 6.87 s
[Task 23/25] Current/Best: 8.37/ 20.08 GFLOPS | Progress: (12/20) | 11.84 s
[Task 23/25] Current/Best: 1.55/ 24.13 GFLOPS | Progress: (16/20) | 15.54 s
[Task 23/25] Current/Best: 10.18/ 24.13 GFLOPS | Progress: (20/20) | 18.91 s Done.
-
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 3.77/ 6.66 GFLOPS | Progress: (4/20) | 4.87 s
[Task 24/25] Current/Best: 2.21/ 6.66 GFLOPS | Progress: (8/20) | 15.54 s
[Task 24/25] Current/Best: 1.90/ 6.66 GFLOPS | Progress: (12/20) | 27.05 s Done.
- Done.
-
[Task 24/25] Current/Best: 0.54/ 7.87 GFLOPS | Progress: (16/20) | 39.22 s
[Task 24/25] Current/Best: 5.93/ 7.87 GFLOPS | Progress: (20/20) | 43.21 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 25/25] Current/Best: 8.74/ 8.74 GFLOPS | Progress: (4/20) | 6.69 s
[Task 25/25] Current/Best: 1.55/ 8.74 GFLOPS | Progress: (8/20) | 17.37 s
[Task 25/25] Current/Best: 5.96/ 8.74 GFLOPS | Progress: (12/20) | 29.08 s
[Task 25/25] Current/Best: 8.18/ 8.74 GFLOPS | Progress: (16/20) | 31.76 s
[Task 25/25] Current/Best: 5.20/ 9.58 GFLOPS | Progress: (20/20) | 40.10 s
+
[Task 22/25] Current/Best: 19.45/ 21.89 GFLOPS | Progress: (8/20) | 7.28 s
[Task 22/25] Current/Best: 20.86/ 21.89 GFLOPS | Progress: (12/20) | 8.79 s
[Task 22/25] Current/Best: 18.93/ 21.89 GFLOPS | Progress: (16/20) | 10.37 s
[Task 22/25] Current/Best: 6.87/ 21.89 GFLOPS | Progress: (20/20) | 13.11 s Done.
+
[Task 23/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 23/25] Current/Best: 19.10/ 19.49 GFLOPS | Progress: (4/20) | 4.55 s
[Task 23/25] Current/Best: 11.91/ 19.49 GFLOPS | Progress: (8/20) | 7.56 s
[Task 23/25] Current/Best: 16.28/ 19.58 GFLOPS | Progress: (12/20) | 10.74 s
[Task 23/25] Current/Best: 5.39/ 19.58 GFLOPS | Progress: (16/20) | 15.31 s
[Task 23/25] Current/Best: 11.79/ 19.58 GFLOPS | Progress: (20/20) | 18.40 s Done.
+
[Task 24/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s
[Task 24/25] Current/Best: 3.79/ 5.57 GFLOPS | Progress: (4/20) | 12.38 s
[Task 24/25] Current/Best: 1.88/ 5.57 GFLOPS | Progress: (8/20) | 19.57 s
[Task 24/25] Current/Best: 1.18/ 5.57 GFLOPS | Progress: (12/20) | 24.14 s
[Task 24/25] Current/Best: 4.04/ 7.02 GFLOPS | Progress: (16/20) | 35.07 s
[Task 24/25] Current/Best: 3.95/ 7.02 GFLOPS | Progress: (20/20) | 46.85 s
[Task 25/25] Current/Best: 0.00/ 0.00 GFLOPS | Progress: (0/20) | 0.00 s Done.
+
[Task 25/25] Current/Best: 3.04/ 9.81 GFLOPS | Progress: (4/20) | 5.22 s
[Task 25/25] Current/Best: 8.38/ 9.81 GFLOPS | Progress: (8/20) | 7.77 s
[Task 25/25] Current/Best: 9.42/ 9.81 GFLOPS | Progress: (12/20) | 18.43 s
[Task 25/25] Current/Best: 5.86/ 9.81 GFLOPS | Progress: (16/20) | 20.93 s
[Task 25/25] Current/Best: 10.04/ 10.04 GFLOPS | Progress: (20/20) | 31.87 s
@@ -678,7 +679,7 @@ Verify that the optimized model runs and produces the same results:
.. code-block:: none
class='n02123045 tabby, tabby cat' with probability=0.621104
- class='n02123159 tiger cat' with probability=0.356378
+ class='n02123159 tiger cat' with probability=0.356379
class='n02124075 Egyptian cat' with probability=0.019712
class='n02129604 tiger, Panthera tigris' with probability=0.001215
class='n04040759 radiator' with probability=0.000262
@@ -735,8 +736,8 @@ improvement in comparing the optimized model to the unoptimized model.
.. code-block:: none
- optimized: {'mean': 407.68921118999515, 'median': 406.46522925001136, 'std': 3.361531823863263}
- unoptimized: {'mean': 510.46957689999545, 'median': 509.98423024998374, 'std': 1.8196836642888992}
+ optimized: {'mean': 421.2595205099933, 'median': 421.0890523499984, 'std': 3.3985261511904916}
+ unoptimized: {'mean': 509.2969556299795, 'median': 509.13396689993533, 'std': 2.566962968850706}
@@ -759,7 +760,7 @@ profiling/benchmarking.
.. rst-class:: sphx-glr-timing
- **Total running time of the script:** ( 11 minutes 48.111 seconds)
+ **Total running time of the script:** ( 11 minutes 26.170 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 a159189997..bef51be12c 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
- 1.229e-07 secs/op
+ 1.24e-07 secs/op
diff --git a/docs/_sources/tutorial/intro_topi.rst.txt b/docs/_sources/tutorial/intro_topi.rst.txt
index f0b60ffb0a..6134dc8f70 100644
--- a/docs/_sources/tutorial/intro_topi.rst.txt
+++ b/docs/_sources/tutorial/intro_topi.rst.txt
@@ -264,7 +264,7 @@ As you can see, scheduled stages of computation have been accumulated and we can
.. code-block:: none
- [stage(a, placeholder(a, 0x1af4ec10)), stage(b, placeholder(b, 0x216d5720)), stage(T_add, compute(T_add, body=[(a[ax0, ax1, ax2] + b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min=0, ext=10))], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[(a[ax0, ax1, ax2]*b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(mi [...]
+ [stage(a, placeholder(a, 0x110be910)), stage(b, placeholder(b, 0x8ee5d10)), stage(T_add, compute(T_add, body=[(a[ax0, ax1, ax2] + b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min=0, ext=10))], reduce_axis=[], tag=broadcast, attrs={})), stage(T_multiply, compute(T_multiply, body=[(a[ax0, ax1, ax2]*b[ax1, ax2])], axis=[iter_var(ax0, range(min=0, ext=100)), iter_var(ax1, range(min=0, ext=10)), iter_var(ax2, range(min [...]
diff --git a/docs/_sources/tutorial/sg_execution_times.rst.txt b/docs/_sources/tutorial/sg_execution_times.rst.txt
index f8af853361..05c52885ef 100644
--- a/docs/_sources/tutorial/sg_execution_times.rst.txt
+++ b/docs/_sources/tutorial/sg_execution_times.rst.txt
@@ -5,32 +5,32 @@
Computation times
=================
-**15:00.811** total execution time for **tutorial** files:
+**14:34.881** total execution time for **tutorial** files:
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 11:48.111 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_relay_x86.py` (``autotvm_relay_x86.py``) | 11:26.170 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 01:18.257 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_auto_scheduler_matmul_x86.py` (``auto_scheduler_matmul_x86.py``) | 01:17.980 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 00:58.485 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_expr_get_started.py` (``tensor_expr_get_started.py``) | 00:57.789 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:33.351 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_relay_quick_start.py` (``relay_quick_start.py``) | 00:33.403 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:21.023 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_autotvm_matmul_x86.py` (``autotvm_matmul_x86.py``) | 00:17.946 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.810 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_intro_topi.py` (``intro_topi.py``) | 00:00.828 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:00.601 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_tensor_ir_blitz_course.py` (``tensor_ir_blitz_course.py``) | 00:00.600 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.164 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_cross_compilation_and_rpc.py` (``cross_compilation_and_rpc.py``) | 00:00.156 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_introduction.py` (``introduction.py``) | 00:00.006 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_uma.py` (``uma.py``) | 00:00.001 | 0.0 MB |
+| :ref:`sphx_glr_tutorial_uma.py` (``uma.py``) | 00:00.002 | 0.0 MB |
++------------------------------------------------------------------------------------------+-----------+--------+
+| :ref:`sphx_glr_tutorial_install.py` (``install.py``) | 00:00.001 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_tvmc_command_line_driver.py` (``tvmc_command_line_driver.py``) | 00:00.001 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
| :ref:`sphx_glr_tutorial_tvmc_python.py` (``tvmc_python.py``) | 00:00.001 | 0.0 MB |
+------------------------------------------------------------------------------------------+-----------+--------+
-| :ref:`sphx_glr_tutorial_install.py` (``install.py``) | 00:00.001 | 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 341e6fd5a2..380da64a6a 100644
--- a/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
+++ b/docs/_sources/tutorial/tensor_expr_get_started.rst.txt
@@ -299,7 +299,7 @@ helper function to run a profile of the TVM generated code.
.. code-block:: none
Numpy running time: 0.000007
- naive: 0.000009
+ naive: 0.000007
@@ -397,7 +397,7 @@ compile and run this new schedule with the parallel operation applied:
.. code-block:: none
- parallel: 0.000008
+ parallel: 0.000007
@@ -503,10 +503,10 @@ We can now compare the different schedules
.. code-block:: none
Operator Timing Performance
- numpy 7.333700000344834e-06 1.0
- naive 8.653299999999999e-06 1.1799364576670872
- parallel 8.0871e-06 1.1027312270231588
- vector 2.45758e-05 3.351077900492853
+ numpy 6.5418699887231925e-06 1.0
+ naive 6.633400000000001e-06 1.0139914139893618
+ parallel 7.002600000000001e-06 1.0704278764437403
+ vector 2.4646299999999998e-05 3.7674701641098087
@@ -927,7 +927,7 @@ matrix multiplication.
.. code-block:: none
- Numpy running time: 0.018009
+ Numpy running time: 0.017432
@@ -985,7 +985,7 @@ optimizations.
.. code-block:: none
- none: 3.227641
+ none: 3.188975
@@ -1087,7 +1087,7 @@ schedule.
.. code-block:: none
- blocking: 0.302309
+ blocking: 0.295732
@@ -1182,7 +1182,7 @@ already cache friendly from our previous optimizations.
.. code-block:: none
- vectorization: 0.332865
+ vectorization: 0.331842
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1024, 1024], []),
@@ -1255,7 +1255,7 @@ more cache friendly.
.. code-block:: none
- loop permutation: 0.114747
+ loop permutation: 0.113142
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1024, 1024], []),
@@ -1353,7 +1353,7 @@ optimized schedule.
.. code-block:: none
- array packing: 0.107988
+ array packing: 0.107231
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1024, 1024], []),
@@ -1445,7 +1445,7 @@ to `C` when all the block results are ready.
.. code-block:: none
- block caching: 0.109868
+ block caching: 0.110199
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1024, 1024], []),
@@ -1530,7 +1530,7 @@ of thread-level parallelization.
.. code-block:: none
- parallelization: 0.145995
+ parallelization: 0.145666
@main = primfn(A_1: handle, B_1: handle, C_1: handle) -> ()
attr = {"from_legacy_te_schedule": True, "global_symbol": "main", "tir.noalias": True}
buffers = {A: Buffer(A_2: Pointer(float32), float32, [1024, 1024], []),
@@ -1610,13 +1610,13 @@ working, we can compare the results.
.. code-block:: none
Operator Timing Performance
- none 3.2276409815999996 1.0
- blocking 0.3023087391 0.09366244288735612
- vectorization 0.3328645036 0.10312934601387824
- loop permutation 0.11474695480000001 0.03555133778947059
- array packing 0.1079881577 0.03345730157586124
- block caching 0.1098684765 0.034039869095210284
- parallelization 0.1459951633 0.04523277654865678
+ none 3.1889749383 1.0
+ blocking 0.2957320337 0.09273576601315368
+ vectorization 0.3318421705 0.1040591967389059
+ loop permutation 0.11314202640000001 0.035479120591745544
+ array packing 0.10723093129999998 0.03362551709395476
+ block caching 0.1101992959 0.0345563380183683
+ parallelization 0.14566599249999998 0.045677998516241894
diff --git a/docs/commit_hash b/docs/commit_hash
index 27764c3f3c..e2f3fcfac9 100644
--- a/docs/commit_hash
+++ b/docs/commit_hash
@@ -1 +1 @@
-77b6f0eec32a10d2ed716bf71a68e59bb97eefc1
+15e185d922b4de567aa2f74c71aedbc0b56952df
diff --git a/docs/genindex.html b/docs/genindex.html
index cda18469d4..00e80ff623 100644
--- a/docs/genindex.html
+++ b/docs/genindex.html
@@ -3264,10 +3264,12 @@
</li>
<li><a href="reference/api/python/topi.html#tvm.topi.nn.qnn_add_alter_layout">qnn_add_alter_layout() (in module tvm.topi.nn)</a>
</li>
- <li><a href="reference/api/python/topi.html#tvm.topi.nn.qnn_requantize_alter_layout">qnn_requantize_alter_layout() (in module tvm.topi.nn)</a>
+ <li><a href="reference/api/python/topi.html#tvm.topi.nn.qnn_conv2d_alter_layout">qnn_conv2d_alter_layout() (in module tvm.topi.nn)</a>
</li>
</ul></td>
<td style="width: 33%; vertical-align: top;"><ul>
+ <li><a href="reference/api/python/topi.html#tvm.topi.nn.qnn_requantize_alter_layout">qnn_requantize_alter_layout() (in module tvm.topi.nn)</a>
+</li>
<li><a href="reference/api/python/relay/frontend.html#tvm.relay.frontend.quantize_conv_bias_mkldnn_from_var">quantize_conv_bias_mkldnn_from_var() (in module tvm.relay.frontend)</a>
</li>
<li><a href="reference/api/python/auto_scheduler.html#tvm.auto_scheduler.ApplyHistoryBestOrSample.query">query() (tvm.auto_scheduler.ApplyHistoryBestOrSample method)</a>
diff --git a/docs/how_to/compile_models/from_darknet.html b/docs/how_to/compile_models/from_darknet.html
index 9da86bd296..acb5f1c722 100644
--- a/docs/how_to/compile_models/from_darknet.html
+++ b/docs/how_to/compile_models/from_darknet.html
@@ -585,7 +585,7 @@ 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 9.382 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 9.878 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_keras.html b/docs/how_to/compile_models/from_keras.html
index 367032ffbe..485364a160 100644
--- a/docs/how_to/compile_models/from_keras.html
+++ b/docs/how_to/compile_models/from_keras.html
@@ -506,7 +506,7 @@ Tensorflow is also required since it’s used as the default backend of keras.</
<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Relay top-1 id: 285, class name: Egyptian cat
1/1 [==============================] - ETA: 0s
-1/1 [==============================] - 1s 902ms/step
+1/1 [==============================] - 1s 926ms/step
Keras top-1 id: 285, class name: Egyptian cat
</pre></div>
</div>
diff --git a/docs/how_to/compile_models/from_mxnet.html b/docs/how_to/compile_models/from_mxnet.html
index 9f9daccd58..28fabd7c88 100644
--- a/docs/how_to/compile_models/from_mxnet.html
+++ b/docs/how_to/compile_models/from_mxnet.html
@@ -439,7 +439,7 @@
<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.zip92ab5ae2-94bf-44fe-a1cc-83745f42673d 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.zipa59764d6-5791-488f-843d-46863756023f 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 9bdf183d42..8a463f7ef4 100644
--- a/docs/how_to/compile_models/from_oneflow.html
+++ b/docs/how_to/compile_models/from_oneflow.html
@@ -449,13 +449,13 @@ 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]
- 19%|#9 | 7.99M/41.5M [00:00<00:00, 51.8MB/s]
- 35%|###4 | 14.3M/41.5M [00:00<00:00, 55.7MB/s]
- 48%|####7 | 19.7M/41.5M [00:00<00:00, 47.6MB/s]
- 59%|#####8 | 24.3M/41.5M [00:00<00:00, 37.4MB/s]
- 77%|#######7 | 32.0M/41.5M [00:00<00:00, 45.9MB/s]
- 92%|#########2| 38.3M/41.5M [00:00<00:00, 46.0MB/s]
-100%|##########| 41.5M/41.5M [00:00<00:00, 47.1MB/s]
+ 19%|#9 | 7.99M/41.5M [00:00<00:00, 45.2MB/s]
+ 35%|###5 | 14.6M/41.5M [00:00<00:00, 55.9MB/s]
+ 49%|####9 | 20.3M/41.5M [00:00<00:00, 53.3MB/s]
+ 62%|######1 | 25.6M/41.5M [00:00<00:00, 45.2MB/s]
+ 82%|########2 | 34.1M/41.5M [00:00<00:00, 48.4MB/s]
+ 96%|#########6| 40.0M/41.5M [00:00<00:00, 48.8MB/s]
+100%|##########| 41.5M/41.5M [00:00<00:00, 50.4MB/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 52e2e097f5..39aeea1669 100644
--- a/docs/how_to/compile_models/from_pytorch.html
+++ b/docs/how_to/compile_models/from_pytorch.html
@@ -432,10 +432,12 @@ 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]
- 32%|###1 | 14.2M/44.7M [00:00<00:00, 149MB/s]
- 64%|######3 | 28.4M/44.7M [00:00<00:00, 115MB/s]
- 89%|########9 | 39.9M/44.7M [00:00<00:00, 109MB/s]
-100%|##########| 44.7M/44.7M [00:00<00:00, 112MB/s]
+ 18%|#7 | 7.99M/44.7M [00:00<00:00, 72.4MB/s]
+ 36%|###5 | 16.0M/44.7M [00:00<00:00, 68.9MB/s]
+ 54%|#####3 | 24.0M/44.7M [00:00<00:00, 71.6MB/s]
+ 72%|#######1 | 32.0M/44.7M [00:00<00:00, 67.2MB/s]
+ 90%|########9 | 40.0M/44.7M [00:00<00:00, 62.6MB/s]
+100%|##########| 44.7M/44.7M [00:00<00:00, 70.2MB/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 ef8f354b46..61b98defee 100644
--- a/docs/how_to/compile_models/from_tensorflow.html
+++ b/docs/how_to/compile_models/from_tensorflow.html
@@ -649,7 +649,7 @@ 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 10.094 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 9.853 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 84a49406b7..b2a6e0937a 100644
--- a/docs/how_to/compile_models/sg_execution_times.html
+++ b/docs/how_to/compile_models/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>05:34.227</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
+<p><strong>05:38.333</strong> total execution time for <strong>how_to_compile_models</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 81%" />
@@ -348,44 +348,44 @@
<col style="width: 8%" />
</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>01:10.094</p></td>
+<tr class="row-odd"><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>01:09.878</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>01:09.382</p></td>
+<tr class="row-even"><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>01:09.853</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>00:45.021</p></td>
+<td><p>00:46.139</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>00:30.859</p></td>
+<td><p>00:30.729</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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>00:27.563</p></td>
+<td><p>00:28.029</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>00:25.941</p></td>
+<td><p>00:25.966</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>00:24.930</p></td>
+<td><p>00:25.881</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>00:21.729</p></td>
+<td><p>00:22.036</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>00:16.284</p></td>
+<td><p>00:17.415</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>00:02.424</p></td>
+<td><p>00:02.408</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 63d9fa73fa..6a7805ebeb 100644
--- a/docs/how_to/deploy_models/deploy_model_on_adreno.html
+++ b/docs/how_to/deploy_models/deploy_model_on_adreno.html
@@ -920,10 +920,9 @@ Top5 predictions:
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 3342.9183 3342.3090 3350.8926 3339.9926 3.0670
+ 2541.4099 2539.7964 2553.2709 2539.0355 4.0725
</pre></div>
</div>
-<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 0.711 seconds)</p>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-deploy-models-deploy-model-on-adreno-py">
<div class="sphx-glr-download sphx-glr-download-python docutils container">
<p><a class="reference download internal" download="" href="../../_downloads/2387d8448da213eb625e6b3d916327d4/deploy_model_on_adreno.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_model_on_adreno.py</span></code></a></p>
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 e66319b1ea..450a8591dc 100644
--- a/docs/how_to/deploy_models/deploy_model_on_android.html
+++ b/docs/how_to/deploy_models/deploy_model_on_android.html
@@ -662,7 +662,7 @@ to the remote android device.</p>
Evaluate inference time cost...
Execution time summary:
mean (ms) median (ms) max (ms) min (ms) std (ms)
- 15.6403 15.5040 16.0189 15.4519 0.2030
+ 15.7695 15.6253 16.5552 15.4373 0.3572
</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 dc194f9c2c..386748f52b 100644
--- a/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
+++ b/docs/how_to/deploy_models/deploy_object_detection_pytorch.html
@@ -454,21 +454,32 @@ 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]
- 6%|5 | 10.1M/170M [00:00<00:01, 105MB/s]
- 12%|#1 | 20.0M/170M [00:00<00:02, 63.0MB/s]
- 19%|#8 | 32.0M/170M [00:00<00:02, 71.2MB/s]
- 28%|##8 | 48.0M/170M [00:00<00:01, 94.2MB/s]
- 36%|###6 | 61.9M/170M [00:00<00:01, 109MB/s]
- 43%|####3 | 73.1M/170M [00:00<00:01, 97.0MB/s]
- 52%|#####1 | 87.8M/170M [00:00<00:00, 112MB/s]
- 58%|#####8 | 99.3M/170M [00:01<00:00, 94.4MB/s]
- 64%|######4 | 109M/170M [00:01<00:00, 88.0MB/s]
- 70%|####### | 119M/170M [00:01<00:00, 92.1MB/s]
- 76%|#######5 | 129M/170M [00:01<00:00, 86.4MB/s]
- 83%|########3 | 141M/170M [00:01<00:00, 97.6MB/s]
- 89%|########8 | 151M/170M [00:01<00:00, 98.6MB/s]
- 95%|#########4| 161M/170M [00:01<00:00, 91.1MB/s]
-100%|##########| 170M/170M [00:01<00:00, 95.2MB/s]
+ 4%|3 | 6.30M/170M [00:00<00:03, 46.7MB/s]
+ 7%|7 | 12.7M/170M [00:00<00:02, 56.9MB/s]
+ 11%|# | 18.3M/170M [00:00<00:03, 42.2MB/s]
+ 14%|#4 | 24.0M/170M [00:00<00:03, 41.8MB/s]
+ 19%|#8 | 32.0M/170M [00:00<00:02, 51.4MB/s]
+ 24%|##3 | 40.0M/170M [00:00<00:02, 60.1MB/s]
+ 28%|##8 | 48.0M/170M [00:00<00:02, 59.9MB/s]
+ 35%|###4 | 58.9M/170M [00:01<00:01, 74.4MB/s]
+ 39%|###9 | 66.5M/170M [00:01<00:01, 64.2MB/s]
+ 43%|####3 | 73.1M/170M [00:01<00:01, 58.9MB/s]
+ 47%|####7 | 80.0M/170M [00:01<00:01, 55.3MB/s]
+ 51%|##### | 86.3M/170M [00:01<00:01, 54.8MB/s]
+ 54%|#####4 | 91.7M/170M [00:01<00:01, 50.1MB/s]
+ 57%|#####6 | 96.7M/170M [00:01<00:01, 49.9MB/s]
+ 60%|###### | 102M/170M [00:02<00:01, 47.7MB/s]
+ 63%|######2 | 107M/170M [00:02<00:01, 40.7MB/s]
+ 66%|######6 | 112M/170M [00:02<00:01, 44.4MB/s]
+ 71%|####### | 120M/170M [00:02<00:01, 47.5MB/s]
+ 75%|#######5 | 128M/170M [00:02<00:00, 47.4MB/s]
+ 80%|######## | 136M/170M [00:02<00:00, 48.3MB/s]
+ 85%|########4 | 144M/170M [00:02<00:00, 55.5MB/s]
+ 88%|########8 | 150M/170M [00:03<00:00, 45.9MB/s]
+ 91%|#########1| 155M/170M [00:03<00:00, 45.4MB/s]
+ 94%|#########4| 160M/170M [00:03<00:00, 41.6MB/s]
+ 98%|#########7| 166M/170M [00:03<00:00, 39.4MB/s]
+100%|##########| 170M/170M [00:03<00:00, 49.8MB/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=& [...]
@@ -566,7 +577,7 @@ 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 8.729 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 3 minutes 8.162 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 58b21fe9a5..4388ca141f 100644
--- a/docs/how_to/deploy_models/deploy_prequantized.html
+++ b/docs/how_to/deploy_models/deploy_prequantized.html
@@ -498,9 +498,8 @@ 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, 49.2MB/s]
- 99%|#########9| 13.4M/13.6M [00:00<00:00, 52.7MB/s]
-100%|##########| 13.6M/13.6M [00:00<00:00, 52.3MB/s]
+ 59%|#####8 | 7.99M/13.6M [00:00<00:00, 51.8MB/s]
+100%|##########| 13.6M/13.6M [00:00<00:00, 62.8MB/s]
</pre></div>
</div>
</div>
@@ -591,7 +590,7 @@ 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)
- 90.0495 89.9418 94.4827 89.7411 0.5090
+ 90.1727 90.0634 92.2951 89.8462 0.3375
</pre></div>
</div>
<div class="admonition note">
@@ -630,7 +629,7 @@ 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 4.623 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 4.723 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 20b3a64f52..e44c322e21 100644
--- a/docs/how_to/deploy_models/deploy_prequantized_tflite.html
+++ b/docs/how_to/deploy_models/deploy_prequantized_tflite.html
@@ -583,7 +583,7 @@ 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)
- 117.7106 117.3661 123.0387 115.9515 1.2004
+ 118.2884 118.1312 120.6947 116.4210 1.1491
</pre></div>
</div>
<div class="admonition note">
@@ -611,7 +611,7 @@ 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 27.708 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 2 minutes 20.990 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 aeaed13120..154b8c721e 100644
--- a/docs/how_to/deploy_models/deploy_quantized.html
+++ b/docs/how_to/deploy_models/deploy_quantized.html
@@ -521,7 +521,7 @@ 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 31.171 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 34.558 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 4cc13fa7e1..132efcb0f4 100644
--- a/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
+++ b/docs/how_to/deploy_models/deploy_ssd_gluoncv.html
@@ -463,23 +463,24 @@ 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]
- 4%|3 | 5232/132723 [00:00<00:02, 52305.96KB/s]
- 10%|# | 13353/132723 [00:00<00:01, 69300.86KB/s]
- 16%|#6 | 21478/132723 [00:00<00:01, 74754.60KB/s]
- 22%|##2 | 29789/132723 [00:00<00:01, 78050.79KB/s]
- 29%|##8 | 38078/132723 [00:00<00:01, 79794.79KB/s]
- 35%|###4 | 46433/132723 [00:00<00:01, 81069.42KB/s]
- 41%|####1 | 54729/132723 [00:00<00:00, 81685.77KB/s]
- 48%|####7 | 63047/132723 [00:00<00:00, 82159.47KB/s]
- 54%|#####3 | 71375/132723 [00:00<00:00, 82509.03KB/s]
- 60%|###### | 79818/132723 [00:01<00:00, 83094.06KB/s]
- 66%|######6 | 88128/132723 [00:01<00:00, 81084.51KB/s]
- 73%|#######2 | 96247/132723 [00:01<00:00, 80809.41KB/s]
- 79%|#######8 | 104335/132723 [00:01<00:00, 80800.48KB/s]
- 85%|########4 | 112420/132723 [00:01<00:00, 80797.92KB/s]
- 91%|######### | 120504/132723 [00:01<00:00, 80759.95KB/s]
- 97%|#########6| 128583/132723 [00:01<00:00, 80639.96KB/s]
-100%|##########| 132723/132723 [00:01<00:00, 79602.00KB/s]
+ 4%|4 | 5468/132723 [00:00<00:02, 54601.03KB/s]
+ 10%|# | 13312/132723 [00:00<00:01, 68609.69KB/s]
+ 15%|#5 | 20174/132723 [00:00<00:02, 45614.49KB/s]
+ 21%|##1 | 28173/132723 [00:00<00:01, 56008.52KB/s]
+ 27%|##6 | 35532/132723 [00:00<00:01, 61311.98KB/s]
+ 33%|###2 | 43500/132723 [00:00<00:01, 66842.14KB/s]
+ 39%|###8 | 51360/132723 [00:00<00:01, 70378.10KB/s]
+ 45%|####4 | 59412/132723 [00:00<00:00, 73424.00KB/s]
+ 51%|##### | 67408/132723 [00:01<00:00, 75386.50KB/s]
+ 57%|#####6 | 75477/132723 [00:01<00:00, 76976.96KB/s]
+ 63%|######2 | 83553/132723 [00:01<00:00, 78110.97KB/s]
+ 69%|######8 | 91530/132723 [00:01<00:00, 78607.72KB/s]
+ 75%|#######4 | 99480/132723 [00:01<00:00, 78874.32KB/s]
+ 81%|######## | 107450/132723 [00:01<00:00, 79115.81KB/s]
+ 87%|########7 | 115488/132723 [00:01<00:00, 79494.38KB/s]
+ 93%|#########3| 123603/132723 [00:01<00:00, 79988.94KB/s]
+ 99%|#########9| 131921/132723 [00:01<00:00, 80942.53KB/s]
+100%|##########| 132723/132723 [00:01<00:00, 72855.81KB/s]
</pre></div>
</div>
<p>Create TVM runtime and do inference
@@ -518,7 +519,7 @@ 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 2.839 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 3.144 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 a0a21ac054..a05f0dc4f6 100644
--- a/docs/how_to/deploy_models/sg_execution_times.html
+++ b/docs/how_to/deploy_models/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>13:40.032</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
+<p><strong>13:25.645</strong> total execution time for <strong>how_to_deploy_models</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 86%" />
@@ -349,39 +349,39 @@
</colgroup>
<tbody>
<tr class="row-odd"><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>03:08.729</p></td>
+<td><p>03:08.162</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>03:02.839</p></td>
+<td><p>03:03.144</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>02:27.708</p></td>
+<td><p>02:20.990</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>01:31.171</p></td>
+<td><p>01:34.558</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>01:04.623</p></td>
+<td><p>01:04.723</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 Adreno</span></a> (<code class="docutils literal notranslate"><span class="pre">deploy_model_on_adreno.py</span></code>)</p></td>
-<td><p>01:00.711</p></td>
+<td><p>00:50.923</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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>00:34.561</p></td>
+<td><p>00:34.624</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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>00:25.021</p></td>
+<td><p>00:24.451</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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>00:24.662</p></td>
+<td><p>00:24.065</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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 d1a077a955..1cdf34d0af 100644
--- a/docs/how_to/extend_tvm/bring_your_own_datatypes.html
+++ b/docs/how_to/extend_tvm/bring_your_own_datatypes.html
@@ -622,7 +622,7 @@ 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.zip9677fdde-9c12-4bd6-aefe-38a5f844cd59 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.zip1440c57f-f2d6-4d23-9bae-eed2f8ecf36f 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 1fc819a0a4..896f12c7da 100644
--- a/docs/how_to/extend_tvm/sg_execution_times.html
+++ b/docs/how_to/extend_tvm/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>00:46.389</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
+<p><strong>00:45.776</strong> total execution time for <strong>how_to_extend_tvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -349,15 +349,15 @@
</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>00:43.064</p></td>
+<td><p>00:42.488</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>00:02.331</p></td>
+<td><p>00:02.290</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>00:00.987</p></td>
+<td><p>00:00.990</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>
diff --git a/docs/how_to/extend_tvm/use_pass_instrument.html b/docs/how_to/extend_tvm/use_pass_instrument.html
index 2d83d04c3a..ded96048f9 100644
--- a/docs/how_to/extend_tvm/use_pass_instrument.html
+++ b/docs/how_to/extend_tvm/use_pass_instrument.html
@@ -526,10 +526,10 @@ 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: 7133us [7133us] (46.57%; 46.57%)
-FoldScaleAxis: 8185us [6us] (53.43%; 53.43%)
- FoldConstant: 8178us [1644us] (53.39%; 99.92%)
- InferType: 6534us [6534us] (42.66%; 79.89%)
+InferType: 7311us [7311us] (46.93%; 46.93%)
+FoldScaleAxis: 8269us [8us] (53.07%; 53.07%)
+ FoldConstant: 8262us [1697us] (53.03%; 99.91%)
+ InferType: 6565us [6565us] (42.13%; 79.46%)
</pre></div>
</div>
</div>
@@ -551,10 +551,10 @@ 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: 6580us [6580us] (44.96%; 44.96%)
-FoldScaleAxis: 8057us [4us] (55.04%; 55.04%)
- FoldConstant: 8053us [1616us] (55.02%; 99.95%)
- InferType: 6436us [6436us] (43.97%; 79.93%)
+InferType: 6603us [6603us] (44.94%; 44.94%)
+FoldScaleAxis: 8091us [5us] (55.06%; 55.06%)
+ FoldConstant: 8086us [1686us] (55.03%; 99.94%)
+ InferType: 6400us [6400us] (43.55%; 79.15%)
</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 2f03911f85..dcf6577c70 100644
--- a/docs/how_to/optimize_operators/opt_conv_cuda.html
+++ b/docs/how_to/optimize_operators/opt_conv_cuda.html
@@ -578,7 +578,7 @@ 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: 38.573760 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Convolution: 54.223007 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 113aef08dc..6b03e79b6e 100644
--- a/docs/how_to/optimize_operators/opt_conv_tensorcore.html
+++ b/docs/how_to/optimize_operators/opt_conv_tensorcore.html
@@ -915,7 +915,7 @@ 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: 13.346890 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>conv2d with tensor core: 11.942332 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 958b7d088a..4738e79255 100644
--- a/docs/how_to/optimize_operators/opt_gemm.html
+++ b/docs/how_to/optimize_operators/opt_gemm.html
@@ -475,8 +475,8 @@ 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: 0.018447
-Baseline: 3.199606
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Numpy running time: 0.017736
+Baseline: 3.189127
</pre></div>
</div>
<p>In TVM, we can always inspect lower level IR to debug or optimize our schedule.
@@ -535,7 +535,7 @@ 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: 0.291128
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt1: 0.295809
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -601,7 +601,7 @@ 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: 0.330108
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt2: 0.330921
</pre></div>
</div>
<p>Here is the generated IR after vectorization.</p>
@@ -661,7 +661,7 @@ 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: 0.115935
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt3: 0.113154
</pre></div>
</div>
<p>Here is the generated IR after loop permutation.</p>
@@ -743,7 +743,7 @@ 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: 0.109205
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt4: 0.109051
</pre></div>
</div>
<p>Here is the generated IR after array packing.</p>
@@ -828,7 +828,7 @@ 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: 0.110408
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt5: 0.111148
</pre></div>
</div>
<p>Here is the generated IR after blocking.</p>
@@ -917,7 +917,7 @@ write to C when all the block results are ready.</p>
<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: 0.146240
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Opt6: 0.146006
</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 4b8c60e31e..a2c5213eef 100644
--- a/docs/how_to/optimize_operators/sg_execution_times.html
+++ b/docs/how_to/optimize_operators/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>00:34.144</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
+<p><strong>00:34.187</strong> total execution time for <strong>how_to_optimize_operators</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -349,15 +349,15 @@
</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>00:31.424</p></td>
+<td><p>00:31.378</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>00:01.566</p></td>
+<td><p>00:01.581</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>00:01.154</p></td>
+<td><p>00:01.227</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 28cba46d8b..a143ac3fdb 100644
--- a/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
+++ b/docs/how_to/tune_with_autoscheduler/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>08:53.425</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
+<p><strong>08:58.603</strong> total execution time for <strong>how_to_tune_with_autoscheduler</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 85%" />
@@ -349,27 +349,27 @@
</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>05:29.809</p></td>
+<td><p>05:26.025</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>01:30.541</p></td>
+<td><p>01:30.113</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>01:00.857</p></td>
+<td><p>01:00.903</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>00:29.525</p></td>
+<td><p>00:38.892</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>00:11.758</p></td>
+<td><p>00:11.784</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>00:10.936</p></td>
+<td><p>00:10.885</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 cb4fdf7311..dd481941b1 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
@@ -504,12 +504,12 @@ cooperative fetching, unrolling and operator fusion.</p>
bias: Buffer(bias_2: Pointer(float32), float32, [1, 512, 1, 1], []),
compute: Buffer(compute_2: Pointer(float32), float32, [1, 512, 7, 7], [])}
buffer_map = {data_1: data, kernel_1: kernel, bias_1: bias, compute_1: compute} {
- attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 64;
- allocate(conv2d_nchw: Pointer(local float32), float32, [8]), storage_scope = local;
- allocate(pad_temp.shared: Pointer(shared float32), float32, [4032]), storage_scope = shared;
+ attr [IterVar(blockIdx.x: int32, (nullptr), "ThreadIndex", "blockIdx.x")] "thread_extent" = 16;
+ allocate(conv2d_nchw: Pointer(local float32), float32, [14]), storage_scope = local;
+ allocate(pad_temp.shared: Pointer(shared float32), float32, [1008]), storage_scope = shared;
allocate(kernel.shared: Pointer(shared float32), float32, [1536]), storage_scope = shared;
- attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- conv2d_nchw_1: Buffer(conv2d_nchw, float32, [8], [], scope="local", align=32)[0] = 0f32
+ attr [IterVar(threadIdx.x: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112 {
+ conv2d_nchw_1: Buffer(conv2d_nchw, float32, [1], [], scope="local", align=4)[0] = 0f32
conv2d_nchw_1[1] = 0f32
conv2d_nchw_1[2] = 0f32
conv2d_nchw_1[3] = 0f32
@@ -517,258 +517,754 @@ cooperative fetching, unrolling and operator fusion.</p>
conv2d_nchw_1[5] = 0f32
conv2d_nchw_1[6] = 0f32
conv2d_nchw_1[7] = 0f32
- for (rc.outer.outer: int32, 0, 8) {
+ conv2d_nchw_1[8] = 0f32
+ conv2d_nchw_1[9] = 0f32
+ conv2d_nchw_1[10] = 0f32
+ conv2d_nchw_1[11] = 0f32
+ conv2d_nchw_1[12] = 0f32
+ conv2d_nchw_1[13] = 0f32
+ for (rc.outer.outer: int32, 0, 32) {
for (rx.outer.outer: int32, 0, 3) {
- let cse_var_2: int32 = (rc.outer.outer*3136)
- let cse_var_1: int32 = (rc.outer.outer*576)
+ let cse_var_2: int32 = (rc.outer.outer*784)
+ let cse_var_1: int32 = (rc.outer.outer*144)
{
- attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1: Buffer(pad_temp.shared, float32, [4032], [], scope="shared")[(threadIdx.x_1*3)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3: Buffer(data_2, float32, [25088], [])[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx. [...]
- pad_temp.shared_1[((threadIdx.x_1*3) + 1)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) - 7)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) - 6)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 49), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floor [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 49), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 49), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 98), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floor [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 98), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 98), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 441)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 335)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 442)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 336)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 443)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 337)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 196), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 196), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 196), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 245), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 245), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 245), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 882)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 678)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 883)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 679)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 884)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 680)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 343), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 343), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 343), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 392), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 392), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 392), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 1323)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1021)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 1324)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1022)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 1325)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1023)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 490), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 490), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 490), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 539), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 539), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 539), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 1764)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1364)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 1765)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1365)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 1766)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1366)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 637), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 637), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 637), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 686), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 686), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 686), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 2205)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1707)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2206)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1708)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2207)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 1709)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 784), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 784), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 784), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 833), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 833), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 833), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 2646)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2050)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2647)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2051)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 2648)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2052)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 931), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 931), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 931), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 980), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (floo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 980), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 980), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3087)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2393)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 3088)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2394)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 3089)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2395)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1078), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (flo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1078), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) &l [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1078), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) &l [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1127), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (flo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1127), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) &l [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1127), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) &l [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3528)] = @tir.if_then_else(((((2 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2736)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 3529)] = @tir.if_then_else(((((2 <= floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2737)], 0f32, dtype=float32)
- pad_temp.shared_1[((threadIdx.x_1*3) + 3530)] = @tir.if_then_else(((((1 < floormod(threadIdx.x_1, 21)) && (floormod(threadIdx.x_1, 21) < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[((((cse_var_2 + (floordiv(threadIdx.x_1, 21)*49)) + (floormod(threadIdx.x_1, 21)*3)) + rx.outer.outer) + 2738)], 0f32, dtype=float32)
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1225), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (flo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1225), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) &l [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1225), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) &l [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1274), 21)*63) + (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)*7)) + floormod((threadIdx.x_1*3), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9)) && (floormod((floordiv((threadIdx.x_1*3), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((((cse_var_2 + (flo [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1274), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 1), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 1), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) &l [...]
- pad_temp.shared_1[(((floordiv((threadIdx.x_1 + 1274), 21)*63) + (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)*7)) + floormod(((threadIdx.x_1*3) + 2), 7))] = @tir.if_then_else(((((1 <= floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9)) && (floormod((floordiv(((threadIdx.x_1*3) + 2), 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) &l [...]
- }
- attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49 {
- if @tir.likely((threadIdx.x_1 < 21), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3969)] = @tir.if_then_else(((((2 < threadIdx.x_1) && (threadIdx.x_1 < 19)) && (1 <= (rx.outer.outer + floormod((threadIdx.x_1*3), 7)))) && ((rx.outer.outer + floormod((threadIdx.x_1*3), 7)) < 8)), data_3[(((cse_var_2 + (threadIdx.x_1*3)) + rx.outer.outer) + 3079)], 0f32, dtype=float32)
- }
- if @tir.likely((threadIdx.x_1 < 21), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3970)] = @tir.if_then_else(((((2 <= threadIdx.x_1) && (threadIdx.x_1 < 19)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 1), 7)) < 8)), data_3[(((cse_var_2 + ((threadIdx.x_1*3) + 1)) + rx.outer.outer) + 3079)], 0f32, dtype=float32)
- }
- if @tir.likely((threadIdx.x_1 < 21), dtype=bool) {
- pad_temp.shared_1[((threadIdx.x_1*3) + 3971)] = @tir.if_then_else(((((1 < threadIdx.x_1) && (threadIdx.x_1 < 18)) && (1 <= (rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)))) && ((rx.outer.outer + floormod(((threadIdx.x_1*3) + 2), 7)) < 8)), data_3[(((cse_var_2 + ((threadIdx.x_1*3) + 2)) + rx.outer.outer) + 3079)], 0f32, dtype=float32)
- }
- }
- attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1: Buffer(kernel.shared, float32, [1536], [], scope="shared")[threadIdx.x_2] = kernel_3: Buffer(kernel_2, float32, [2359296], [])[((((blockIdx.x*36864) + cse_var_1) + (threadIdx.x_2*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 49)] = kernel_3[(((((blockIdx.x*36864) + cse_var_1) + (floordiv((threadIdx.x_2 + 49), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 98)] = kernel_3[(((((blockIdx.x*36864) + cse_var_1) + (floordiv((threadIdx.x_2 + 98), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 147)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 147), 192)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 49), 64)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 196)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 196), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 4), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 245)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 245), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 53), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 294)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 294), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 34)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 343)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 343), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 151), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 392)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 392), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 8), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 441)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 441), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 19)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 490)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 490), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 106), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 539)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 539), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 155), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 588)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 588), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 4)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 637)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 637), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 61), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 686)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 686), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 110), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 735)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 735), 192)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 53), 64)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 784)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 784), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 833)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 833), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 65), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 882)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 882), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 38)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 931)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 931), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 163), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 980)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 980), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 20), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1029)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1029), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 23)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1078)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1078), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 118), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1127)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1127), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 167), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1176)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1176), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 8)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1225)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1225), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 73), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1274)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1274), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 122), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1323)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1323), 192)*4608)) + cse_var_1) + (floormod((floordiv(threadIdx.x_2, 3) + 57), 64)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1372)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1372), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 28), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1421)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1421), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 77), 192), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- kernel.shared_1[(threadIdx.x_2 + 1470)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1470), 192)*4608)) + cse_var_1) + ((floordiv(threadIdx.x_2, 3) + 42)*9)) + (floormod(threadIdx.x_2, 3)*3)) + rx.outer.outer)]
- attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 49;
- if @tir.likely((threadIdx.x_2 < 17), dtype=bool) {
- kernel.shared_1[(threadIdx.x_2 + 1519)] = kernel_3[((((((blockIdx.x*36864) + (floordiv((threadIdx.x_2 + 1519), 192)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 175), 192), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
- }
- for (rc.inner: int32, 0, 64) {
- let cse_var_3: int32 = (rc.inner*3)
- {
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[cse_var_3]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 192)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 384)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 576)]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 768)]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 960)]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 1152)]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((rc.inner*63) + threadIdx.x)]*kernel.shared_1[(cse_var_3 + 1344)]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 1)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 193)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 385)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 577)]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 769)]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 961)]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 1153)]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 7)]*kernel.shared_1[(cse_var_3 + 1345)]))
- conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 2)]))
- conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 194)]))
- conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 386)]))
- conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 578)]))
- conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 770)]))
- conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 962)]))
- conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 1154)]))
- conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(((rc.inner*63) + threadIdx.x) + 14)]*kernel.shared_1[(cse_var_3 + 1346)]))
- }
+ attr [IterVar(threadIdx.x_1: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1: Buffer(pad_temp.shared, float32, [1008], [], scope="shared")[threadIdx.x_1] = @tir.if_then_else(((((7 <= floormod(threadIdx.x_1, 63)) && (floormod(threadIdx.x_1, 63) < 56)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3: Buffer(data_2, float32, [25088], [])[((((cse_var_2 + (floordiv(threadIdx.x_1, 63)*49)) + rx.outer.outer) + floormod(thread [...]
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 112)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 7), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 7), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 112), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 7), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x [...]
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 224)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 5), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 5), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 224), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 5), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x [...]
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 336)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 3), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 3), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 336), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 3), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x [...]
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 448)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 1), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 1), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 448), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 1), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x [...]
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 560)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 8), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 8), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 560), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 8), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x [...]
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 672)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 6), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 6), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 672), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 6), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x [...]
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 784)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 4), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 4), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 784), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 4), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x [...]
+ attr [IterVar(threadIdx.x_1, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ pad_temp.shared_1[(threadIdx.x_1 + 896)] = @tir.if_then_else(((((1 <= floormod((floordiv(threadIdx.x_1, 7) + 2), 9)) && (floormod((floordiv(threadIdx.x_1, 7) + 2), 9) < 8)) && (1 <= (rx.outer.outer + floormod(threadIdx.x_1, 7)))) && ((rx.outer.outer + floormod(threadIdx.x_1, 7)) < 8)), data_3[(((((cse_var_2 + (floordiv((threadIdx.x_1 + 896), 63)*49)) + (floormod((floordiv(threadIdx.x_1, 7) + 2), 9)*7)) + rx.outer.outer) + floormod(threadIdx.x [...]
+ attr [IterVar(threadIdx.x_2: int32, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1: Buffer(kernel.shared, float32, [1536], [], scope="shared")[threadIdx.x_2] = kernel_3: Buffer(kernel_2, float32, [2359296], [])[(((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 112)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 112), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 224)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 224), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 48), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 336)] = kernel_3[((((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer) + 32256)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 448)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 448), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 560)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 560), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 48), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 672)] = kernel_3[((((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer) + 64512)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 784)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 784), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 896)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 896), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 48), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1008)] = kernel_3[((((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer) + 96768)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1120)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 1120), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1232)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 1232), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 32), 48), 3)*9)) + (floormod((threadIdx.x_2 + 2), 3)*3)) + rx.outer.outer)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ kernel.shared_1[(threadIdx.x_2 + 1344)] = kernel_3[((((((blockIdx.x*147456) + (floordiv(threadIdx.x_2, 48)*4608)) + cse_var_1) + (floormod(threadIdx.x_2, 48)*3)) + rx.outer.outer) + 129024)]
+ attr [IterVar(threadIdx.x_2, (nullptr), "ThreadIndex", "threadIdx.x")] "thread_extent" = 112;
+ if @tir.likely((threadIdx.x_2 < 80), dtype=bool) {
+ kernel.shared_1[(threadIdx.x_2 + 1456)] = kernel_3[((((((blockIdx.x*147456) + (floordiv((threadIdx.x_2 + 1456), 48)*4608)) + cse_var_1) + (floordiv(floormod((threadIdx.x_2 + 16), 48), 3)*9)) + (floormod((threadIdx.x_2 + 1), 3)*3)) + rx.outer.outer)]
}
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[(floormod(threadIdx.x, 7)*7)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 1)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 2)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 3)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 4)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 5)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 6)]*kernel.shared_1[(floordiv(threadIdx.x, 7)*48)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[(floormod(threadIdx.x, 7)*7)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 1)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 2)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 3)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 4)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 5)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 6)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 768)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 7)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 8)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 9)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 10)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 11)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 12)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 13)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 1)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 7)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 8)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 9)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 10)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 11)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 12)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 13)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 769)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 14)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 15)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 16)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 17)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 18)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 19)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 20)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 2)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 14)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 15)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 16)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 17)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 18)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 19)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 20)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 770)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 63)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 64)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 65)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 66)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 67)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 68)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 69)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 3)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 63)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 64)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 65)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 66)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 67)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 68)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 69)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 771)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 70)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 71)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 72)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 73)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 74)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 75)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 76)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 4)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 70)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 71)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 72)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 73)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 74)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 75)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 76)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 772)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 77)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 78)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 79)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 80)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 81)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 82)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 83)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 5)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 77)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 78)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 79)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 80)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 81)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 82)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 83)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 773)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 126)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 127)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 128)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 129)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 130)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 131)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 132)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 6)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 126)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 127)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 128)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 129)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 130)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 131)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 132)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 774)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 133)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 134)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 135)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 136)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 137)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 138)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 139)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 7)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 133)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 134)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 135)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 136)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 137)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 138)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 139)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 775)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 140)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 141)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 142)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 143)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 144)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 145)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 146)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 8)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 140)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 141)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 142)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 143)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 144)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 145)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 146)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 776)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 189)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 190)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 191)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 192)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 193)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 194)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 195)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 9)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 189)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 190)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 191)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 192)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 193)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 194)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 195)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 777)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 196)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 197)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 198)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 199)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 200)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 201)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 202)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 10)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 196)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 197)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 198)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 199)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 200)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 201)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 202)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 778)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 203)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 204)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 205)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 206)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 207)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 208)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 209)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 11)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 203)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 204)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 205)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 206)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 207)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 208)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 209)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 779)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 252)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 253)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 254)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 255)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 256)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 257)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 258)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 12)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 252)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 253)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 254)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 255)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 256)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 257)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 258)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 780)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 259)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 260)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 261)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 262)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 263)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 264)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 265)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 13)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 259)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 260)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 261)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 262)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 263)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 264)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 265)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 781)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 266)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 267)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 268)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 269)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 270)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 271)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 272)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 14)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 266)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 267)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 268)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 269)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 270)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 271)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 272)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 782)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 315)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 316)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 317)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 318)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 319)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 320)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 321)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 15)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 315)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 316)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 317)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 318)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 319)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 320)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 321)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 783)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 322)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 323)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 324)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 325)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 326)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 327)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 328)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 16)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 322)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 323)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 324)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 325)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 326)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 327)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 328)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 784)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 329)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 330)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 331)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 332)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 333)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 334)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 335)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 17)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 329)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 330)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 331)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 332)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 333)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 334)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 335)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 785)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 378)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 379)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 380)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 381)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 382)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 383)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 384)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 18)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 378)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 379)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 380)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 381)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 382)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 383)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 384)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 786)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 385)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 386)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 387)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 388)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 389)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 390)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 391)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 19)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 385)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 386)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 387)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 388)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 389)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 390)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 391)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 787)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 392)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 393)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 394)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 395)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 396)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 397)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 398)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 20)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 392)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 393)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 394)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 395)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 396)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 397)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 398)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 788)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 441)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 442)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 443)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 444)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 445)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 446)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 447)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 21)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 441)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 442)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 443)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 444)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 445)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 446)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 447)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 789)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 448)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 449)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 450)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 451)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 452)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 453)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 454)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 22)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 448)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 449)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 450)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 451)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 452)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 453)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 454)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 790)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 455)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 456)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 457)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 458)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 459)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 460)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 461)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 23)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 455)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 456)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 457)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 458)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 459)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 460)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 461)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 791)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 504)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 505)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 506)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 507)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 508)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 509)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 510)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 24)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 504)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 505)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 506)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 507)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 508)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 509)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 510)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 792)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 511)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 512)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 513)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 514)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 515)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 516)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 517)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 25)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 511)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 512)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 513)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 514)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 515)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 516)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 517)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 793)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 518)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 519)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 520)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 521)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 522)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 523)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 524)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 26)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 518)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 519)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 520)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 521)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 522)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 523)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 524)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 794)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 567)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 568)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 569)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 570)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 571)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 572)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 573)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 27)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 567)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 568)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 569)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 570)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 571)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 572)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 573)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 795)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 574)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 575)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 576)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 577)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 578)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 579)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 580)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 28)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 574)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 575)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 576)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 577)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 578)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 579)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 580)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 796)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 581)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 582)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 583)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 584)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 585)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 586)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 587)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 29)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 581)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 582)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 583)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 584)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 585)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 586)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 587)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 797)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 630)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 631)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 632)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 633)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 634)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 635)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 636)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 30)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 630)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 631)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 632)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 633)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 634)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 635)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 636)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 798)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 637)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 638)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 639)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 640)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 641)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 642)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 643)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 31)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 637)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 638)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 639)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 640)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 641)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 642)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 643)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 799)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 644)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 645)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 646)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 647)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 648)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 649)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 650)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 32)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 644)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 645)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 646)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 647)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 648)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 649)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 650)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 800)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 693)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 694)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 695)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 696)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 697)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 698)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 699)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 33)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 693)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 694)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 695)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 696)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 697)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 698)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 699)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 801)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 700)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 701)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 702)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 703)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 704)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 705)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 706)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 34)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 700)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 701)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 702)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 703)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 704)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 705)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 706)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 802)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 707)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 708)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 709)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 710)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 711)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 712)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 713)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 35)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 707)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 708)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 709)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 710)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 711)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 712)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 713)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 803)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 756)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 757)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 758)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 759)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 760)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 761)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 762)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 36)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 756)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 757)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 758)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 759)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 760)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 761)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 762)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 804)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 763)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 764)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 765)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 766)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 767)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 768)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 769)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 37)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 763)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 764)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 765)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 766)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 767)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 768)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 769)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 805)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 770)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 771)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 772)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 773)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 774)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 775)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 776)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 38)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 770)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 771)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 772)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 773)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 774)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 775)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 776)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 806)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 819)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 820)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 821)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 822)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 823)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 824)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 825)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 39)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 819)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 820)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 821)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 822)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 823)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 824)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 825)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 807)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 826)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 827)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 828)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 829)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 830)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 831)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 832)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 40)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 826)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 827)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 828)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 829)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 830)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 831)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 832)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 808)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 833)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 834)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 835)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 836)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 837)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 838)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 839)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 41)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 833)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 834)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 835)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 836)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 837)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 838)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 839)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 809)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 882)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 883)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 884)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 885)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 886)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 887)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 888)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 42)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 882)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 883)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 884)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 885)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 886)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 887)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 888)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 810)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 889)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 890)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 891)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 892)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 893)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 894)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 895)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 43)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 889)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 890)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 891)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 892)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 893)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 894)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 895)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 811)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 896)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 897)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 898)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 899)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 900)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 901)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 902)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 44)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 896)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 897)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 898)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 899)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 900)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 901)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 902)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 812)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 945)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 946)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 947)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 948)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 949)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 950)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 951)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 45)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 945)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 946)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 947)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 948)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 949)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 950)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 951)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 813)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 952)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 953)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 954)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 955)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 956)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 957)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 958)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 46)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 952)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 953)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 954)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 955)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 956)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 957)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 958)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 814)]))
+ conv2d_nchw_1[0] = (conv2d_nchw_1[0] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 959)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[1] = (conv2d_nchw_1[1] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 960)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[2] = (conv2d_nchw_1[2] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 961)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[3] = (conv2d_nchw_1[3] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 962)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[4] = (conv2d_nchw_1[4] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 963)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[5] = (conv2d_nchw_1[5] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 964)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[6] = (conv2d_nchw_1[6] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 965)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 47)]))
+ conv2d_nchw_1[7] = (conv2d_nchw_1[7] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 959)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[8] = (conv2d_nchw_1[8] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 960)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[9] = (conv2d_nchw_1[9] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 961)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[10] = (conv2d_nchw_1[10] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 962)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[11] = (conv2d_nchw_1[11] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 963)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[12] = (conv2d_nchw_1[12] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 964)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
+ conv2d_nchw_1[13] = (conv2d_nchw_1[13] + (pad_temp.shared_1[((floormod(threadIdx.x, 7)*7) + 965)]*kernel.shared_1[((floordiv(threadIdx.x, 7)*48) + 815)]))
}
}
}
- for (i1.inner: int32, 0, 8) {
- compute_3: Buffer(compute_2, float32, [25088], [])[(((blockIdx.x*392) + (i1.inner*49)) + threadIdx.x)] = max((conv2d_nchw_1[i1.inner] + bias_3: Buffer(bias_2, float32, [512], [])[((blockIdx.x*8) + i1.inner)]), 0f32)
- }
+ compute_3: Buffer(compute_2, float32, [25088], [])[((blockIdx.x*1568) + (threadIdx.x*7))] = max((conv2d_nchw_1[0] + bias_3: Buffer(bias_2, float32, [512], [])[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 1)] = max((conv2d_nchw_1[1] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 2)] = max((conv2d_nchw_1[2] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 3)] = max((conv2d_nchw_1[3] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 4)] = max((conv2d_nchw_1[4] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 5)] = max((conv2d_nchw_1[5] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 6)] = max((conv2d_nchw_1[6] + bias_3[((blockIdx.x*32) + floordiv(threadIdx.x, 7))]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 784)] = max((conv2d_nchw_1[7] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 785)] = max((conv2d_nchw_1[8] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 786)] = max((conv2d_nchw_1[9] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 787)] = max((conv2d_nchw_1[10] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 788)] = max((conv2d_nchw_1[11] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 789)] = max((conv2d_nchw_1[12] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
+ compute_3[(((blockIdx.x*1568) + (threadIdx.x*7)) + 790)] = max((conv2d_nchw_1[13] + bias_3[(((blockIdx.x*32) + floordiv(threadIdx.x, 7)) + 16)]), 0f32)
}
}
</pre></div>
@@ -804,7 +1300,7 @@ cooperative fetching, unrolling and operator fusion.</p>
<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: 0.289 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 0.388 ms
</pre></div>
</div>
</div>
@@ -833,37 +1329,37 @@ conv2d_nchw_nn_o_i, conv2d_nchw_nn_i = s[conv2d_nchw].split(conv2d_nchw_nn, fact
conv2d_nchw_nn_o_o_i, conv2d_nchw_nn_o_i = s[conv2d_nchw].split(conv2d_nchw_nn_o_i, factor=1)
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=8)
+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=1)
-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_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=2)
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=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=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=64)
-conv2d_nchw_rc_o_o, conv2d_nchw_rc_o_i = s[conv2d_nchw].split(conv2d_nchw_rc_o_i, factor=1)
-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_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_o, conv2d_nchw_xx_o_o_o_i = s[conv2d_nchw].split(conv2d_nchw_xx_o_o_o_i, factor=7)
+conv2d_nchw_rc_o_i, conv2d_nchw_rc_i = s[conv2d_nchw].split(conv2d_nchw_rc, factor=1)
+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_ry_o_o, conv2d_nchw_ry_o_i = s[conv2d_nchw].split(conv2d_nchw_ry_o_i, factor=3)
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)
s[conv2d_nchw].reorder(conv2d_nchw_nn_o_o_o_o, conv2d_nchw_ff_o_o_o_o, conv2d_nchw_yy_o_o_o_o, conv2d_nchw_xx_o_o_o_o, conv2d_nchw_nn_o_o_o_i, conv2d_nchw_ff_o_o_o_i, conv2d_nchw_yy_o_o_o_i, conv2d_nchw_xx_o_o_o_i, conv2d_nchw_nn_o_o_i, conv2d_nchw_ff_o_o_i, conv2d_nchw_yy_o_o_i, conv2d_nchw_xx_o_o_i, conv2d_nchw_rc_o_o, conv2d_nchw_ry_o_o, conv2d_nchw_rx_o_o, conv2d_nchw_rc_o_i, conv2d_nchw_ry_o_i, conv2d_nchw_rx_o_i, conv2d_nchw_nn_o_i, conv2d_nchw_ff_o_i, conv2d_nchw_yy_o_i, 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=8)
-compute_i1_o_o_i, compute_i1_o_i = s[compute].split(compute_i1_o_i, factor=1)
-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=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=2)
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=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)
+compute_i3_o_o_i, compute_i3_o_i = s[compute].split(compute_i3_o_i, factor=1)
+compute_i3_o_o_o, compute_i3_o_o_i = s[compute].split(compute_i3_o_o_i, factor=7)
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)
kernel_shared = s.cache_read(kernel, "shared", [conv2d_nchw])
@@ -882,12 +1378,12 @@ s[compute].bind(compute_i0_o_i_i1_o_i_fused_i2_o_i_fused_i3_o_i_fused, te.thread
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=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=49)
+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)
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=3)
+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)
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=49)
+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)
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", 1024)
s[conv2d_nchw].pragma(conv2d_nchw_nn_o_o_o_o, "unroll_explicit", True)
@@ -907,9 +1403,9 @@ CUDA source code:
#define int64_t long long
#define uint64_t unsigned long long
#endif
-extern "C" __global__ void __launch_bounds__(49) 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];
+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[14];
+ __shared__ float pad_temp_shared[1008];
__shared__ float kernel_shared[1536];
conv2d_nchw[0] = 0.000000e+00f;
conv2d_nchw[1] = 0.000000e+00f;
@@ -919,165 +1415,729 @@ extern "C" __global__ void __launch_bounds__(49) default_function_kern
conv2d_nchw[5] = 0.000000e+00f;
conv2d_nchw[6] = 0.000000e+00f;
conv2d_nchw[7] = 0.000000e+00f;
- for (int rc_outer_outer = 0; rc_outer_outer < 8; ++rc_outer_outer) {
+ conv2d_nchw[8] = 0.000000e+00f;
+ conv2d_nchw[9] = 0.000000e+00f;
+ conv2d_nchw[10] = 0.000000e+00f;
+ conv2d_nchw[11] = 0.000000e+00f;
+ conv2d_nchw[12] = 0.000000e+00f;
+ conv2d_nchw[13] = 0.000000e+00f;
+ for (int rc_outer_outer = 0; rc_outer_outer < 32; ++rc_outer_outer) {
for (int rx_outer_outer = 0; rx_outer_outer < 3; ++rx_outer_outer) {
__syncthreads();
- pad_temp_shared[(((int)threadIdx.x) * 3)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) - 8)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) - 7)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) - 6)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 49) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 49) / 21) * 4 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 49) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ( [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 49) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ( [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 98) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 98) / 21) * 4 [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 98) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ( [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 98) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + ( [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 441)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 335)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 442)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 336)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 443)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 337)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 196) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 196) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 196) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 196) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 245) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 245) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 245) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 245) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 882)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 678)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 883)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 679)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 884)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 680)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 343) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 343) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 343) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 343) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 392) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 392) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 392) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 392) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1323)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1021)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1324)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1022)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1325)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1023)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 490) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 490) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 490) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 490) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 539) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 539) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 539) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 539) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1764)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1364)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1765)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1365)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 1766)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1366)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 637) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 637) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 637) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 637) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 686) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 686) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 686) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 686) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2205)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1707)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2206)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1708)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2207)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 1709)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 784) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 784) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 784) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 784) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 833) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 833) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 833) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 833) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2646)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2050)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2647)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2051)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 2648)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2052)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 931) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 931) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 931) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 931) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 980) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 980) / 21) * [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 980) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 980) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3087)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2393)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3088)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2394)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3089)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2395)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 1078) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1078) / 21) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1078) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1078) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1127) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1127) / 21) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1127) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1127) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3528)] = (((((2 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2736)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3529)] = (((((2 <= (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2737)] : 0.000000e+00f);
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3530)] = (((((1 < (((int)threadIdx.x) % 21)) && ((((int)threadIdx.x) % 21) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[(((((rc_outer_outer * 3136) + ((((int)threadIdx.x) / 21) * 49)) + ((((int)threadIdx.x) % 21) * 3)) + rx_outer_outer) + 2738)] : 0.000000e+00f);
- pad_temp_shared[(((((((int)threadIdx.x) + 1225) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 3) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1225) / 21) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1225) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1225) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1274) / 21) * 63) + (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) * 7)) + ((((int)threadIdx.x) * 3) % 7))] = (((((1 <= ((((((int)threadIdx.x) * 3) / 7) + 6) % 9)) && (((((((int)threadIdx.x) * 3) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + (((((int)threadIdx.x) + 1274) / 21) [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1274) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 1) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 1) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- pad_temp_shared[(((((((int)threadIdx.x) + 1274) / 21) * 63) + ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) * 7)) + (((((int)threadIdx.x) * 3) + 2) % 7))] = (((((1 <= (((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9)) && ((((((((int)threadIdx.x) * 3) + 2) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((((rc_outer_outer * 3136) + [...]
- if (((int)threadIdx.x) < 21) {
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3969)] = (((((2 < ((int)threadIdx.x)) && (((int)threadIdx.x) < 19)) && (1 <= (rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)))) && ((rx_outer_outer + ((((int)threadIdx.x) * 3) % 7)) < 8)) ? data[((((rc_outer_outer * 3136) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 3079)] : 0.000000e+00f);
- }
- if (((int)threadIdx.x) < 21) {
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3970)] = (((((2 <= ((int)threadIdx.x)) && (((int)threadIdx.x) < 19)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 1) % 7)) < 8)) ? data[((((rc_outer_outer * 3136) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 3080)] : 0.000000e+00f);
- }
- if (((int)threadIdx.x) < 21) {
- pad_temp_shared[((((int)threadIdx.x) * 3) + 3971)] = (((((1 < ((int)threadIdx.x)) && (((int)threadIdx.x) < 18)) && (1 <= (rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)))) && ((rx_outer_outer + (((((int)threadIdx.x) * 3) + 2) % 7)) < 8)) ? data[((((rc_outer_outer * 3136) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 3081)] : 0.000000e+00f);
- }
- kernel_shared[((int)threadIdx.x)] = kernel[((((((int)blockIdx.x) * 36864) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 49)] = kernel[(((((((int)blockIdx.x) * 36864) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 49) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 98)] = kernel[(((((((int)blockIdx.x) * 36864) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 98) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 147)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 147) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) / 3) + 49) & 63) * 9)) + ((((int)threadIdx.x) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 196)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 196) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 4) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 245)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 245) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 53) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 294)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 294) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 306)];
- kernel_shared[(((int)threadIdx.x) + 343)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 343) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) + 151) % 192) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 392)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 392) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 8) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 441)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 441) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 171)];
- kernel_shared[(((int)threadIdx.x) + 490)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 490) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 106) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 539)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 539) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) + 155) % 192) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 588)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 588) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 36)];
- kernel_shared[(((int)threadIdx.x) + 637)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 637) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 61) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 686)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 686) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 110) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 735)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 735) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) / 3) + 53) & 63) * 9)) + ((((int)threadIdx.x) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 784) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 16) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 833)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 833) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 65) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 882)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 882) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 342)];
- kernel_shared[(((int)threadIdx.x) + 931)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 931) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) + 163) % 192) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 980)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 980) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 20) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1029)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1029) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 207)];
- kernel_shared[(((int)threadIdx.x) + 1078)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1078) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 118) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1127)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1127) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) + 167) % 192) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1176)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1176) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 72)];
- kernel_shared[(((int)threadIdx.x) + 1225)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1225) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 73) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1274)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1274) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 122) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1323)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1323) / 192) * 4608)) + (rc_outer_outer * 576)) + ((((((int)threadIdx.x) / 3) + 57) & 63) * 9)) + ((((int)threadIdx.x) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1372)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1372) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 28) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1421)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1421) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 77) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
- kernel_shared[(((int)threadIdx.x) + 1470)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1470) / 192) * 4608)) + (rc_outer_outer * 576)) + (((int)threadIdx.x) * 3)) + rx_outer_outer) + 378)];
- if (((int)threadIdx.x) < 17) {
- kernel_shared[(((int)threadIdx.x) + 1519)] = kernel[((((((((int)blockIdx.x) * 36864) + (((((int)threadIdx.x) + 1519) / 192) * 4608)) + (rc_outer_outer * 576)) + (((((int)threadIdx.x) + 175) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ pad_temp_shared[((int)threadIdx.x)] = (((((7 <= (((int)threadIdx.x) % 63)) && ((((int)threadIdx.x) % 63) < 56)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[(((((rc_outer_outer * 784) + ((((int)threadIdx.x) / 63) * 49)) + rx_outer_outer) + (((int)threadIdx.x) % 63)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 112)] = (((((1 <= (((((int)threadIdx.x) / 7) + 7) % 9)) && ((((((int)threadIdx.x) / 7) + 7) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 112) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 7) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 224)] = (((((1 <= (((((int)threadIdx.x) / 7) + 5) % 9)) && ((((((int)threadIdx.x) / 7) + 5) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 224) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 5) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 336)] = (((((1 <= (((((int)threadIdx.x) / 7) + 3) % 9)) && ((((((int)threadIdx.x) / 7) + 3) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 336) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 3) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 448)] = (((((1 <= (((((int)threadIdx.x) / 7) + 1) % 9)) && ((((((int)threadIdx.x) / 7) + 1) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 448) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 1) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 560)] = (((((1 <= (((((int)threadIdx.x) / 7) + 8) % 9)) && ((((((int)threadIdx.x) / 7) + 8) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 560) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 8) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 672)] = (((((1 <= (((((int)threadIdx.x) / 7) + 6) % 9)) && ((((((int)threadIdx.x) / 7) + 6) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 672) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 6) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 784)] = (((((1 <= (((((int)threadIdx.x) / 7) + 4) % 9)) && ((((((int)threadIdx.x) / 7) + 4) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 784) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 4) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ pad_temp_shared[(((int)threadIdx.x) + 896)] = (((((1 <= (((((int)threadIdx.x) / 7) + 2) % 9)) && ((((((int)threadIdx.x) / 7) + 2) % 9) < 8)) && (1 <= (rx_outer_outer + (((int)threadIdx.x) % 7)))) && ((rx_outer_outer + (((int)threadIdx.x) % 7)) < 8)) ? data[((((((rc_outer_outer * 784) + (((((int)threadIdx.x) + 896) / 63) * 49)) + ((((((int)threadIdx.x) / 7) + 2) % 9) * 7)) + rx_outer_outer) + (((int)threadIdx.x) % 7)) - 8)] : 0.000000e+00f);
+ kernel_shared[((int)threadIdx.x)] = kernel[(((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 112)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 112) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 224)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 224) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 336)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer) + 32256)];
+ kernel_shared[(((int)threadIdx.x) + 448)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 448) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 560)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 560) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 672)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer) + 64512)];
+ kernel_shared[(((int)threadIdx.x) + 784)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 784) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 896)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 896) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1008)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer) + 96768)];
+ kernel_shared[(((int)threadIdx.x) + 1120)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1120) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1232)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1232) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 32) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 2) % 3) * 3)) + rx_outer_outer)];
+ kernel_shared[(((int)threadIdx.x) + 1344)] = kernel[((((((((int)blockIdx.x) * 147456) + ((((int)threadIdx.x) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((int)threadIdx.x) % 48) * 3)) + rx_outer_outer) + 129024)];
+ if (((int)threadIdx.x) < 80) {
+ kernel_shared[(((int)threadIdx.x) + 1456)] = kernel[((((((((int)blockIdx.x) * 147456) + (((((int)threadIdx.x) + 1456) / 48) * 4608)) + (rc_outer_outer * 144)) + ((((((int)threadIdx.x) + 16) % 48) / 3) * 9)) + (((((int)threadIdx.x) + 1) % 3) * 3)) + rx_outer_outer)];
}
__syncthreads();
- for (int rc_inner = 0; rc_inner < 64; ++rc_inner) {
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[(rc_inner * 3)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 192)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 384)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 576)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 768)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 960)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 1152)]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((rc_inner * 63) + ((int)threadIdx.x))] * kernel_shared[((rc_inner * 3) + 1344)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 1)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 193)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 385)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 577)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 769)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 961)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 1153)]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 7)] * kernel_shared[((rc_inner * 3) + 1345)]));
- conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 2)]));
- conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 194)]));
- conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 386)]));
- conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 578)]));
- conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 770)]));
- conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 962)]));
- conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 1154)]));
- conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((rc_inner * 63) + ((int)threadIdx.x)) + 14)] * kernel_shared[((rc_inner * 3) + 1346)]));
- }
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[((((int)threadIdx.x) % 7) * 7)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 1)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 2)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 3)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 4)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 5)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 6)] * kernel_shared[((((int)threadIdx.x) / 7) * 48)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[((((int)threadIdx.x) % 7) * 7)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 1)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 2)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 3)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 4)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 5)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 6)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 768)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 7)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 8)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 9)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 10)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 11)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 12)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 13)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 1)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 7)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 8)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 9)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 10)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 11)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 12)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 13)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 769)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 14)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 15)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 16)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 17)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 18)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 19)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 20)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 2)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 14)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 15)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 16)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 17)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 18)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 19)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 20)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 770)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 63)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 64)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 65)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 66)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 67)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 68)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 69)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 3)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 63)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 64)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 65)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 66)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 67)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 68)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 69)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 771)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 70)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 71)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 72)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 73)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 74)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 75)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 76)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 4)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 70)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 71)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 72)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 73)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 74)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 75)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 76)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 772)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 77)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 78)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 79)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 80)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 81)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 82)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 83)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 5)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 77)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 78)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 79)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 80)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 81)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 82)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 83)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 773)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 126)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 127)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 128)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 129)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 130)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 131)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 132)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 6)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 126)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 127)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 128)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 129)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 130)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 131)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 132)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 774)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 133)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 134)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 135)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 136)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 137)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 138)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 139)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 7)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 133)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 134)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 135)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 136)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 137)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 138)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 139)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 775)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 140)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 141)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 142)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 143)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 144)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 145)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 146)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 8)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 140)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 141)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 142)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 143)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 144)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 145)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 146)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 776)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 189)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 190)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 191)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 192)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 193)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 194)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 195)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 9)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 189)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 190)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 191)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 192)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 193)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 194)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 195)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 777)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 196)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 197)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 198)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 199)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 200)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 201)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 202)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 10)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 196)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 197)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 198)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 199)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 200)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 201)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 202)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 778)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 203)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 204)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 205)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 206)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 207)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 208)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 209)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 11)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 203)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 204)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 205)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 206)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 207)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 208)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 209)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 779)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 252)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 253)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 254)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 255)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 256)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 257)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 258)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 12)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 252)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 253)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 254)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 255)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 256)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 257)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 258)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 780)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 259)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 260)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 261)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 262)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 263)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 264)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 265)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 13)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 259)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 260)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 261)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 262)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 263)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 264)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 265)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 781)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 266)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 267)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 268)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 269)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 270)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 271)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 272)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 14)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 266)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 267)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 268)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 269)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 270)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 271)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 272)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 782)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 315)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 316)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 317)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 318)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 319)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 320)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 321)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 15)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 315)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 316)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 317)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 318)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 319)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 320)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 321)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 783)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 322)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 323)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 324)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 325)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 326)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 327)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 328)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 16)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 322)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 323)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 324)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 325)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 326)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 327)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 328)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 784)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 329)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 330)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 331)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 332)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 333)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 334)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 335)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 17)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 329)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 330)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 331)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 332)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 333)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 334)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 335)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 785)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 378)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 379)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 380)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 381)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 382)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 383)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 384)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 18)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 378)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 379)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 380)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 381)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 382)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 383)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 384)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 786)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 385)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 386)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 387)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 388)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 389)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 390)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 391)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 19)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 385)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 386)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 387)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 388)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 389)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 390)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 391)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 787)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 392)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 393)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 394)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 395)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 396)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 397)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 398)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 20)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 392)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 393)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 394)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 395)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 396)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 397)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 398)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 788)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 441)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 442)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 443)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 444)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 445)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 446)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 447)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 21)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 441)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 442)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 443)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 444)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 445)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 446)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 447)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 789)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 448)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 449)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 450)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 451)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 452)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 453)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 454)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 22)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 448)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 449)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 450)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 451)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 452)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 453)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 454)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 790)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 455)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 456)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 457)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 458)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 459)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 460)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 461)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 23)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 455)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 456)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 457)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 458)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 459)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 460)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 461)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 791)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 504)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 505)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 506)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 507)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 508)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 509)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 510)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 24)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 504)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 505)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 506)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 507)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 508)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 509)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 510)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 792)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 511)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 512)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 513)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 514)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 515)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 516)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 517)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 25)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 511)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 512)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 513)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 514)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 515)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 516)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 517)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 793)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 518)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 519)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 520)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 521)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 522)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 523)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 524)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 26)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 518)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 519)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 520)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 521)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 522)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 523)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 524)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 794)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 567)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 568)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 569)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 570)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 571)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 572)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 573)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 27)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 567)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 568)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 569)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 570)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 571)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 572)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 573)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 795)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 574)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 575)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 576)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 577)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 578)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 579)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 580)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 28)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 574)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 575)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 576)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 577)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 578)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 579)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 580)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 796)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 581)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 582)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 583)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 584)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 585)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 586)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 587)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 29)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 581)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 582)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 583)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 584)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 585)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 586)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 587)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 797)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 630)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 631)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 632)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 633)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 634)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 635)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 636)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 30)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 630)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 631)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 632)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 633)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 634)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 635)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 636)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 798)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 637)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 638)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 639)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 640)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 641)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 642)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 643)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 31)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 637)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 638)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 639)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 640)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 641)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 642)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 643)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 799)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 644)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 645)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 646)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 647)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 648)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 649)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 650)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 32)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 644)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 645)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 646)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 647)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 648)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 649)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 650)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 800)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 693)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 694)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 695)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 696)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 697)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 698)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 699)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 33)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 693)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 694)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 695)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 696)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 697)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 698)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 699)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 801)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 700)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 701)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 702)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 703)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 704)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 705)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 706)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 34)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 700)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 701)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 702)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 703)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 704)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 705)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 706)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 802)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 707)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 708)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 709)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 710)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 711)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 712)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 713)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 35)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 707)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 708)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 709)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 710)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 711)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 712)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 713)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 803)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 756)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 757)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 758)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 759)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 760)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 761)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 762)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 36)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 756)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 757)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 758)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 759)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 760)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 761)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 762)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 804)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 763)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 764)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 765)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 766)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 767)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 768)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 769)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 37)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 763)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 764)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 765)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 766)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 767)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 768)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 769)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 805)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 770)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 771)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 772)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 773)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 774)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 775)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 776)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 38)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 770)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 771)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 772)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 773)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 774)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 775)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 776)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 806)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 819)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 820)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 821)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 822)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 823)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 824)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 825)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 39)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 819)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 820)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 821)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 822)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 823)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 824)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 825)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 807)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 826)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 827)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 828)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 829)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 830)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 831)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 832)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 40)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 826)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 827)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 828)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 829)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 830)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 831)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 832)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 808)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 833)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 834)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 835)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 836)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 837)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 838)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 839)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 41)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 833)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 834)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 835)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 836)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 837)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 838)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 839)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 809)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 882)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 883)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 884)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 885)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 886)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 887)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 888)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 42)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 882)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 883)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 884)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 885)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 886)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 887)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 888)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 810)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 889)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 890)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 891)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 892)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 893)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 894)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 895)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 43)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 889)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 890)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 891)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 892)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 893)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 894)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 895)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 811)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 896)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 897)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 898)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 899)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 900)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 901)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 902)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 44)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 896)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 897)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 898)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 899)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 900)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 901)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 902)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 812)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 945)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 946)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 947)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 948)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 949)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 950)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 951)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 45)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 945)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 946)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 947)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 948)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 949)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 950)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 951)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 813)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 952)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 953)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 954)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 955)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 956)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 957)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 958)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 46)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 952)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 953)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 954)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 955)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 956)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 957)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 958)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 814)]));
+ conv2d_nchw[0] = (conv2d_nchw[0] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 959)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[1] = (conv2d_nchw[1] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 960)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[2] = (conv2d_nchw[2] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 961)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[3] = (conv2d_nchw[3] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 962)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[4] = (conv2d_nchw[4] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 963)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[5] = (conv2d_nchw[5] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 964)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[6] = (conv2d_nchw[6] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 965)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 47)]));
+ conv2d_nchw[7] = (conv2d_nchw[7] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 959)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[8] = (conv2d_nchw[8] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 960)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[9] = (conv2d_nchw[9] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 961)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[10] = (conv2d_nchw[10] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 962)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[11] = (conv2d_nchw[11] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 963)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[12] = (conv2d_nchw[12] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 964)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
+ conv2d_nchw[13] = (conv2d_nchw[13] + (pad_temp_shared[(((((int)threadIdx.x) % 7) * 7) + 965)] * kernel_shared[(((((int)threadIdx.x) / 7) * 48) + 815)]));
}
}
- for (int i1_inner = 0; i1_inner < 8; ++i1_inner) {
- compute[(((((int)blockIdx.x) * 392) + (i1_inner * 49)) + ((int)threadIdx.x))] = max((conv2d_nchw[i1_inner] + bias[((((int)blockIdx.x) * 8) + i1_inner)]), 0.000000e+00f);
- }
+ compute[((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7))] = max((conv2d_nchw[0] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 1)] = max((conv2d_nchw[1] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 2)] = max((conv2d_nchw[2] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 3)] = max((conv2d_nchw[3] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 4)] = max((conv2d_nchw[4] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 5)] = max((conv2d_nchw[5] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 6)] = max((conv2d_nchw[6] + bias[((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7))]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 784)] = max((conv2d_nchw[7] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 785)] = max((conv2d_nchw[8] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 786)] = max((conv2d_nchw[9] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 787)] = max((conv2d_nchw[10] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 788)] = max((conv2d_nchw[11] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 789)] = max((conv2d_nchw[12] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
+ compute[(((((int)blockIdx.x) * 1568) + (((int)threadIdx.x) * 7)) + 790)] = max((conv2d_nchw[13] + bias[(((((int)blockIdx.x) * 32) + (((int)threadIdx.x) / 7)) + 16)]), 0.000000e+00f);
}
</pre></div>
</div>
@@ -1113,7 +2173,7 @@ 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> ( 5 minutes 29.809 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 5 minutes 26.025 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 89b7f379b1..7c84ec51a8 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_cuda.html
@@ -916,7 +916,7 @@ 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)
- 7.9005 7.9059 7.9113 7.8842 0.0117
+ 7.9106 7.9078 7.9204 7.9036 0.0071
</pre></div>
</div>
</div>
@@ -938,7 +938,7 @@ 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 0.857 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 0.903 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 53f3bf865c..1bea72ea5f 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_network_x86.html
@@ -935,7 +935,7 @@ 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)
- 746.5433 746.6337 747.1823 745.8140 0.5622
+ 743.2264 743.2485 743.7502 742.6803 0.4371
</pre></div>
</div>
</div>
@@ -957,7 +957,7 @@ 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 30.541 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 30.113 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 25d1c1c9e7..66ae72e004 100644
--- a/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
+++ b/docs/how_to/tune_with_autoscheduler/tune_sparse_x86.html
@@ -633,28 +633,29 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
placeholder_4: Buffer(placeholder_14: Pointer(float32), float32, [128, 512], []),
compute: Buffer(compute_2: Pointer(float32), float32, [128, 512], [])}
buffer_map = {placeholder_5: placeholder, placeholder_6: placeholder_1, placeholder_7: placeholder_2, placeholder_8: placeholder_3, placeholder_9: placeholder_4, compute_1: compute} {
- for (i0.outer.i1.outer.fused: int32, 0, 32) "parallel" {
- allocate(compute_3: Pointer(global float32), float32, [2048]), storage_scope = global {
+ for (i0.outer.i1.outer.fused: int32, 0, 16) "parallel" {
+ allocate(compute_3: Pointer(global float32), float32, [4096]), storage_scope = global {
for (i.outer.inner: int32, 0, 8) {
- for (i.inner.init: int32, 0, 16) {
- for (j.init: int32, 0, 16) {
- compute_4: Buffer(compute_3, float32, [2048], [])[(((i.outer.inner*256) + (i.inner.init*16)) + j.init)] = 0f32
+ for (nb_j.inner: int32, 0, 2) {
+ for (i.inner.init: int32, 0, 16) {
+ for (j.init: int32, 0, 16) {
+ compute_4: Buffer(compute_3, float32, [4096], [])[((((i.outer.inner*512) + (i.inner.init*32)) + (nb_j.inner*16)) + j.init)] = 0f32
+ }
}
- }
- for (elem_idx: int32, 0, (placeholder_15: Buffer(placeholder_13, int32, [33], [])[(i0.outer.i1.outer.fused + 1)] - placeholder_15[i0.outer.i1.outer.fused])) {
- for (i.inner: int32, 0, 16) {
- for (j: int32, 0, 16) {
- if @tir.likely((elem_idx < (placeholder_15[(i0.outer.i1.outer.fused + 1)] - placeholder_15[i0.outer.i1.outer.fused])), dtype=bool) {
- let cse_var_1: int32 = (((i.outer.inner*256) + (i.inner*16)) + j)
- compute_4[cse_var_1] = (compute_4[cse_var_1] + (placeholder_16: Buffer(placeholder_11, float32, [78656], [])[(((placeholder_15[i0.outer.i1.outer.fused]*16) + (elem_idx*16)) + j)]*max(placeholder_17: Buffer(placeholder_10, float32, [32768], [])[(((i.outer.inner*4096) + (i.inner*256)) + placeholder_18: Buffer(placeholder_12, int32, [4916], [])[(placeholder_15[i0.outer.i1.outer.fused] + elem_idx)])], 0f32)))
+ for (elem_idx: int32, 0, let cse_var_1: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner) in (placeholder_15: Buffer(placeholder_13, int32, [33], [])[(cse_var_1 + 1)] - placeholder_15[cse_var_1])) {
+ for (i.inner: int32, 0, 16) {
+ for (j: int32, 0, 16) {
+ let cse_var_3: int32 = ((i0.outer.i1.outer.fused*2) + nb_j.inner)
+ let cse_var_2: int32 = ((((i.outer.inner*512) + (i.inner*32)) + (nb_j.inner*16)) + j)
+ compute_4[cse_var_2] = (compute_4[cse_var_2] + (placeholder_16: Buffer(placeholder_11, float32, [78656], [])[(((placeholder_15[cse_var_3]*16) + (elem_idx*16)) + j)]*max(placeholder_17: Buffer(placeholder_10, float32, [32768], [])[(((i.outer.inner*4096) + (i.inner*256)) + placeholder_18: Buffer(placeholder_12, int32, [4916], [])[(placeholder_15[cse_var_3] + elem_idx)])], 0f32)))
}
}
}
}
}
for (i0.inner: int32, 0, 128) {
- let cse_var_2: int32 = ((i0.inner*512) + (i0.outer.i1.outer.fused*16))
- compute_5: Buffer(compute_2, float32, [65536], [])[ramp(cse_var_2, 1, 16)] = max((compute_4[ramp((i0.inner*16), 1, 16)] + placeholder_19: Buffer(placeholder_14, float32, [65536], [])[ramp(cse_var_2, 1, 16)]), broadcast(0f32, 16))
+ let cse_var_4: int32 = ((i0.inner*512) + (i0.outer.i1.outer.fused*32))
+ compute_5: Buffer(compute_2, float32, [65536], [])[ramp(cse_var_4, 1, 32)] = max((compute_4[ramp((i0.inner*32), 1, 32)] + placeholder_19: Buffer(placeholder_14, float32, [65536], [])[ramp(cse_var_4, 1, 32)]), broadcast(0f32, 32))
}
}
}
@@ -692,7 +693,7 @@ layout transformation, parallelization, vectorization, unrolling, and operator f
<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: 1.526 ms
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution time of this operator: 1.508 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 07f8f359df..a6d1a5c588 100644
--- a/docs/how_to/tune_with_autotvm/sg_execution_times.html
+++ b/docs/how_to/tune_with_autotvm/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>00:48.546</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
+<p><strong>00:37.776</strong> total execution time for <strong>how_to_tune_with_autotvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -349,11 +349,11 @@
</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>00:48.511</p></td>
+<td><p>00:37.741</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>00:00.020</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>
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 da9abfc925..77f5f2aebb 100644
--- a/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
+++ b/docs/how_to/tune_with_autotvm/tune_conv2d_cuda.html
@@ -690,8 +690,9 @@ 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, 8, 16, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10062228
-No: 2 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 4, 64, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3174427
+No: 2 GFLOPS: 32.47/32.47 result: MeasureResult(costs=(0.007130298227272727,), error_no=MeasureErrorNo.NO_ERROR, all_cost=2.338937997817993, timestamp=1673430625.906357) [('tile_f', [-1, 16, 16, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1754369
+No: 3 GFLOPS: 0.00/32.47 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
@@ -813,8 +814,8 @@ 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, 4, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2510569
-No: 3 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 16, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6948515
+No: 4 GFLOPS: 0.00/32.47 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
@@ -936,27 +937,747 @@ 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, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5644349
-No: 4 GFLOPS: 0.00/0.00 result: Traceback (most recent call last):
- File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 142, in build
- res = future.result()
- File "/usr/lib/python3.7/concurrent/futures/_base.py", line 435, in result
- return self.__get_result()
- File "/usr/lib/python3.7/concurrent/futures/_base.py", line 384, in __get_result
- raise self._exception
- File "/usr/lib/python3.7/concurrent/futures/thread.py", line 57, in run
- result = self.fn(*self.args, **self.kwargs)
- File "/workspace/python/tvm/contrib/popen_pool.py", line 432, in <lambda>
- worker = lambda *args: self._worker_run(*args)
- File "/workspace/python/tvm/contrib/popen_pool.py", line 401, in _worker_run
- return proc.recv()
- File "/workspace/python/tvm/contrib/popen_pool.py", line 309, in recv
- raise TimeoutError()
-TimeoutError
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 64, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9888828
+No: 5 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 4, 32, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1190022
+No: 6 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 8, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1098144
+No: 7 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 32, 1, 4]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 64]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10035625
+No: 8 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 1, 256, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9941412
+No: 9 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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
+
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 1, 8, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 256, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9710167
+No: 10 GFLOPS: 0.00/32.47 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_host=task.target_host, 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
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 32, 1, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7173461
-No: 5 GFLOPS: 84.05/84.05 result: MeasureResult(costs=(0.002754245216216216,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5222408771514893, timestamp=1673414749.1550088) [('tile_f', [-1, 1, 4, 16]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2392895
-No: 6 GFLOPS: 0.00/84.05 result: Traceback (most recent call last):
+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:1730
+ 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:1670
+ 19: run<>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 18: run<tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 17: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 16: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 15: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1630
+ 14: run<tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_, tvm::runtime::TVMMovableArgValueWithContext_>
+ at ../include/tvm/runtime/packed_func.h:1645
+ 13: operator()
+ at ../src/driver/driver_api.cc:395
+ 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:381
+ 11: tvm::LowerWithPassList(tvm::IRModule, tvm::runtime::Array<tvm::transform::Pass, void>)
+ at ../src/driver/driver_api.cc:276
+ 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:454
+ 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:1749
+ 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:1693
+ 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:1617
+ 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, 32, 1, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 128, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1031145
+No: 11 GFLOPS: 1.40/32.47 result: MeasureResult(costs=(0.1651578825,), error_no=MeasureErrorNo.NO_ERROR, all_cost=4.256547689437866, timestamp=1673430632.4948845) [('tile_f', [-1, 1, 1, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,8887835
+No: 12 GFLOPS: 0.00/32.47 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
@@ -1078,8 +1799,8 @@ 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, 4, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 1, 32]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5949874
-No: 7 GFLOPS: 0.00/84.05 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 8, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 16, 4]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10343820
+No: 13 GFLOPS: 0.00/32.47 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
@@ -1201,8 +1922,8 @@ 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, 4, 32]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 64]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8103235
-No: 8 GFLOPS: 0.00/84.05 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 32]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9632850
+No: 14 GFLOPS: 0.00/32.47 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
@@ -1324,8 +2045,8 @@ 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, 32, 2, 2]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 2, 256]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,962789
-No: 9 GFLOPS: 0.00/84.05 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 32, 1, 2]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 128, 4]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,288480
+No: 15 GFLOPS: 0.00/32.47 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
@@ -1447,10 +2168,8 @@ 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, 32, 1, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 32, 4]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2991400
-No: 10 GFLOPS: 87.66/87.66 result: MeasureResult(costs=(0.002640982239130435,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.2351536750793457, timestamp=1673414751.758597) [('tile_f', [-1, 8, 32, 1]), ('tile_y', [-1, 1, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1744203
-No: 11 GFLOPS: 134.20/134.20 result: MeasureResult(costs=(0.001725022275862069,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.235008955001831, timestamp=1673414752.4624658) [('tile_f', [-1, 2, 32, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,489981
-No: 12 GFLOPS: 0.00/134.20 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 2, 4, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4498535
+No: 16 GFLOPS: 0.00/32.47 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
@@ -1572,9 +2291,8 @@ 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, 16, 2, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 1, 128]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,3079354
-No: 13 GFLOPS: 94.54/134.20 result: MeasureResult(costs=(0.002448800608695652,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.6984493732452393, timestamp=1673414756.3338737) [('tile_f', [-1, 2, 8, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 16]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7478615
-No: 14 GFLOPS: 0.00/134.20 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 128, 1, 4]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,1212307
+No: 17 GFLOPS: 0.00/32.47 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
@@ -1696,8 +2414,8 @@ 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, 1]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 1, 64]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,8289826
-No: 15 GFLOPS: 0.00/134.20 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 2, 32]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 8, 16]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 1)],None,7296933
+No: 18 GFLOPS: 0.00/32.47 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
@@ -1819,12 +2537,26 @@ 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, 2]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 128, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,2735327
-No: 16 GFLOPS: 8.07/134.20 result: MeasureResult(costs=(0.02869964775,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.5496540069580078, timestamp=1673414757.147663) [('tile_f', [-1, 2, 4, 8]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 3, 1]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,6008570
-No: 17 GFLOPS: 4.92/134.20 result: MeasureResult(costs=(0.047058536500000005,), error_no=MeasureErrorNo.NO_ERROR, all_cost=7.601166486740112, timestamp=1673414764.9139814) [('tile_f', [-1, 2, 4, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,3537090
-No: 18 GFLOPS: 46.92/134.20 result: MeasureResult(costs=(0.004934022047619048,), error_no=MeasureErrorNo.NO_ERROR, all_cost=1.2691385746002197, timestamp=1673414765.66909) [('tile_f', [-1, 2, 32, 8]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 1)],None,5229342
-No: 19 GFLOPS: 1.63/134.20 result: MeasureResult(costs=(0.142306076,), error_no=MeasureErrorNo.NO_ERROR, all_cost=3.1770129203796387, timestamp=1673414767.9414265) [('tile_f', [-1, 8, 2, 32]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 2, 1]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,391353
-No: 20 GFLOPS: 0.00/134.20 result: Traceback (most recent call last):
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 8, 4, 4]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 16, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,10311738
+No: 19 GFLOPS: 0.00/32.47 result: Traceback (most recent call last):
+ File "/workspace/python/tvm/autotvm/measure/measure_methods.py", line 142, in build
+ res = future.result()
+ File "/usr/lib/python3.7/concurrent/futures/_base.py", line 435, in result
+ return self.__get_result()
+ File "/usr/lib/python3.7/concurrent/futures/_base.py", line 384, in __get_result
+ raise self._exception
+ File "/usr/lib/python3.7/concurrent/futures/thread.py", line 57, in run
+ result = self.fn(*self.args, **self.kwargs)
+ File "/workspace/python/tvm/contrib/popen_pool.py", line 432, in <lambda>
+ worker = lambda *args: self._worker_run(*args)
+ File "/workspace/python/tvm/contrib/popen_pool.py", line 401, in _worker_run
+ return proc.recv()
+ File "/workspace/python/tvm/contrib/popen_pool.py", line 309, in recv
+ raise TimeoutError()
+TimeoutError
+
+ [('tile_f', [-1, 16, 2, 8]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 4, 2]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 3]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 1)],None,9919067
+No: 20 GFLOPS: 0.00/32.47 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
@@ -1946,7 +2678,7 @@ 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, 8, 64, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 7]), ('tile_rc', [-1, 2, 2]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 3, 1]), ('auto_unroll_max_step', 1500), ('unroll_explicit', 0)],None,4494868
+tvm.autotvm.task.space.InstantiationError: Skipped because of invalid gpu kernel [('tile_f', [-1, 1, 32, 16]), ('tile_y', [-1, 7, 1, 1]), ('tile_x', [-1, 1, 7, 1]), ('tile_rc', [-1, 1, 512]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,192244
</pre></div>
</div>
<p>Finally we can inspect the best config from log file, check correctness,
@@ -1985,9 +2717,9 @@ 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, 2, 32, 1]), ('tile_y', [-1, 1, 1, 7]), ('tile_x', [-1, 1, 1, 1]), ('tile_rc', [-1, 4, 8]), ('tile_ry', [-1, 1, 3]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 0), ('unroll_explicit', 0)],None,489981
+[('tile_f', [-1, 16, 16, 2]), ('tile_y', [-1, 1, 7, 1]), ('tile_x', [-1, 7, 1, 1]), ('tile_rc', [-1, 8, 1]), ('tile_ry', [-1, 1, 1]), ('tile_rx', [-1, 1, 1]), ('auto_unroll_max_step', 512), ('unroll_explicit', 0)],None,1754369
Finish loading 20 records
-Time cost of this operator: 0.002178
+Time cost of this operator: 0.007549
</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 60e31cd701..5ba336eebc 100644
--- a/docs/how_to/work_with_microtvm/micro_autotune.html
+++ b/docs/how_to/work_with_microtvm/micro_autotune.html
@@ -663,10 +663,10 @@ 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 309.4 98.717 (1, 2, 10, 10, 3) 2 1 [309.4]
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.035 0.968 (1, 6, 10, 10) 1 1 [3.035]
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.987 0.315 (1, 1, 10, 10, 3) 1 1 [0.987]
-Total_time - 313.421 - - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 311.2 98.735 (1, 2, 10, 10, 3) 2 1 [311.2]
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 3.018 0.958 (1, 6, 10, 10) 1 1 [3.018]
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.969 0.307 (1, 1, 10, 10, 3) 1 1 [0.969]
+Total_time - 315.187 - - - - -
</pre></div>
</div>
</div>
@@ -718,10 +718,10 @@ 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 103.4 97.494 (1, 6, 10, 10, 1) 2 1 [103.4]
-tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.811 1.708 (1, 6, 10, 10) 1 1 [1.811]
-tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.847 0.798 (1, 3, 10, 10, 1) 1 1 [0.847]
-Total_time - 106.058 - - - - -
+tvmgen_default_fused_nn_contrib_conv2d_NCHWc tvmgen_default_fused_nn_contrib_conv2d_NCHWc 103.6 97.529 (1, 6, 10, 10, 1) 2 1 [103.6]
+tvmgen_default_fused_layout_transform_1 tvmgen_default_fused_layout_transform_1 1.774 1.67 (1, 6, 10, 10) 1 1 [1.774]
+tvmgen_default_fused_layout_transform tvmgen_default_fused_layout_transform 0.851 0.801 (1, 3, 10, 10, 1) 1 1 [0.851]
+Total_time - 106.225 - - - - -
</pre></div>
</div>
<div class="sphx-glr-footer sphx-glr-footer-example docutils container" id="sphx-glr-download-how-to-work-with-microtvm-micro-autotune-py">
diff --git a/docs/how_to/work_with_microtvm/micro_pytorch.html b/docs/how_to/work_with_microtvm/micro_pytorch.html
index dde90f9075..b250eec3c0 100644
--- a/docs/how_to/work_with_microtvm/micro_pytorch.html
+++ b/docs/how_to/work_with_microtvm/micro_pytorch.html
@@ -453,8 +453,7 @@ 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, 18.4MB/s]
-100%|##########| 3.42M/3.42M [00:00<00:00, 29.0MB/s]
+100%|##########| 3.42M/3.42M [00:00<00:00, 38.8MB/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.
@@ -578,7 +577,7 @@ 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 1.827 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 1 minutes 1.602 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 5d56891cd9..3ce1ee702e 100644
--- a/docs/how_to/work_with_microtvm/micro_train.html
+++ b/docs/how_to/work_with_microtvm/micro_train.html
@@ -523,7 +523,7 @@ 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/tmpy1qmbn_8/images/random'
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>'/tmp/tmpsi_54jvx/images/random'
</pre></div>
</div>
</div>
@@ -583,8 +583,8 @@ 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], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [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/tmpy1qmbn_8/images/target contains 8144 images
-/tmp/tmpy1qmbn_8/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], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0], [0.0, 1.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [1.0, 0.0], [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/tmpsi_54jvx/images/target contains 8144 images
+/tmp/tmpsi_54jvx/images/random contains 5000 images
</pre></div>
</div>
</div>
@@ -696,13 +696,13 @@ 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 - 46s - loss: 0.2135 - accuracy: 0.9272 - val_loss: 0.1840 - val_accuracy: 0.9475 - 46s/epoch - 141ms/step
+328/328 - 46s - loss: 0.2090 - accuracy: 0.9301 - val_loss: 0.1511 - val_accuracy: 0.9524 - 46s/epoch - 141ms/step
Epoch 2/3
-328/328 - 43s - loss: 0.0903 - accuracy: 0.9667 - val_loss: 0.1057 - val_accuracy: 0.9611 - 43s/epoch - 130ms/step
+328/328 - 43s - loss: 0.1003 - accuracy: 0.9636 - val_loss: 0.1229 - val_accuracy: 0.9532 - 43s/epoch - 130ms/step
Epoch 3/3
-328/328 - 43s - loss: 0.0702 - accuracy: 0.9750 - val_loss: 0.1234 - val_accuracy: 0.9581 - 43s/epoch - 130ms/step
+328/328 - 43s - loss: 0.0682 - accuracy: 0.9737 - val_loss: 0.1307 - val_accuracy: 0.9615 - 43s/epoch - 130ms/step
-<keras.callbacks.History object at 0x7f21ade11f10>
+<keras.callbacks.History object at 0x7fa812b525d0>
</pre></div>
</div>
</div>
@@ -962,7 +962,7 @@ 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 23.439 seconds)</p>
+<p class="sphx-glr-timing"><strong>Total running time of the script:</strong> ( 4 minutes 45.819 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 8d5064e23b..5a76f9883c 100644
--- a/docs/how_to/work_with_microtvm/sg_execution_times.html
+++ b/docs/how_to/work_with_microtvm/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>06:27.123</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
+<p><strong>06:49.719</strong> total execution time for <strong>how_to_work_with_microtvm</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -349,23 +349,23 @@
</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">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>04:23.439</p></td>
+<td><p>04:45.819</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><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">microTVM PyTorch Tutorial</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_pytorch.py</span></code>)</p></td>
-<td><p>01:01.827</p></td>
+<td><p>01:01.602</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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">Autotuning with microTVM</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_autotune.py</span></code>)</p></td>
-<td><p>00:50.305</p></td>
+<td><p>00:50.647</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">microTVM Host-Driven AoT</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_aot.py</span></code>)</p></td>
-<td><p>00:07.820</p></td>
+<td><p>00:07.884</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-odd"><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">microTVM with TFLite Models</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_tflite.py</span></code>)</p></td>
-<td><p>00:03.731</p></td>
+<td><p>00:03.765</p></td>
<td><p>0.0 MB</p></td>
</tr>
<tr class="row-even"><td><p><a class="reference internal" href="micro_reference_vm.html#sphx-glr-how-to-work-with-microtvm-micro-reference-vm-py"><span class="std std-ref">microTVM Reference Virtual Machines</span></a> (<code class="docutils literal notranslate"><span class="pre">micro_reference_vm.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 d45b7c9d93..553145634b 100644
--- a/docs/how_to/work_with_relay/sg_execution_times.html
+++ b/docs/how_to/work_with_relay/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>00:43.473</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
+<p><strong>00:43.599</strong> total execution time for <strong>how_to_work_with_relay</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 84%" />
@@ -349,15 +349,15 @@
</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>00:31.860</p></td>
+<td><p>00:32.055</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>00:10.126</p></td>
+<td><p>00:10.096</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>00:01.480</p></td>
+<td><p>00:01.441</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>
diff --git a/docs/how_to/work_with_schedules/intrin_math.html b/docs/how_to/work_with_schedules/intrin_math.html
index da45166462..e4b434ca27 100644
--- a/docs/how_to/work_with_schedules/intrin_math.html
+++ b/docs/how_to/work_with_schedules/intrin_math.html
@@ -536,7 +536,7 @@ 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 0x7f21ae94db90>
+<div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span><function my_cuda_math_rule at 0x7fa7b5ce87a0>
</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 997b10275e..cc2faf4b21 100644
--- a/docs/how_to/work_with_schedules/sg_execution_times.html
+++ b/docs/how_to/work_with_schedules/sg_execution_times.html
@@ -340,7 +340,7 @@
<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>00:06.741</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
+<p><strong>00:04.921</strong> total execution time for <strong>how_to_work_with_schedules</strong> files:</p>
<table class="docutils align-default">
<colgroup>
<col style="width: 83%" />
@@ -349,19 +349,19 @@
</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>00:04.280</p></td>
+<td><p>00:02.410</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>00:01.120</p></td>
+<td><p>00:01.140</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>00:00.575</p></td>
+<td><p>00:00.588</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>00:00.554</p></td>
+<td><p>00:00.569</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>
diff --git a/docs/how_to/work_with_schedules/tensorize.html b/docs/how_to/work_with_schedules/tensorize.html
index 94673d0bcc..92f0539e4a 100644
--- a/docs/how_to/work_with_schedules/tensorize.html
+++ b/docs/how_to/work_with_schedules/tensorize.html
@@ -587,7 +587,7 @@ The importing needs to happen before the tensorized GEMV being executed.</p>
B: Buffer(B_2: Pointer(float32), float32, [512, 64], []),
C: Buffer(C_2: Pointer(float32), float32, [1024, 512], [])}
buffer_map = {A_1: A, B_1: B, C_1: C} {
- attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp6vzbr1y2/input0.cc'\nsource_filename = \"/tmp/tmp6vzbr1y2/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n %7 = allo [...]
+ attr [IterVar(i: int32, (nullptr), "DataPar", "")] "pragma_import_llvm" = "; ModuleID = '/tmp/tmp6evazljz/input0.cc'\nsource_filename = \"/tmp/tmp6evazljz/input0.cc\"\ntarget datalayout = \"e-m:e-i64:64-f80:128-n8:16:32:64-S128\"\ntarget triple = \"x86_64-pc-linux-gnu\"\n\n; Function Attrs: noinline nounwind optnone uwtable\ndefine dso_local i32 @gemv_update(float*, float*, float*, i32, i32, i32) #0 {\n %7 = allo [...]
for (i, 0, 1024) {
for (j.outer: int32, 0, 32) {
@tir.call_extern("gemv_update", @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), C_2, ((i*512) + (j.outer*16)), 16, 2, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), A_2, (i*64), 64, 1, dtype=handle), @tir.tvm_access_ptr(@tir.type_annotation(, dtype=float32), B_2, (j.outer*1024), 1024, 1, dtype=handle), 16, 64, 64, dtype=int32)
diff --git a/docs/objects.inv b/docs/objects.inv
index ccb2ee1ff0..f810a42fb8 100644
Binary files a/docs/objects.inv and b/docs/objects.inv differ
diff --git a/docs/reference/api/doxygen/namespacemembers_func_l.html b/docs/reference/api/doxygen/namespacemembers_func_l.html
index 8cdadc0b77..3e0a7dc294 100644
--- a/docs/reference/api/doxygen/namespacemembers_func_l.html
+++ b/docs/reference/api/doxygen/namespacemembers_func_l.html
@@ -90,7 +90,8 @@ $(function() {
, <a class="el" href="namespacetvm_1_1topi.html#a46057526edd4bbd0a291edf7f0c863b4">tvm::topi</a>
</li>
<li>Legalize()
-: <a class="el" href="namespacetvm_1_1relay_1_1qnn_1_1transform.html#a2323e3c38cc9ae1626cd98295b83e906">tvm::relay::qnn::transform</a>
+: <a class="el" href="namespacetvm_1_1relay_1_1legalize.html#aa09b6a5f5aef5c92628b13be0f8cf0a8">tvm::relay::legalize</a>
+, <a class="el" href="namespacetvm_1_1relay_1_1qnn_1_1transform.html#a2323e3c38cc9ae1626cd98295b83e906">tvm::relay::qnn::transform</a>
, <a class="el" href="namespacetvm_1_1relay_1_1transform.html#aae623a28eda64b60c6ee90edde103891">tvm::relay::transform</a>
</li>
<li>LegalizePackedCalls()
@@ -101,7 +102,7 @@ $(function() {
, <a class="el" href="namespacetvm_1_1topi.html#a28aa974fb51f2e262413811cab7f969e">tvm::topi</a>
</li>
<li>less_equal()
-: <a class="el" href="namespacetvm.html#ad4734f467b4107f0da21a510788479c1">tvm</a>
+: <a class="el" href="namespacetvm.html#a5cee73ced0a40ed261dc3beec9f8247c">tvm</a>
, <a class="el" href="namespacetvm_1_1topi.html#aab3e2cccec9b8e90aa01e64c2587d8d9">tvm::topi</a>
</li>
<li>Let()
@@ -136,8 +137,8 @@ $(function() {
: <a class="el" href="namespacetvm_1_1topi_1_1nn.html#ac0e20b6b30ec8296c1f037866d3bf772">tvm::topi::nn</a>
</li>
<li>logical_and()
-: <a class="el" href="namespacetvm.html#a27d5567b95675d383c4675fdcd85346c">tvm</a>
-, <a class="el" href="namespacetvm_1_1topi.html#a81fb1c1bf730c4824470aac242c083ff">tvm::topi</a>
+: <a class="el" href="namespacetvm.html#a4bf2f9a44cbd664ba4a0be9a0d35f0b5">tvm</a>
+, <a class="el" href="namespacetvm_1_1topi.html#aef92757524e9e87b1eb646098fceac7a">tvm::topi</a>
</li>
<li>logical_not()
: <a class="el" href="namespacetvm.html#a62955df1df48917116efe39d4cd18fec">tvm</a>
@@ -148,7 +149,7 @@ $(function() {
, <a class="el" href="namespacetvm_1_1topi.html#a40dc56569ff5aad86ca82762ffb11180">tvm::topi</a>
</li>
<li>logical_xor()
-: <a class="el" href="namespacetvm_1_1topi.html#af2f8fcdbbf2ebc9e4acc822325f5c768">tvm::topi</a>
+: <a class="el" href="namespacetvm_1_1topi.html#a1fa63c5475fc107ee0be36929f99463d">tvm::topi</a>
</li>
<li>lookup_param()
: <a class="el" href="namespacetvm_1_1tir_1_1builtin.html#af5125377c482e5b34bf8710b7b3dc1e6">tvm::tir::builtin</a>
@@ -187,7 +188,7 @@ $(function() {
: <a class="el" href="namespacetvm.html#ac22649049328c5f3704e3592e6ffd6c5">tvm</a>
</li>
<li>LowerSchedule()
-: <a class="el" href="namespacetvm.html#a57e287f55feee72b4f61b9944878f757">tvm</a>
+: <a class="el" href="namespacetvm.html#aee93eceb651e8509a1bf340a982148de">tvm</a>
</li>
<li>LowerThreadAllreduce()
: <a class="el" href="namespacetvm_1_1tir_1_1transform.html#a16d42050efec51126d5b90eb2f60171f">tvm::tir::transform</a>
diff --git a/docs/reference/api/doxygen/namespacemembers_l.html b/docs/reference/api/doxygen/namespacemembers_l.html
index 2f339108a7..9ae19f3257 100644
--- a/docs/reference/api/doxygen/namespacemembers_l.html
+++ b/docs/reference/api/doxygen/namespacemembers_l.html
@@ -99,7 +99,8 @@ $(function() {
, <a class="el" href="namespacetvm_1_1topi.html#a3e83ff0bafdc8bc4d92683c67fd3bbef">tvm::topi</a>
</li>
<li>Legalize()
-: <a class="el" href="namespacetvm_1_1relay_1_1qnn_1_1transform.html#a2323e3c38cc9ae1626cd98295b83e906">tvm::relay::qnn::transform</a>
+: <a class="el" href="namespacetvm_1_1relay_1_1legalize.html#aa09b6a5f5aef5c92628b13be0f8cf0a8">tvm::relay::legalize</a>
+, <a class="el" href="namespacetvm_1_1relay_1_1qnn_1_1transform.html#a2323e3c38cc9ae1626cd98295b83e906">tvm::relay::qnn::transform</a>
, <a class="el" href="namespacetvm_1_1relay_1_1transform.html#aae623a28eda64b60c6ee90edde103891">tvm::relay::transform</a>
</li>
<li>LegalizePackedCalls()
@@ -107,7 +108,7 @@ $(function() {
</li>
<li>less()
: <a class="el" href="namespacetvm.html#a52fa1dc57423a077eb098960162e7b85">tvm</a>
-, <a class="el" href="namespacetvm_1_1topi.html#a28aa974fb51f2e262413811cab7f969e">tvm::topi</a>
+, <a class="el" href="namespacetvm_1_1topi.html#aac9102587dd015eb4929628bfe9dc0eb">tvm::topi</a>
</li>
<li>less_equal()
: <a class="el" href="namespacetvm.html#a6dfe80d16a7b4f551c87a8901d366d08">tvm</a>
@@ -146,7 +147,7 @@ $(function() {
</li>
<li>logical_and()
: <a class="el" href="namespacetvm.html#a27d5567b95675d383c4675fdcd85346c">tvm</a>
-, <a class="el" href="namespacetvm_1_1topi.html#ac1de3a16468101e6f61e9f099956754f">tvm::topi</a>
+, <a class="el" href="namespacetvm_1_1topi.html#a81fb1c1bf730c4824470aac242c083ff">tvm::topi</a>
</li>
<li>logical_not()
: <a class="el" href="namespacetvm.html#a62955df1df48917116efe39d4cd18fec">tvm</a>
@@ -154,10 +155,10 @@ $(function() {
</li>
<li>logical_or()
: <a class="el" href="namespacetvm.html#a4509dece1af96338cc25097855fcecd7">tvm</a>
-, <a class="el" href="namespacetvm_1_1topi.html#a6ee600bfc4bf51acfb382cf8bea41540">tvm::topi</a>
+, <a class="el" href="namespacetvm_1_1topi.html#a40dc56569ff5aad86ca82762ffb11180">tvm::topi</a>
</li>
<li>logical_xor()
-: <a class="el" href="namespacetvm_1_1topi.html#af431e260780c9070d4b4fcd264e0bd23">tvm::topi</a>
+: <a class="el" href="namespacetvm_1_1topi.html#af2f8fcdbbf2ebc9e4acc822325f5c768">tvm::topi</a>
</li>
<li>lookup_param()
: <a class="el" href="namespacetvm_1_1tir_1_1builtin.html#af5125377c482e5b34bf8710b7b3dc1e6">tvm::tir::builtin</a>
@@ -199,7 +200,7 @@ $(function() {
: <a class="el" href="namespacetvm.html#ac22649049328c5f3704e3592e6ffd6c5">tvm</a>
</li>
<li>LowerSchedule()
-: <a class="el" href="namespacetvm.html#a57e287f55feee72b4f61b9944878f757">tvm</a>
+: <a class="el" href="namespacetvm.html#aee93eceb651e8509a1bf340a982148de">tvm</a>
</li>
<li>LowerThreadAllreduce()
: <a class="el" href="namespacetvm_1_1tir_1_1transform.html#a16d42050efec51126d5b90eb2f60171f">tvm::tir::transform</a>
diff --git a/docs/reference/api/doxygen/namespacemembers_m.html b/docs/reference/api/doxygen/namespacemembers_m.html
index bd65a0e598..6856026e49 100644
--- a/docs/reference/api/doxygen/namespacemembers_m.html
+++ b/docs/reference/api/doxygen/namespacemembers_m.html
@@ -173,6 +173,9 @@ $(function() {
<li>meta_schedule_cooperative_fetch
: <a class="el" href="namespacetvm_1_1tir_1_1attr.html#a82bba3064a40ce62c958db4b120471a7">tvm::tir::attr</a>
</li>
+<li>meta_schedule_inline_rule
+: <a class="el" href="namespacetvm_1_1tir_1_1attr.html#ae4b1468e0d830201631fd97e1f944e07">tvm::tir::attr</a>
+</li>
<li>meta_schedule_layout_rewrite_preproc
: <a class="el" href="namespacetvm_1_1tir_1_1attr.html#a763184e47fc7f1423d12e5f919d932be">tvm::tir::attr</a>
</li>
@@ -213,14 +216,14 @@ $(function() {
: <a class="el" href="namespacetvm_1_1parser.html#a41a4c15c99064626719acc9332a1d039">tvm::parser</a>
</li>
<li>min()
-: <a class="el" href="namespacetvm.html#aac2abc149c1a47944c37b560181b15c0">tvm</a>
+: <a class="el" href="namespacetvm.html#acfa7fdecbf7391561b96ab5ad4ef21ed">tvm</a>
, <a class="el" href="namespacetvm_1_1topi.html#ae488679377c78cd5411b7df11c297673">tvm::topi</a>
</li>
<li>min_value()
: <a class="el" href="namespacetvm.html#a3b37fa55ea93d6868751a2441996b072">tvm</a>
</li>
<li>minimum()
-: <a class="el" href="namespacetvm_1_1topi.html#a0e19dc06a2b1ecbb83b0942fdf836169">tvm::topi</a>
+: <a class="el" href="namespacetvm_1_1topi.html#a28d4ef4b3426bff237215ce356dd5681">tvm::topi</a>
</li>
<li>MinOp()
: <a class="el" href="namespacetvm_1_1topi.html#aea9a989b0aaa2aef03fe8ee237d8257e">tvm::topi</a>
@@ -235,13 +238,13 @@ $(function() {
: <a class="el" href="namespacetvm_1_1tir_1_1builtin.html#a772fb68f083e71e635c50bb503903f22">tvm::tir::builtin</a>
</li>
<li>mod()
-: <a class="el" href="namespacetvm_1_1topi.html#aaa95d3ad68932ab206efbe0a326db6a2">tvm::topi</a>
+: <a class="el" href="namespacetvm_1_1topi.html#a87d479472e2846030990c87731f7a759">tvm::topi</a>
</li>
<li>mul()
-: <a class="el" href="namespacetvm.html#a40c70817dccaa589da0562bc8f179008">tvm</a>
+: <a class="el" href="namespacetvm.html#aaa28e92b677086d89ebfb77204bf92a2">tvm</a>
</li>
<li>multiply()
-: <a class="el" href="namespacetvm_1_1topi.html#a513a52077bacc5cb7170aab7031c0465">tvm::topi</a>
+: <a class="el" href="namespacetvm_1_1topi.html#a5a5d59f85892da838847bced0990cb38">tvm::topi</a>
</li>
</ul>
</div><!-- contents -->
diff --git a/docs/reference/api/doxygen/namespacemembers_vars.html b/docs/reference/api/doxygen/namespacemembers_vars.html
index 0105c43656..e21b56f817 100644
--- a/docs/reference/api/doxygen/namespacemembers_vars.html
+++ b/docs/reference/api/doxygen/namespacemembers_vars.html
@@ -374,6 +374,9 @@ $(function() {
<li>meta_schedule_cooperative_fetch
: <a class="el" href="namespacetvm_1_1tir_1_1attr.html#a82bba3064a40ce62c958db4b120471a7">tvm::tir::attr</a>
</li>
+<li>meta_schedule_inline_rule
+: <a class="el" href="namespacetvm_1_1tir_1_1attr.html#ae4b1468e0d830201631fd97e1f944e07">tvm::tir::attr</a>
+</li>
<li>meta_schedule_layout_rewrite_preproc
: <a class="el" href="namespacetvm_1_1tir_1_1attr.html#a763184e47fc7f1423d12e5f919d932be">tvm::tir::attr</a>
</li>
diff --git a/docs/reference/api/doxygen/namespaces.html b/docs/reference/api/doxygen/namespaces.html
index a2a2970019..cfddadd70c 100644
--- a/docs/reference/api/doxygen/namespaces.html
+++ b/docs/reference/api/doxygen/namespaces.html
@@ -77,42 +77,43 @@ $(function() {
<tr id="row_1_7_" class="even"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1parser.html" target="_self">parser</a></td><td class="desc"></td></tr>
<tr id="row_1_8_"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_8_" class="arrow" onclick="toggleFolder('1_8_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay.html" target="_self">relay</a></td><td class="desc">Relay: a high level functional IR for TVM </td></tr>
<tr id="row_1_8_0_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay_1_1attr.html" target="_self">attr</a></td><td class="desc">Namespace of the attributes that can be attached to a <a class="el" href="classtvm_1_1relay_1_1Function.html" title="Managed reference to FunctionNode. ">relay::Function</a> </td></tr>
-<tr id="row_1_8_1_"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span id="arr_1_8_1_" class="arrow" onclick="toggleFolder('1_8_1_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay_1_1qnn.html" target="_self">qnn</a></td><td class="desc"></td></tr>
-<tr id="row_1_8_1_0_" class="even"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay_1_1qnn_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
-<tr id="row_1_8_2_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
-<tr id="row_1_9_" class="even"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_9_" class="arrow" onclick="toggleFolder('1_9_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime.html" target="_self">runtime</a></td><td class="desc"></td></tr>
-<tr id="row_1_9_0_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1contrib.html" target="_self">contrib</a></td><td class="desc"></td></tr>
-<tr id="row_1_9_1_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1metadata.html" target="_self">metadata</a></td><td class="desc"></td></tr>
-<tr id="row_1_9_2_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1micro__rpc.html" target="_self">micro_rpc</a></td><td class="desc"></td></tr>
-<tr id="row_1_9_3_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1profiling.html" target="_self">profiling</a></td><td class="desc"></td></tr>
-<tr id="row_1_9_4_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1symbol.html" target="_self">symbol</a></td><td class="desc">Namespace for constant symbols </td></tr>
-<tr id="row_1_9_5_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1threading.html" target="_self">threading</a></td><td class="desc"></td></tr>
-<tr id="row_1_9_6_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1vm.html" target="_self">vm</a></td><td class="desc"></td></tr>
-<tr id="row_1_10_" class="even"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_10_" class="arrow" onclick="toggleFolder('1_10_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script.html" target="_self">script</a></td><td class="desc"></td></tr>
-<tr id="row_1_10_0_"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span id="arr_1_10_0_" class="arrow" onclick="toggleFolder('1_10_0_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder.html" target="_self">ir_builder</a></td><td class="desc"></td></tr>
-<tr id="row_1_10_0_0_" class="even"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1details.html" target="_self">details</a></td><td class="desc"></td></tr>
-<tr id="row_1_10_0_1_"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1ir.html" target="_self">ir</a></td><td class="desc"></td></tr>
-<tr id="row_1_10_0_2_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span id="arr_1_10_0_2_" class="arrow" onclick="toggleFolder('1_10_0_2_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html" target="_self">tir</a></td><td class="desc"></td></tr>
-<tr id="row_1_10_0_2_0_"><td class="entry"><span style="width:80px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir_1_1axis.html" target="_self">axis</a></td><td class="desc"></td></tr>
-<tr id="row_1_10_1_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1printer.html" target="_self">printer</a></td><td class="desc"></td></tr>
-<tr id="row_1_11_"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1support.html" target="_self">support</a></td><td class="desc"></td></tr>
-<tr id="row_1_12_" class="even"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1te.html" target="_self">te</a></td><td class="desc"><a class="el" href="classtvm_1_1te_1_1Tensor.html" title="Tensor structure representing a possible input, or intermediate computation result. ">Tensor</a> expression language DSL </td></tr>
-<tr id="row_1_13_"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_13_" class="arrow" onclick="toggleFolder('1_13_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir.html" target="_self">tir</a></td><td class="desc"></td></tr>
-<tr id="row_1_13_0_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1attr.html" target="_self">attr</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1PrimFunc.html" title="Managed reference to PrimFuncNode. ">PrimFunc</a> specific attribute names </td></tr>
-<tr id="row_1_13_1_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1builtin.html" target="_self">builtin</a></td><td class="desc">Collection of builtin intrinsics as ops </td></tr>
-<tr id="row_1_13_2_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
-<tr id="row_1_13_3_"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span id="arr_1_13_3_" class="arrow" onclick="toggleFolder('1_13_3_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1usmp.html" target="_self">usmp</a></td><td class="desc"></td></tr>
-<tr id="row_1_13_3_0_" class="even"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1usmp_1_1algo.html" target="_self">algo</a></td><td class="desc"></td></tr>
-<tr id="row_1_13_3_1_"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1usmp_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
-<tr id="row_1_14_" class="even"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_14_" class="arrow" onclick="toggleFolder('1_14_')">▼</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_14_0_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1contrib.html" target="_self">contrib</a></td><td class="desc"></td></tr>
-<tr id="row_1_14_1_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1cuda.html" target="_self">cuda</a></td><td class="desc"></td></tr>
-<tr id="row_1_14_2_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1generic.html" target="_self">generic</a></td><td class="desc"></td></tr>
-<tr id="row_1_14_3_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1nn.html" target="_self">nn</a></td><td class="desc"></td></tr>
-<tr id="row_1_14_4_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1rocm.html" target="_self">rocm</a></td><td class="desc"></td></tr>
-<tr id="row_1_14_5_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1vision.html" target="_self">vision</a></td><td class="desc"></td></tr>
-<tr id="row_1_14_6_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1x86.html" target="_self">x86</a></td><td class="desc"></td></tr>
-<tr id="row_1_15_" class="even"><td class="entry"><span style="width:32px;display:inline-block;"> </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>
+<tr id="row_1_8_1_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay_1_1legalize.html" target="_self">legalize</a></td><td class="desc"></td></tr>
+<tr id="row_1_8_2_" class="even"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span id="arr_1_8_2_" class="arrow" onclick="toggleFolder('1_8_2_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay_1_1qnn.html" target="_self">qnn</a></td><td class="desc"></td></tr>
+<tr id="row_1_8_2_0_"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay_1_1qnn_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
+<tr id="row_1_8_3_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1relay_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
+<tr id="row_1_9_"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_9_" class="arrow" onclick="toggleFolder('1_9_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime.html" target="_self">runtime</a></td><td class="desc"></td></tr>
+<tr id="row_1_9_0_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1contrib.html" target="_self">contrib</a></td><td class="desc"></td></tr>
+<tr id="row_1_9_1_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1metadata.html" target="_self">metadata</a></td><td class="desc"></td></tr>
+<tr id="row_1_9_2_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1micro__rpc.html" target="_self">micro_rpc</a></td><td class="desc"></td></tr>
+<tr id="row_1_9_3_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1profiling.html" target="_self">profiling</a></td><td class="desc"></td></tr>
+<tr id="row_1_9_4_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1symbol.html" target="_self">symbol</a></td><td class="desc">Namespace for constant symbols </td></tr>
+<tr id="row_1_9_5_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1threading.html" target="_self">threading</a></td><td class="desc"></td></tr>
+<tr id="row_1_9_6_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1runtime_1_1vm.html" target="_self">vm</a></td><td class="desc"></td></tr>
+<tr id="row_1_10_"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_10_" class="arrow" onclick="toggleFolder('1_10_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script.html" target="_self">script</a></td><td class="desc"></td></tr>
+<tr id="row_1_10_0_" class="even"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span id="arr_1_10_0_" class="arrow" onclick="toggleFolder('1_10_0_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder.html" target="_self">ir_builder</a></td><td class="desc"></td></tr>
+<tr id="row_1_10_0_0_"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1details.html" target="_self">details</a></td><td class="desc"></td></tr>
+<tr id="row_1_10_0_1_" class="even"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1ir.html" target="_self">ir</a></td><td class="desc"></td></tr>
+<tr id="row_1_10_0_2_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span id="arr_1_10_0_2_" class="arrow" onclick="toggleFolder('1_10_0_2_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html" target="_self">tir</a></td><td class="desc"></td></tr>
+<tr id="row_1_10_0_2_0_" class="even"><td class="entry"><span style="width:80px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1ir__builder_1_1tir_1_1axis.html" target="_self">axis</a></td><td class="desc"></td></tr>
+<tr id="row_1_10_1_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1script_1_1printer.html" target="_self">printer</a></td><td class="desc"></td></tr>
+<tr id="row_1_11_" class="even"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1support.html" target="_self">support</a></td><td class="desc"></td></tr>
+<tr id="row_1_12_"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1te.html" target="_self">te</a></td><td class="desc"><a class="el" href="classtvm_1_1te_1_1Tensor.html" title="Tensor structure representing a possible input, or intermediate computation result. ">Tensor</a> expression language DSL </td></tr>
+<tr id="row_1_13_" class="even"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_13_" class="arrow" onclick="toggleFolder('1_13_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir.html" target="_self">tir</a></td><td class="desc"></td></tr>
+<tr id="row_1_13_0_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1attr.html" target="_self">attr</a></td><td class="desc"><a class="el" href="classtvm_1_1tir_1_1PrimFunc.html" title="Managed reference to PrimFuncNode. ">PrimFunc</a> specific attribute names </td></tr>
+<tr id="row_1_13_1_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1builtin.html" target="_self">builtin</a></td><td class="desc">Collection of builtin intrinsics as ops </td></tr>
+<tr id="row_1_13_2_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
+<tr id="row_1_13_3_" class="even"><td class="entry"><span style="width:32px;display:inline-block;"> </span><span id="arr_1_13_3_" class="arrow" onclick="toggleFolder('1_13_3_')">▼</span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1usmp.html" target="_self">usmp</a></td><td class="desc"></td></tr>
+<tr id="row_1_13_3_0_"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1usmp_1_1algo.html" target="_self">algo</a></td><td class="desc"></td></tr>
+<tr id="row_1_13_3_1_" class="even"><td class="entry"><span style="width:64px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1tir_1_1usmp_1_1transform.html" target="_self">transform</a></td><td class="desc"></td></tr>
+<tr id="row_1_14_"><td class="entry"><span style="width:16px;display:inline-block;"> </span><span id="arr_1_14_" class="arrow" onclick="toggleFolder('1_14_')">▼</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_14_0_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1contrib.html" target="_self">contrib</a></td><td class="desc"></td></tr>
+<tr id="row_1_14_1_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1cuda.html" target="_self">cuda</a></td><td class="desc"></td></tr>
+<tr id="row_1_14_2_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1generic.html" target="_self">generic</a></td><td class="desc"></td></tr>
+<tr id="row_1_14_3_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1nn.html" target="_self">nn</a></td><td class="desc"></td></tr>
+<tr id="row_1_14_4_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1rocm.html" target="_self">rocm</a></td><td class="desc"></td></tr>
+<tr id="row_1_14_5_"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1vision.html" target="_self">vision</a></td><td class="desc"></td></tr>
+<tr id="row_1_14_6_" class="even"><td class="entry"><span style="width:48px;display:inline-block;"> </span><span class="icona"><span class="icon">N</span></span><a class="el" href="namespacetvm_1_1topi_1_1x86.html" target="_self">x86</a></td><td class="desc"></td></tr>
+<tr id="row_1_15_"><td class="entry"><span style="width:32px;display:inline-block;"> </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>
</table>
</div><!-- directory -->
</div><!-- contents -->
diff --git a/docs/reference/api/doxygen/namespacetvm_1_1relay.html b/docs/reference/api/doxygen/namespacetvm_1_1relay.html
index 33c2336152..1d1e117d8e 100644
--- a/docs/reference/api/doxygen/namespacetvm_1_1relay.html
+++ b/docs/reference/api/doxygen/namespacetvm_1_1relay.html
@@ -82,6 +82,8 @@ Namespaces</h2></td></tr>
<tr class="memitem:namespacetvm_1_1relay_1_1attr"><td class="memItemLeft" align="right" valign="top">  </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay_1_1attr.html">attr</a></td></tr>
<tr class="memdesc:namespacetvm_1_1relay_1_1attr"><td class="mdescLeft"> </td><td class="mdescRight">namespace of the attributes that can be attached to a <a class="el" href="classtvm_1_1relay_1_1Function.html" title="Managed reference to FunctionNode. ">relay::Function</a>. <br /></td></tr>
<tr class="separator:"><td class="memSeparator" colspan="2"> </td></tr>
+<tr class="memitem:namespacetvm_1_1relay_1_1legalize"><td class="memItemLeft" align="right" valign="top">  </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay_1_1legalize.html">legalize</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2"> </td></tr>
<tr class="memitem:namespacetvm_1_1relay_1_1qnn"><td class="memItemLeft" align="right" valign="top">  </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay_1_1qnn.html">qnn</a></td></tr>
<tr class="separator:"><td class="memSeparator" colspan="2"> </td></tr>
<tr class="memitem:namespacetvm_1_1relay_1_1transform"><td class="memItemLeft" align="right" valign="top">  </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay_1_1transform.html">transform</a></td></tr>
diff --git a/docs/reference/api/doxygen/namespacetvm_1_1relay_1_1legalize.html b/docs/reference/api/doxygen/namespacetvm_1_1relay_1_1legalize.html
new file mode 100644
index 0000000000..7e0232dd8f
--- /dev/null
+++ b/docs/reference/api/doxygen/namespacetvm_1_1relay_1_1legalize.html
@@ -0,0 +1,114 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+<head>
+<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
+<meta http-equiv="X-UA-Compatible" content="IE=9"/>
+<meta name="generator" content="Doxygen 1.8.13"/>
+<meta name="viewport" content="width=device-width, initial-scale=1"/>
+<title>tvm: tvm::relay::legalize Namespace Reference</title>
+<link href="tabs.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="jquery.js"></script>
+<script type="text/javascript" src="dynsections.js"></script>
+<link href="search/search.css" rel="stylesheet" type="text/css"/>
+<script type="text/javascript" src="search/searchdata.js"></script>
+<script type="text/javascript" src="search/search.js"></script>
+<link href="doxygen.css" rel="stylesheet" type="text/css" />
+</head>
+<body>
+<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
+<div id="titlearea">
+<table cellspacing="0" cellpadding="0">
+ <tbody>
+ <tr style="height: 56px;">
+ <td id="projectalign" style="padding-left: 0.5em;">
+ <div id="projectname">tvm
+ </div>
+ </td>
+ </tr>
+ </tbody>
+</table>
+</div>
+<!-- end header part -->
+<!-- Generated by Doxygen 1.8.13 -->
+<script type="text/javascript">
+var searchBox = new SearchBox("searchBox", "search",false,'Search');
+</script>
+<script type="text/javascript" src="menudata.js"></script>
+<script type="text/javascript" src="menu.js"></script>
+<script type="text/javascript">
+$(function() {
+ initMenu('',true,false,'search.php','Search');
+ $(document).ready(function() { init_search(); });
+});
+</script>
+<div id="main-nav"></div>
+<!-- window showing the filter options -->
+<div id="MSearchSelectWindow"
+ onmouseover="return searchBox.OnSearchSelectShow()"
+ onmouseout="return searchBox.OnSearchSelectHide()"
+ onkeydown="return searchBox.OnSearchSelectKey(event)">
+</div>
+
+<!-- iframe showing the search results (closed by default) -->
+<div id="MSearchResultsWindow">
+<iframe src="javascript:void(0)" frameborder="0"
+ name="MSearchResults" id="MSearchResults">
+</iframe>
+</div>
+
+<div id="nav-path" class="navpath">
+ <ul>
+<li class="navelem"><a class="el" href="namespacetvm.html">tvm</a></li><li class="navelem"><a class="el" href="namespacetvm_1_1relay.html">relay</a></li><li class="navelem"><a class="el" href="namespacetvm_1_1relay_1_1legalize.html">legalize</a></li> </ul>
+</div>
+</div><!-- top -->
+<div class="header">
+ <div class="summary">
+<a href="#func-members">Functions</a> </div>
+ <div class="headertitle">
+<div class="title">tvm::relay::legalize Namespace Reference</div> </div>
+</div><!--header-->
+<div class="contents">
+<table class="memberdecls">
+<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="func-members"></a>
+Functions</h2></td></tr>
+<tr class="memitem:aa09b6a5f5aef5c92628b13be0f8cf0a8"><td class="memItemLeft" align="right" valign="top"><a class="el" href="namespacetvm_1_1relay.html#a5b84e3790f89bb3fad5c7911eeb99531">Expr</a> </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay_1_1legalize.html#aa09b6a5f5aef5c92628b13be0f8cf0a8">Legalize</a> (const <a class="el" href="namespacetvm_1_1relay.html#a5b84e3790f89bb3fad5c7911eeb99531">Expr</a> &expr, const std::string &legali [...]
+<tr class="separator:aa09b6a5f5aef5c92628b13be0f8cf0a8"><td class="memSeparator" colspan="2"> </td></tr>
+</table>
+<h2 class="groupheader">Function Documentation</h2>
+<a id="aa09b6a5f5aef5c92628b13be0f8cf0a8"></a>
+<h2 class="memtitle"><span class="permalink"><a href="#aa09b6a5f5aef5c92628b13be0f8cf0a8">◆ </a></span>Legalize()</h2>
+
+<div class="memitem">
+<div class="memproto">
+ <table class="memname">
+ <tr>
+ <td class="memname"><a class="el" href="namespacetvm_1_1relay.html#a5b84e3790f89bb3fad5c7911eeb99531">Expr</a> tvm::relay::legalize::Legalize </td>
+ <td>(</td>
+ <td class="paramtype">const <a class="el" href="namespacetvm_1_1relay.html#a5b84e3790f89bb3fad5c7911eeb99531">Expr</a> & </td>
+ <td class="paramname"><em>expr</em>, </td>
+ </tr>
+ <tr>
+ <td class="paramkey"></td>
+ <td></td>
+ <td class="paramtype">const std::string & </td>
+ <td class="paramname"><em>legalize_map_attr_name</em> </td>
+ </tr>
+ <tr>
+ <td></td>
+ <td>)</td>
+ <td></td><td></td>
+ </tr>
+ </table>
+</div><div class="memdoc">
+
+</div>
+</div>
+</div><!-- contents -->
+<!-- start footer part -->
+<hr class="footer"/><address class="footer"><small>
+Generated by  <a href="http://www.doxygen.org/index.html">
+<img class="footer" src="doxygen.png" alt="doxygen"/>
+</a> 1.8.13
+</small></address>
+</body>
+</html>
diff --git a/docs/reference/api/doxygen/namespacetvm_1_1tir_1_1attr.html b/docs/reference/api/doxygen/namespacetvm_1_1tir_1_1attr.html
index 741b46e637..e4d2f6c81f 100644
--- a/docs/reference/api/doxygen/namespacetvm_1_1tir_1_1attr.html
+++ b/docs/reference/api/doxygen/namespacetvm_1_1tir_1_1attr.html
@@ -286,6 +286,9 @@ Variables</h2></td></tr>
<tr class="memitem:a350f417c4c3ed61f4578c5e5cb72d667"><td class="memItemLeft" align="right" valign="top">constexpr const char * </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1tir_1_1attr.html#a350f417c4c3ed61f4578c5e5cb72d667">warp_execution</a> = "warp_execution"</td></tr>
<tr class="memdesc:a350f417c4c3ed61f4578c5e5cb72d667"><td class="mdescLeft"> </td><td class="mdescRight">Mark that a block is executed by a warp. This implies the extend of threadIdx.x is warp size. <a href="#a350f417c4c3ed61f4578c5e5cb72d667">More...</a><br /></td></tr>
<tr class="separator:a350f417c4c3ed61f4578c5e5cb72d667"><td class="memSeparator" colspan="2"> </td></tr>
+<tr class="memitem:ae4b1468e0d830201631fd97e1f944e07"><td class="memItemLeft" align="right" valign="top">constexpr const char * </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1tir_1_1attr.html#ae4b1468e0d830201631fd97e1f944e07">meta_schedule_inline_rule</a> = "meta_schedule.inline_rule"</td></tr>
+<tr class="memdesc:ae4b1468e0d830201631fd97e1f944e07"><td class="mdescLeft"> </td><td class="mdescRight">Mark that a block is disallowed in auto inline. <a href="#ae4b1468e0d830201631fd97e1f944e07">More...</a><br /></td></tr>
+<tr class="separator:ae4b1468e0d830201631fd97e1f944e07"><td class="memSeparator" colspan="2"> </td></tr>
</table>
<a name="details" id="details"></a><h2 class="groupheader">Detailed Description</h2>
<div class="textblock"><p><a class="el" href="classtvm_1_1tir_1_1PrimFunc.html" title="Managed reference to PrimFuncNode. ">PrimFunc</a> specific attribute names. </p>
@@ -916,6 +919,22 @@ Variables</h2></td></tr>
<p>Mark that the loop should be further skip and bound to environment threads to enable cooperative fetching. </p>
+</div>
+</div>
+<a id="ae4b1468e0d830201631fd97e1f944e07"></a>
+<h2 class="memtitle"><span class="permalink"><a href="#ae4b1468e0d830201631fd97e1f944e07">◆ </a></span>meta_schedule_inline_rule</h2>
+
+<div class="memitem">
+<div class="memproto">
+ <table class="memname">
+ <tr>
+ <td class="memname">constexpr const char* tvm::tir::attr::meta_schedule_inline_rule = "meta_schedule.inline_rule"</td>
+ </tr>
+ </table>
+</div><div class="memdoc">
+
+<p>Mark that a block is disallowed in auto inline. </p>
+
</div>
</div>
<a id="a763184e47fc7f1423d12e5f919d932be"></a>
diff --git a/docs/reference/api/doxygen/relay_2transform_8h.html b/docs/reference/api/doxygen/relay_2transform_8h.html
index 9e44f72431..f90d7a89a9 100644
--- a/docs/reference/api/doxygen/relay_2transform_8h.html
+++ b/docs/reference/api/doxygen/relay_2transform_8h.html
@@ -108,6 +108,8 @@ Namespaces</h2></td></tr>
<tr class="separator:"><td class="memSeparator" colspan="2"> </td></tr>
<tr class="memitem:namespacetvm_1_1relay_1_1transform"><td class="memItemLeft" align="right" valign="top">  </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay_1_1transform.html">tvm::relay::transform</a></td></tr>
<tr class="separator:"><td class="memSeparator" colspan="2"> </td></tr>
+<tr class="memitem:namespacetvm_1_1relay_1_1legalize"><td class="memItemLeft" align="right" valign="top">  </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay_1_1legalize.html">tvm::relay::legalize</a></td></tr>
+<tr class="separator:"><td class="memSeparator" colspan="2"> </td></tr>
</table><table class="memberdecls">
<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="typedef-members"></a>
Typedefs</h2></td></tr>
@@ -292,6 +294,8 @@ Functions</h2></td></tr>
<tr class="memitem:a1ecbcbe35c7abd82b9eabf94f6b797d2"><td class="memItemLeft" align="right" valign="top">Expr </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay.html#a1ecbcbe35c7abd82b9eabf94f6b797d2">tvm::relay::DeDup</a> (const Expr &e)</td></tr>
<tr class="memdesc:a1ecbcbe35c7abd82b9eabf94f6b797d2"><td class="mdescLeft"> </td><td class="mdescRight">Deduplicate the bound variables and type variables in the expression. <a href="namespacetvm_1_1relay.html#a1ecbcbe35c7abd82b9eabf94f6b797d2">More...</a><br /></td></tr>
<tr class="separator:a1ecbcbe35c7abd82b9eabf94f6b797d2"><td class="memSeparator" colspan="2"> </td></tr>
+<tr class="memitem:aa09b6a5f5aef5c92628b13be0f8cf0a8"><td class="memItemLeft" align="right" valign="top">Expr </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1relay_1_1legalize.html#aa09b6a5f5aef5c92628b13be0f8cf0a8">tvm::relay::legalize::Legalize</a> (const Expr &expr, const std::string &legalize_map_attr_name)</td></tr>
+<tr class="separator:aa09b6a5f5aef5c92628b13be0f8cf0a8"><td class="memSeparator" colspan="2"> </td></tr>
</table>
<a name="details" id="details"></a><h2 class="groupheader">Detailed Description</h2>
<div class="textblock"><p>Relay specific transformation passes. </p>
diff --git a/docs/reference/api/doxygen/relay_2transform_8h_source.html b/docs/reference/api/doxygen/relay_2transform_8h_source.html
index 5b5ea24c92..397e1a9ec1 100644
--- a/docs/reference/api/doxygen/relay_2transform_8h_source.html
+++ b/docs/reference/api/doxygen/relay_2transform_8h_source.html
@@ -66,7 +66,7 @@ $(function() {
<div class="title">transform.h</div> </div>
</div><!--header-->
<div class="contents">
-<a href="relay_2transform_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="relay_2transform_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="classtvm_1_1CompilationConfig_html"><div class="ttname"><a href="classtvm_1_1CompilationConfig.html">tvm::CompilationConfig</a></div><div class="ttdoc">Managed reference class to CompilationConfig. </div><div class="ttdef"><b>Definition:</b> compilation_config.h:191</div></div>
<div class="ttc" id="namespacetvm_1_1relay_1_1transform_html_ad90e4d6ac08b62ef553755e759d398fa"><div class="ttname"><a href="namespacetvm_1_1relay_1_1transform.html#ad90e4d6ac08b62ef553755e759d398fa">tvm::relay::transform::ToCPS</a></div><div class="ttdeci">Pass ToCPS()</div><div class="ttdoc">Turn an expression into continuation passing style(CPS). </div></div>
<div class="ttc" id="namespacetvm_1_1relay_1_1transform_html_a93bbf7ab3f612d4f38a6832d6b53b4fd"><div class="ttname"><a href="namespacetvm_1_1relay_1_1transform.html#a93bbf7ab3f612d4f38a6832d6b53b4fd">tvm::relay::transform::CanonicalizeCast</a></div><div class="ttdeci">Pass CanonicalizeCast()</div><div class="ttdoc">Canonicalize cast expressions to make operator fusion more efficient. </div></div>
diff --git a/docs/reference/api/doxygen/search/all_15.js b/docs/reference/api/doxygen/search/all_15.js
index 92a58f7ef8..a59ebaaa43 100644
--- a/docs/reference/api/doxygen/search/all_15.js
+++ b/docs/reference/api/doxygen/search/all_15.js
@@ -16,6 +16,7 @@ var searchData=
['instrument',['instrument',['../namespacetvm_1_1instrument.html',1,'tvm']]],
['ir',['ir',['../namespacetvm_1_1script_1_1ir__builder_1_1ir.html',1,'tvm::script::ir_builder']]],
['ir_5fbuilder',['ir_builder',['../namespacetvm_1_1script_1_1ir__builder.html',1,'tvm::script']]],
+ ['legalize',['legalize',['../namespacetvm_1_1relay_1_1legalize.html',1,'tvm::relay']]],
['meta_5fschedule',['meta_schedule',['../namespacetvm_1_1meta__schedule.html',1,'tvm']]],
['metadata',['metadata',['../namespacetvm_1_1runtime_1_1metadata.html',1,'tvm::runtime']]],
['micro_5frpc',['micro_rpc',['../namespacetvm_1_1runtime_1_1micro__rpc.html',1,'tvm::runtime']]],
diff --git a/docs/reference/api/doxygen/search/all_d.js b/docs/reference/api/doxygen/search/all_d.js
index faa0a58f08..5483e0bf9b 100644
--- a/docs/reference/api/doxygen/search/all_d.js
+++ b/docs/reference/api/doxygen/search/all_d.js
@@ -36,7 +36,7 @@ var searchData=
['leakyreluattrs',['LeakyReluAttrs',['../structtvm_1_1relay_1_1LeakyReluAttrs.html',1,'tvm::relay']]],
['left',['left',['../classtvm_1_1relay_1_1AltPatternNode.html#a02ffb32a51710af5bf595069c2dd7a33',1,'tvm::relay::AltPatternNode']]],
['left_5fshift',['left_shift',['../namespacetvm.html#ad4fceb4266c6e7644fa373eacf73359f',1,'tvm::left_shift(PrimExpr a, PrimExpr b, Span span=Span())'],['../namespacetvm.html#a6bc108896d74f5f3b5cc3b98e9780e1c',1,'tvm::left_shift(const PrimExpr &a, int b, Span span=Span())'],['../namespacetvm.html#a58fbf68a58a7f32935d6c4539d292a08',1,'tvm::left_shift(int a, const PrimExpr &b, Span span=Span())'],['../namespacetvm_1_1topi.html#a46057526edd4bbd0a291edf7f0c863b4',1,'tvm::topi::left_ [...]
- ['legalize',['Legalize',['../namespacetvm_1_1relay_1_1qnn_1_1transform.html#a2323e3c38cc9ae1626cd98295b83e906',1,'tvm::relay::qnn::transform::Legalize()'],['../namespacetvm_1_1relay_1_1transform.html#aae623a28eda64b60c6ee90edde103891',1,'tvm::relay::transform::Legalize()']]],
+ ['legalize',['Legalize',['../namespacetvm_1_1relay_1_1qnn_1_1transform.html#a2323e3c38cc9ae1626cd98295b83e906',1,'tvm::relay::qnn::transform::Legalize()'],['../namespacetvm_1_1relay_1_1transform.html#aae623a28eda64b60c6ee90edde103891',1,'tvm::relay::transform::Legalize()'],['../namespacetvm_1_1relay_1_1legalize.html#aa09b6a5f5aef5c92628b13be0f8cf0a8',1,'tvm::relay::legalize::Legalize()']]],
['legalizepackedcalls',['LegalizePackedCalls',['../namespacetvm_1_1tir_1_1transform.html#acc84aad38c2cacfae333c8970ece933d',1,'tvm::tir::transform']]],
['length',['length',['../classtvm_1_1runtime_1_1String.html#a812045bfefb0f80f850d1b39f0e40489',1,'tvm::runtime::String::length()'],['../classtvm_1_1ObjectPathNode.html#a7fb8c3f4a9510e3893cb04bd045f93db',1,'tvm::ObjectPathNode::Length()']]],
['lengths',['lengths',['../classtvm_1_1auto__scheduler_1_1SplitStepNode.html#aac7336b4924fc50da9178c271f1cb2ed',1,'tvm::auto_scheduler::SplitStepNode']]],
diff --git a/docs/reference/api/doxygen/search/all_e.js b/docs/reference/api/doxygen/search/all_e.js
index d7c08078cc..d1709a2583 100644
--- a/docs/reference/api/doxygen/search/all_e.js
+++ b/docs/reference/api/doxygen/search/all_e.js
@@ -133,6 +133,7 @@ var searchData=
['meta_5fschedule_5fauto_5ftensorize',['meta_schedule_auto_tensorize',['../namespacetvm_1_1tir_1_1attr.html#a16f0de00900235f9c158f20f345604a4',1,'tvm::tir::attr']]],
['meta_5fschedule_5fauto_5ftensorize_5finit',['meta_schedule_auto_tensorize_init',['../namespacetvm_1_1tir_1_1attr.html#a82af28d290971bb889ea37caca78ff46',1,'tvm::tir::attr']]],
['meta_5fschedule_5fcooperative_5ffetch',['meta_schedule_cooperative_fetch',['../namespacetvm_1_1tir_1_1attr.html#a82bba3064a40ce62c958db4b120471a7',1,'tvm::tir::attr']]],
+ ['meta_5fschedule_5finline_5frule',['meta_schedule_inline_rule',['../namespacetvm_1_1tir_1_1attr.html#ae4b1468e0d830201631fd97e1f944e07',1,'tvm::tir::attr']]],
['meta_5fschedule_5flayout_5frewrite_5fpreproc',['meta_schedule_layout_rewrite_preproc',['../namespacetvm_1_1tir_1_1attr.html#a763184e47fc7f1423d12e5f919d932be',1,'tvm::tir::attr']]],
['meta_5fschedule_5flayout_5ftransform',['meta_schedule_layout_transform',['../namespacetvm_1_1topi.html#a3df09fe365656f953c8145a8cf0b3fa3',1,'tvm::topi']]],
['meta_5fschedule_5foriginal_5fshape',['meta_schedule_original_shape',['../structtvm_1_1relay_1_1Conv2DAttrs.html#a42f48616a9edb0b7e6f2a1041d9b85af',1,'tvm::relay::Conv2DAttrs::meta_schedule_original_shape()'],['../structtvm_1_1relay_1_1Conv2DWinogradAttrs.html#a1f0b75f43dc9d59e76eff063a9d6efbb',1,'tvm::relay::Conv2DWinogradAttrs::meta_schedule_original_shape()'],['../structtvm_1_1relay_1_1Conv3DAttrs.html#ab6fe84df014b5dd01caa27ef52cd3f2c',1,'tvm::relay::Conv3DAttrs::meta_schedule_ori [...]
diff --git a/docs/reference/api/doxygen/search/functions_c.js b/docs/reference/api/doxygen/search/functions_c.js
index 527b33a805..d6cff57620 100644
--- a/docs/reference/api/doxygen/search/functions_c.js
+++ b/docs/reference/api/doxygen/search/functions_c.js
@@ -15,7 +15,7 @@ var searchData=
['le',['LE',['../classtvm_1_1tir_1_1LE.html#aa55713e8c1af53b5442602821c1da5fa',1,'tvm::tir::LE']]],
['leaky_5frelu',['leaky_relu',['../namespacetvm_1_1topi.html#a54fc6163dddcf865940e1d030058a2f1',1,'tvm::topi']]],
['left_5fshift',['left_shift',['../namespacetvm.html#ad4fceb4266c6e7644fa373eacf73359f',1,'tvm::left_shift(PrimExpr a, PrimExpr b, Span span=Span())'],['../namespacetvm.html#a6bc108896d74f5f3b5cc3b98e9780e1c',1,'tvm::left_shift(const PrimExpr &a, int b, Span span=Span())'],['../namespacetvm.html#a58fbf68a58a7f32935d6c4539d292a08',1,'tvm::left_shift(int a, const PrimExpr &b, Span span=Span())'],['../namespacetvm_1_1topi.html#a46057526edd4bbd0a291edf7f0c863b4',1,'tvm::topi::left_ [...]
- ['legalize',['Legalize',['../namespacetvm_1_1relay_1_1qnn_1_1transform.html#a2323e3c38cc9ae1626cd98295b83e906',1,'tvm::relay::qnn::transform::Legalize()'],['../namespacetvm_1_1relay_1_1transform.html#aae623a28eda64b60c6ee90edde103891',1,'tvm::relay::transform::Legalize()']]],
+ ['legalize',['Legalize',['../namespacetvm_1_1relay_1_1qnn_1_1transform.html#a2323e3c38cc9ae1626cd98295b83e906',1,'tvm::relay::qnn::transform::Legalize()'],['../namespacetvm_1_1relay_1_1transform.html#aae623a28eda64b60c6ee90edde103891',1,'tvm::relay::transform::Legalize()'],['../namespacetvm_1_1relay_1_1legalize.html#aa09b6a5f5aef5c92628b13be0f8cf0a8',1,'tvm::relay::legalize::Legalize()']]],
['legalizepackedcalls',['LegalizePackedCalls',['../namespacetvm_1_1tir_1_1transform.html#acc84aad38c2cacfae333c8970ece933d',1,'tvm::tir::transform']]],
['length',['length',['../classtvm_1_1runtime_1_1String.html#a812045bfefb0f80f850d1b39f0e40489',1,'tvm::runtime::String::length()'],['../classtvm_1_1ObjectPathNode.html#a7fb8c3f4a9510e3893cb04bd045f93db',1,'tvm::ObjectPathNode::Length()']]],
['less',['less',['../namespacetvm.html#a52fa1dc57423a077eb098960162e7b85',1,'tvm::less(PrimExpr a, PrimExpr b, Span span=Span())'],['../namespacetvm.html#ab69bb15adb82d166504b9cbeef21a4dc',1,'tvm::less(const PrimExpr &a, float b, Span span=Span())'],['../namespacetvm.html#a0d5dc442dde0e69657c40803d394ea73',1,'tvm::less(float a, const PrimExpr &b, Span span=Span())'],['../namespacetvm.html#a042ab9c595a8a0f63c07dbbdf75ecf9c',1,'tvm::less(int a, const PrimExpr &b, Span span=Sp [...]
diff --git a/docs/reference/api/doxygen/search/namespaces_1.js b/docs/reference/api/doxygen/search/namespaces_1.js
index ef695be60e..e6d0140d9b 100644
--- a/docs/reference/api/doxygen/search/namespaces_1.js
+++ b/docs/reference/api/doxygen/search/namespaces_1.js
@@ -15,6 +15,7 @@ var searchData=
['instrument',['instrument',['../namespacetvm_1_1instrument.html',1,'tvm']]],
['ir',['ir',['../namespacetvm_1_1script_1_1ir__builder_1_1ir.html',1,'tvm::script::ir_builder']]],
['ir_5fbuilder',['ir_builder',['../namespacetvm_1_1script_1_1ir__builder.html',1,'tvm::script']]],
+ ['legalize',['legalize',['../namespacetvm_1_1relay_1_1legalize.html',1,'tvm::relay']]],
['meta_5fschedule',['meta_schedule',['../namespacetvm_1_1meta__schedule.html',1,'tvm']]],
['metadata',['metadata',['../namespacetvm_1_1runtime_1_1metadata.html',1,'tvm::runtime']]],
['micro_5frpc',['micro_rpc',['../namespacetvm_1_1runtime_1_1micro__rpc.html',1,'tvm::runtime']]],
diff --git a/docs/reference/api/doxygen/search/variables_c.js b/docs/reference/api/doxygen/search/variables_c.js
index f56c903571..6936134563 100644
--- a/docs/reference/api/doxygen/search/variables_c.js
+++ b/docs/reference/api/doxygen/search/variables_c.js
@@ -29,6 +29,7 @@ var searchData=
['meta_5fschedule_5fauto_5ftensorize',['meta_schedule_auto_tensorize',['../namespacetvm_1_1tir_1_1attr.html#a16f0de00900235f9c158f20f345604a4',1,'tvm::tir::attr']]],
['meta_5fschedule_5fauto_5ftensorize_5finit',['meta_schedule_auto_tensorize_init',['../namespacetvm_1_1tir_1_1attr.html#a82af28d290971bb889ea37caca78ff46',1,'tvm::tir::attr']]],
['meta_5fschedule_5fcooperative_5ffetch',['meta_schedule_cooperative_fetch',['../namespacetvm_1_1tir_1_1attr.html#a82bba3064a40ce62c958db4b120471a7',1,'tvm::tir::attr']]],
+ ['meta_5fschedule_5finline_5frule',['meta_schedule_inline_rule',['../namespacetvm_1_1tir_1_1attr.html#ae4b1468e0d830201631fd97e1f944e07',1,'tvm::tir::attr']]],
['meta_5fschedule_5flayout_5frewrite_5fpreproc',['meta_schedule_layout_rewrite_preproc',['../namespacetvm_1_1tir_1_1attr.html#a763184e47fc7f1423d12e5f919d932be',1,'tvm::tir::attr']]],
['meta_5fschedule_5foriginal_5fshape',['meta_schedule_original_shape',['../structtvm_1_1relay_1_1Conv2DAttrs.html#a42f48616a9edb0b7e6f2a1041d9b85af',1,'tvm::relay::Conv2DAttrs::meta_schedule_original_shape()'],['../structtvm_1_1relay_1_1Conv2DWinogradAttrs.html#a1f0b75f43dc9d59e76eff063a9d6efbb',1,'tvm::relay::Conv2DWinogradAttrs::meta_schedule_original_shape()'],['../structtvm_1_1relay_1_1Conv3DAttrs.html#ab6fe84df014b5dd01caa27ef52cd3f2c',1,'tvm::relay::Conv3DAttrs::meta_schedule_ori [...]
['meta_5fschedule_5fparallel',['meta_schedule_parallel',['../namespacetvm_1_1tir_1_1attr.html#afdc8510abcf9d0c3b3dd9f30046b5c0f',1,'tvm::tir::attr']]],
diff --git a/docs/reference/api/doxygen/stmt_8h.html b/docs/reference/api/doxygen/stmt_8h.html
index 60cedc29a0..9f1b009443 100644
--- a/docs/reference/api/doxygen/stmt_8h.html
+++ b/docs/reference/api/doxygen/stmt_8h.html
@@ -463,6 +463,9 @@ Variables</h2></td></tr>
<tr class="memitem:a350f417c4c3ed61f4578c5e5cb72d667"><td class="memItemLeft" align="right" valign="top">constexpr const char * </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1tir_1_1attr.html#a350f417c4c3ed61f4578c5e5cb72d667">tvm::tir::attr::warp_execution</a> = "warp_execution"</td></tr>
<tr class="memdesc:a350f417c4c3ed61f4578c5e5cb72d667"><td class="mdescLeft"> </td><td class="mdescRight">Mark that a block is executed by a warp. This implies the extend of threadIdx.x is warp size. <a href="namespacetvm_1_1tir_1_1attr.html#a350f417c4c3ed61f4578c5e5cb72d667">More...</a><br /></td></tr>
<tr class="separator:a350f417c4c3ed61f4578c5e5cb72d667"><td class="memSeparator" colspan="2"> </td></tr>
+<tr class="memitem:ae4b1468e0d830201631fd97e1f944e07"><td class="memItemLeft" align="right" valign="top">constexpr const char * </td><td class="memItemRight" valign="bottom"><a class="el" href="namespacetvm_1_1tir_1_1attr.html#ae4b1468e0d830201631fd97e1f944e07">tvm::tir::attr::meta_schedule_inline_rule</a> = "meta_schedule.inline_rule"</td></tr>
+<tr class="memdesc:ae4b1468e0d830201631fd97e1f944e07"><td class="mdescLeft"> </td><td class="mdescRight">Mark that a block is disallowed in auto inline. <a href="namespacetvm_1_1tir_1_1attr.html#ae4b1468e0d830201631fd97e1f944e07">More...</a><br /></td></tr>
+<tr class="separator:ae4b1468e0d830201631fd97e1f944e07"><td class="memSeparator" colspan="2"> </td></tr>
</table>
<a name="details" id="details"></a><h2 class="groupheader">Detailed Description</h2>
<div class="textblock"><p>TIR statements. </p>
diff --git a/docs/reference/api/doxygen/stmt_8h_source.html b/docs/reference/api/doxygen/stmt_8h_source.html
index d761498398..450af077e8 100644
--- a/docs/reference/api/doxygen/stmt_8h_source.html
+++ b/docs/reference/api/doxygen/stmt_8h_source.html
@@ -66,7 +66,7 @@ $(function() {
<div class="title">stmt.h</div> </div>
</div><!--header-->
<div class="contents">
-<a href="stmt_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 more contri [...]
+<a href="stmt_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 more contri [...]
<div class="ttc" id="classtvm_1_1tir_1_1WhileNode_html_ae897db210a499a29a66aef4eaafcd043"><div class="ttname"><a href="classtvm_1_1tir_1_1WhileNode.html#ae897db210a499a29a66aef4eaafcd043">tvm::tir::WhileNode::SEqualReduce</a></div><div class="ttdeci">bool SEqualReduce(const WhileNode *other, SEqualReducer equal) const</div><div class="ttdef"><b>Definition:</b> stmt.h:1045</div></div>
<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a2823f2e8c3ae9eec6c8f797752d1f9b5"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a2823f2e8c3ae9eec6c8f797752d1f9b5">tvm::tir::attr::pragma_import_c</a></div><div class="ttdeci">constexpr const char * pragma_import_c</div><div class="ttdoc">Import C source or file into the final code gen module. </div><div class="ttdef"><b>Definition:</b> stmt.h:1417</div></div>
<div class="ttc" id="classtvm_1_1tir_1_1Store_html"><div class="ttname"><a href="classtvm_1_1tir_1_1Store.html">tvm::tir::Store</a></div><div class="ttdoc">Managed reference to StoreNode. </div><div class="ttdef"><b>Definition:</b> stmt.h:271</div></div>
@@ -146,7 +146,7 @@ $(function() {
<div class="ttc" id="classtvm_1_1tir_1_1ForNode_html_a4fe09a4b1fb71a8ae8d5e7c807d8540b"><div class="ttname"><a href="classtvm_1_1tir_1_1ForNode.html#a4fe09a4b1fb71a8ae8d5e7c807d8540b">tvm::tir::ForNode::kind</a></div><div class="ttdeci">ForKind kind</div><div class="ttdoc">The kind of the for loop. </div><div class="ttdef"><b>Definition:</b> stmt.h:959</div></div>
<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_ad73f6ad441e1be3b2e71dead1e89f4bb"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#ad73f6ad441e1be3b2e71dead1e89f4bb">tvm::tir::attr::meta_schedule_thread_extent_high_inclusive</a></div><div class="ttdeci">constexpr const char * meta_schedule_thread_extent_high_inclusive</div><div class="ttdoc">The allowed range of thread extent in thread bindings. </div><div class="ttdef"><b>Definition:</b> stmt.h:1581</div></div>
<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_af00ba402645b1def7c543af3c48be80d"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#af00ba402645b1def7c543af3c48be80d">tvm::tir::attr::pragma_import_llvm</a></div><div class="ttdeci">constexpr const char * pragma_import_llvm</div><div class="ttdoc">Import llvm source or file into the final code gen module. </div><div class="ttdef"><b>Definition:</b> stmt.h:1419</div></div>
-<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a385e883a7cecc309d063786e5fdf2c4b"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a385e883a7cecc309d063786e5fdf2c4b">tvm::tir::attr::IsPragmaKey</a></div><div class="ttdeci">bool IsPragmaKey(const std::string &attr_key)</div><div class="ttdoc">Check if attr_key is a pragma key extension. </div><div class="ttdef"><b>Definition:</b> stmt.h:1621</div></div>
+<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a385e883a7cecc309d063786e5fdf2c4b"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a385e883a7cecc309d063786e5fdf2c4b">tvm::tir::attr::IsPragmaKey</a></div><div class="ttdeci">bool IsPragmaKey(const std::string &attr_key)</div><div class="ttdoc">Check if attr_key is a pragma key extension. </div><div class="ttdef"><b>Definition:</b> stmt.h:1624</div></div>
<div class="ttc" id="namespacetvm_1_1script_1_1ir__builder_1_1tir_html_aeb707d56c770edb33ebf73da27ebc1b9"><div class="ttname"><a href="namespacetvm_1_1script_1_1ir__builder_1_1tir.html#aeb707d56c770edb33ebf73da27ebc1b9">tvm::script::ir_builder::tir::Prefetch</a></div><div class="ttdeci">void Prefetch(Buffer buffer, Array< Range > bounds)</div><div class="ttdoc">The prefetch hint for a buffer. </div></div>
<div class="ttc" id="classtvm_1_1tir_1_1AllocateNode_html_a05bb509b87de2d9de5bfbbcc7f71f97b"><div class="ttname"><a href="classtvm_1_1tir_1_1AllocateNode.html#a05bb509b87de2d9de5bfbbcc7f71f97b">tvm::tir::AllocateNode::SEqualReduce</a></div><div class="ttdeci">bool SEqualReduce(const AllocateNode *other, SEqualReducer equal) const</div><div class="ttdef"><b>Definition:</b> stmt.h:549</div></div>
<div class="ttc" id="classtvm_1_1tir_1_1ForNode_html_a014d871983a05a0c16efd6deb83a7472"><div class="ttname"><a href="classtvm_1_1tir_1_1ForNode.html#a014d871983a05a0c16efd6deb83a7472">tvm::tir::ForNode::VisitAttrs</a></div><div class="ttdeci">void VisitAttrs(AttrVisitor *v)</div><div class="ttdef"><b>Definition:</b> stmt.h:977</div></div>
@@ -355,8 +355,9 @@ $(function() {
<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a93d76d80fd7252d66991dc650693c0ef"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a93d76d80fd7252d66991dc650693c0ef">tvm::tir::attr::fragment_shape</a></div><div class="ttdeci">constexpr const char * fragment_shape</div><div class="ttdoc">Mark that the shape of TensorCore fragment. </div><div class="ttdef"><b>Definition:</b> stmt.h:1523</div></div>
<div class="ttc" id="classtvm_1_1tir_1_1ForNode_html_ab54798257255b682a1aad74cec33e070"><div class="ttname"><a href="classtvm_1_1tir_1_1ForNode.html#ab54798257255b682a1aad74cec33e070">tvm::tir::ForNode::extent</a></div><div class="ttdeci">PrimExpr extent</div><div class="ttdoc">The extent of the iteration. </div><div class="ttdef"><b>Definition:</b> stmt.h:957</div></div>
<div class="ttc" id="classtvm_1_1runtime_1_1ObjectRef_html_a2d76fa1fb628ff276a284e61123589c5"><div class="ttname"><a href="classtvm_1_1runtime_1_1ObjectRef.html#a2d76fa1fb628ff276a284e61123589c5">tvm::runtime::ObjectRef::as</a></div><div class="ttdeci">const ObjectType * as() const</div><div class="ttdoc">Try to downcast the internal Object to a raw pointer of a corresponding type. </div><div class="ttdef"><b>Definition:</b> object.h:865</div></div>
-<div class="ttc" id="namespacetvm_1_1tir_html_a03c36414c1be2960099e023ffba09f6e"><div class="ttname"><a href="namespacetvm_1_1tir.html#a03c36414c1be2960099e023ffba09f6e">tvm::tir::ForKind2String</a></div><div class="ttdeci">const char * ForKind2String(ForKind t)</div><div class="ttdef"><b>Definition:</b> stmt.h:1638</div></div>
+<div class="ttc" id="namespacetvm_1_1tir_html_a03c36414c1be2960099e023ffba09f6e"><div class="ttname"><a href="namespacetvm_1_1tir.html#a03c36414c1be2960099e023ffba09f6e">tvm::tir::ForKind2String</a></div><div class="ttdeci">const char * ForKind2String(ForKind t)</div><div class="ttdef"><b>Definition:</b> stmt.h:1641</div></div>
<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_af08d3d2b645a914f1a64d81e45f3b86a"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#af08d3d2b645a914f1a64d81e45f3b86a">tvm::tir::attr::pragma_scope_prefix</a></div><div class="ttdeci">constexpr const char * pragma_scope_prefix</div><div class="ttdoc">Mark region is guarded by the pragma extension. </div><div class="ttdef"><b>Definition:</b> stmt.h:1415</div></div>
+<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_ae4b1468e0d830201631fd97e1f944e07"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#ae4b1468e0d830201631fd97e1f944e07">tvm::tir::attr::meta_schedule_inline_rule</a></div><div class="ttdeci">constexpr const char * meta_schedule_inline_rule</div><div class="ttdoc">Mark that a block is disallowed in auto inline. </div><div class="ttdef"><b>Definition:</b> stmt.h:1617</div></div>
<div class="ttc" id="classtvm_1_1tir_1_1BufferRealizeNode_html_ac111908806003589f64a8eb7b068272f"><div class="ttname"><a href="classtvm_1_1tir_1_1BufferRealizeNode.html#ac111908806003589f64a8eb7b068272f">tvm::tir::BufferRealizeNode::bounds</a></div><div class="ttdeci">Array< Range > bounds</div><div class="ttdoc">Bounds to be realized. </div><div class="ttdef"><b>Definition:</b> stmt.h:350</div></div>
<div class="ttc" id="namespacetvm_1_1tir_1_1attr_html_a84f5d42e968fd8f4cdd7a4aac7ba2137"><div class="ttname"><a href="namespacetvm_1_1tir_1_1attr.html#a84f5d42e968fd8f4cdd7a4aac7ba2137">tvm::tir::attr::scan_update_scope</a></div><div class="ttdeci">constexpr const char * scan_update_scope</div><div class="ttdoc">Mark of scan update scope. </div><div class="ttdef"><b>Definition:</b> stmt.h:1453</div></div>
<div class="ttc" id="classtvm_1_1tir_1_1MatchBufferRegionNode_html_a50ec46359fc889149f40ed39e0da006e"><div class="ttname"><a href="classtvm_1_1tir_1_1MatchBufferRegionNode.html#a50ec46359fc889149f40ed39e0da006e">tvm::tir::MatchBufferRegionNode::buffer</a></div><div class="ttdeci">Buffer buffer</div><div class="ttdoc">The target buffer. </div><div class="ttdef"><b>Definition:</b> stmt.h:1184</div></div>
diff --git a/docs/reference/api/python/auto_scheduler.html b/docs/reference/api/python/auto_scheduler.html
index cd7e0f338d..9a5cd92057 100644
--- a/docs/reference/api/python/auto_scheduler.html
+++ b/docs/reference/api/python/auto_scheduler.html
@@ -1615,7 +1615,7 @@ history states as starting point to perform Evolutionary Search).</p></li>
<dl class="py class">
<dt class="sig sig-object py" id="tvm.auto_scheduler.SketchPolicy">
-<em class="property"><span class="pre">class</span> </em><span class="sig-prename descclassname"><span class="pre">tvm.auto_scheduler.</span></span><span class="sig-name descname"><span class="pre">SketchPolicy</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">task</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">program_cost_model</span></span><span class="o"><span class="pre">=</span></span><span class="defau [...]
+<em class="property"><span class="pre">class</span> </em><span class="sig-prename descclassname"><span class="pre">tvm.auto_scheduler.</span></span><span class="sig-name descname"><span class="pre">SketchPolicy</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">task</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">program_cost_model</span></span><span class="o"><span class="pre">=</span></span><span class="defau [...]
<dd><p>The search policy that searches in a hierarchical search space defined by sketches.
The policy randomly samples programs from the space defined by sketches and use evolutionary
search to fine-tune them.</p>
@@ -1899,7 +1899,7 @@ Candidates:
<dl class="py function">
<dt class="sig sig-object py" id="tvm.auto_scheduler.auto_schedule">
-<span class="sig-prename descclassname"><span class="pre">tvm.auto_scheduler.</span></span><span class="sig-name descname"><span class="pre">auto_schedule</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">task</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">search_policy</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em clas [...]
+<span class="sig-prename descclassname"><span class="pre">tvm.auto_scheduler.</span></span><span class="sig-name descname"><span class="pre">auto_schedule</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">task</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">search_policy</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em clas [...]
<dd><p>THIS API IS DEPRECATED.</p>
<p>Run auto scheduling search for a task.</p>
<dl class="field-list simple">
diff --git a/docs/reference/api/python/topi.html b/docs/reference/api/python/topi.html
index f992ccca57..8a8f264f42 100644
--- a/docs/reference/api/python/topi.html
+++ b/docs/reference/api/python/topi.html
@@ -4500,85 +4500,88 @@ by the index and False otherwise</p>
<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.qnn_add_alter_layout" title="tvm.topi.nn.qnn_add_alter_layout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">qnn_add_alter_layout</span></code></a>(_attrs, _inputs, ...)</p></td>
<td><p>Change add layout.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.qnn_requantize_alter_layout" title="tvm.topi.nn.qnn_requantize_alter_layout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">qnn_requantize_alter_layout</span></code></a>(_attrs, _inputs, ...)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.qnn_conv2d_alter_layout" title="tvm.topi.nn.qnn_conv2d_alter_layout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">qnn_conv2d_alter_layout</span></code></a>(_attrs, _inputs, ...)</p></td>
+<td><p>Change qnn.conv2D layout.</p></td>
+</tr>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.qnn_requantize_alter_layout" title="tvm.topi.nn.qnn_requantize_alter_layout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">qnn_requantize_alter_layout</span></code></a>(_attrs, _inputs, ...)</p></td>
<td><p>Change requantize layout.</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.relu" title="tvm.topi.nn.relu"><code class="xref py py-obj docutils literal notranslate"><span class="pre">relu</span></code></a>(x)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.relu" title="tvm.topi.nn.relu"><code class="xref py py-obj docutils literal notranslate"><span class="pre">relu</span></code></a>(x)</p></td>
<td><p>Take relu of input x.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.scale_shift_nchw" title="tvm.topi.nn.scale_shift_nchw"><code class="xref py py-obj docutils literal notranslate"><span class="pre">scale_shift_nchw</span></code></a>(Input, Scale, Shift)</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.scale_shift_nchw" title="tvm.topi.nn.scale_shift_nchw"><code class="xref py py-obj docutils literal notranslate"><span class="pre">scale_shift_nchw</span></code></a>(Input, Scale, Shift)</p></td>
<td><p>Batch normalization operator in inference.</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.scale_shift_nchwc" title="tvm.topi.nn.scale_shift_nchwc"><code class="xref py py-obj docutils literal notranslate"><span class="pre">scale_shift_nchwc</span></code></a>(Input, Scale, Shift)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.scale_shift_nchwc" title="tvm.topi.nn.scale_shift_nchwc"><code class="xref py py-obj docutils literal notranslate"><span class="pre">scale_shift_nchwc</span></code></a>(Input, Scale, Shift)</p></td>
<td><p>Batch normalization operator in inference.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.scale_shift_nhwc" title="tvm.topi.nn.scale_shift_nhwc"><code class="xref py py-obj docutils literal notranslate"><span class="pre">scale_shift_nhwc</span></code></a>(Input, Scale, Shift)</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.scale_shift_nhwc" title="tvm.topi.nn.scale_shift_nhwc"><code class="xref py py-obj docutils literal notranslate"><span class="pre">scale_shift_nhwc</span></code></a>(Input, Scale, Shift)</p></td>
<td><p>Batch normalization operator in inference.</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.simplify" title="tvm.topi.nn.simplify"><code class="xref py py-obj docutils literal notranslate"><span class="pre">simplify</span></code></a>(expr)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.simplify" title="tvm.topi.nn.simplify"><code class="xref py py-obj docutils literal notranslate"><span class="pre">simplify</span></code></a>(expr)</p></td>
<td><p>Simplify the expression if it is Expr, directly return if it is int.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.simulated_dequantize" title="tvm.topi.nn.simulated_dequantize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">simulated_dequantize</span></code></a>(data, in_dtype[, ...])</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.simulated_dequantize" title="tvm.topi.nn.simulated_dequantize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">simulated_dequantize</span></code></a>(data, in_dtype[, ...])</p></td>
<td><p>Simulated QNN dequantize operator that mimics QNN outputs without changing datatype.</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.simulated_quantize" title="tvm.topi.nn.simulated_quantize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">simulated_quantize</span></code></a>(data, out_dtype[, ...])</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.simulated_quantize" title="tvm.topi.nn.simulated_quantize"><code class="xref py py-obj docutils literal notranslate"><span class="pre">simulated_quantize</span></code></a>(data, out_dtype[, ...])</p></td>
<td><p>Simulated QNN quantize operator that mimics QNN outputs without changing datatype.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.softmax" title="tvm.topi.nn.softmax"><code class="xref py py-obj docutils literal notranslate"><span class="pre">softmax</span></code></a>(x[, axis])</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.softmax" title="tvm.topi.nn.softmax"><code class="xref py py-obj docutils literal notranslate"><span class="pre">softmax</span></code></a>(x[, axis])</p></td>
<td><p>Perform softmax activation on the data.</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.softmax_common" title="tvm.topi.nn.softmax_common"><code class="xref py py-obj docutils literal notranslate"><span class="pre">softmax_common</span></code></a>(x, axis, use_fast_exp)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.softmax_common" title="tvm.topi.nn.softmax_common"><code class="xref py py-obj docutils literal notranslate"><span class="pre">softmax_common</span></code></a>(x, axis, use_fast_exp)</p></td>
<td><p>The common part of softmax and fast_softmax</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.space_to_batch_nd" title="tvm.topi.nn.space_to_batch_nd"><code class="xref py py-obj docutils literal notranslate"><span class="pre">space_to_batch_nd</span></code></a>(data, block_shape, ...[, ...])</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.space_to_batch_nd" title="tvm.topi.nn.space_to_batch_nd"><code class="xref py py-obj docutils literal notranslate"><span class="pre">space_to_batch_nd</span></code></a>(data, block_shape, ...[, ...])</p></td>
<td><p>Perform batch to space transformation on the data</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.space_to_depth" title="tvm.topi.nn.space_to_depth"><code class="xref py py-obj docutils literal notranslate"><span class="pre">space_to_depth</span></code></a>(data, block_size[, layout])</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.space_to_depth" title="tvm.topi.nn.space_to_depth"><code class="xref py py-obj docutils literal notranslate"><span class="pre">space_to_depth</span></code></a>(data, block_size[, layout])</p></td>
<td><p>Perform space to depth transformation on the data</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_add" title="tvm.topi.nn.sparse_add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_add</span></code></a>(dense_data, sparse_data, ...)</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_add" title="tvm.topi.nn.sparse_add"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_add</span></code></a>(dense_data, sparse_data, ...)</p></td>
<td><p>Computes sparse-dense addition</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_conv2d" title="tvm.topi.nn.sparse_conv2d"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_conv2d</span></code></a>(dense_data, sparse_data, ...)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_conv2d" title="tvm.topi.nn.sparse_conv2d"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_conv2d</span></code></a>(dense_data, sparse_data, ...)</p></td>
<td><p>Computes sparse-conv2d(1*1) of <code class="docutils literal notranslate"><span class="pre">data</span></code> and <code class="docutils literal notranslate"><span class="pre">(weight_data,</span> <span class="pre">weight_indices,</span> <span class="pre">weight_indptr)</span></code></p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_dense" title="tvm.topi.nn.sparse_dense"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_dense</span></code></a>(dense_data, sparse_data, ...[, ...])</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_dense" title="tvm.topi.nn.sparse_dense"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_dense</span></code></a>(dense_data, sparse_data, ...[, ...])</p></td>
<td><p>Computes sparse-dense matrix multiplication of <cite>data</cite> and <cite>(weight_data, weight_indices, weight_indptr).T</cite>, if sparse_lhs=False or Computes sparse-dense matrix multiplication of <cite>(data_data, data_indices, data_indptr)</cite> and <cite>weight.T</cite>, if sparse_lhs=True</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_dense_alter_layout" title="tvm.topi.nn.sparse_dense_alter_layout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_dense_alter_layout</span></code></a>(_attrs, _inputs, ...)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_dense_alter_layout" title="tvm.topi.nn.sparse_dense_alter_layout"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_dense_alter_layout</span></code></a>(_attrs, _inputs, ...)</p></td>
<td><p>Change Sparse Dense layout.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_dense_sp_lhs" title="tvm.topi.nn.sparse_dense_sp_lhs"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_dense_sp_lhs</span></code></a>(data_data, data_indices, ...)</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_dense_sp_lhs" title="tvm.topi.nn.sparse_dense_sp_lhs"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_dense_sp_lhs</span></code></a>(data_data, data_indices, ...)</p></td>
<td><p>Computes sparse-dense matrix multiplication of <cite>(data_data, data_indices, data_indptr)</cite> and <cite>weight.T</cite></p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_dense_sp_rhs" title="tvm.topi.nn.sparse_dense_sp_rhs"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_dense_sp_rhs</span></code></a>(data, weight_data, ...)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_dense_sp_rhs" title="tvm.topi.nn.sparse_dense_sp_rhs"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_dense_sp_rhs</span></code></a>(data, weight_data, ...)</p></td>
<td><p>Computes sparse-dense matrix multiplication of <cite>data</cite> and <cite>(weight_data, weight_indices, weight_indptr).T</cite></p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_transpose" title="tvm.topi.nn.sparse_transpose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_transpose</span></code></a>(sparse_data, ...)</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.sparse_transpose" title="tvm.topi.nn.sparse_transpose"><code class="xref py py-obj docutils literal notranslate"><span class="pre">sparse_transpose</span></code></a>(sparse_data, ...)</p></td>
<td><p>Transpose a square sparse matrix, <cite>A</cite> is an n-by-n sparse matrix in the CSR format.</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.strided_slice" title="tvm.topi.nn.strided_slice"><code class="xref py py-obj docutils literal notranslate"><span class="pre">strided_slice</span></code></a>(a, begin, end[, strides, ...])</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.strided_slice" title="tvm.topi.nn.strided_slice"><code class="xref py py-obj docutils literal notranslate"><span class="pre">strided_slice</span></code></a>(a, begin, end[, strides, ...])</p></td>
<td><p>Slice of an array.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.try_get_conv2d_sparse_input" title="tvm.topi.nn.try_get_conv2d_sparse_input"><code class="xref py py-obj docutils literal notranslate"><span class="pre">try_get_conv2d_sparse_input</span></code></a>(args)</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.try_get_conv2d_sparse_input" title="tvm.topi.nn.try_get_conv2d_sparse_input"><code class="xref py py-obj docutils literal notranslate"><span class="pre">try_get_conv2d_sparse_input</span></code></a>(args)</p></td>
<td><p>Analyze the input data from the given args.</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.try_get_sparse_input" title="tvm.topi.nn.try_get_sparse_input"><code class="xref py py-obj docutils literal notranslate"><span class="pre">try_get_sparse_input</span></code></a>(args)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.try_get_sparse_input" title="tvm.topi.nn.try_get_sparse_input"><code class="xref py py-obj docutils literal notranslate"><span class="pre">try_get_sparse_input</span></code></a>(args)</p></td>
<td><p>Analyze the input data from the given args.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.unpack_NCHWc_to_nchw" title="tvm.topi.nn.unpack_NCHWc_to_nchw"><code class="xref py py-obj docutils literal notranslate"><span class="pre">unpack_NCHWc_to_nchw</span></code></a>(packed_out, out_dtype)</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.unpack_NCHWc_to_nchw" title="tvm.topi.nn.unpack_NCHWc_to_nchw"><code class="xref py py-obj docutils literal notranslate"><span class="pre">unpack_NCHWc_to_nchw</span></code></a>(packed_out, out_dtype)</p></td>
<td><p>Unpack conv2d_NCHWc output from layout NCHWc to NCHW</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.upsampling" title="tvm.topi.nn.upsampling"><code class="xref py py-obj docutils literal notranslate"><span class="pre">upsampling</span></code></a>(data, scale_h, scale_w[, layout, ...])</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.upsampling" title="tvm.topi.nn.upsampling"><code class="xref py py-obj docutils literal notranslate"><span class="pre">upsampling</span></code></a>(data, scale_h, scale_w[, layout, ...])</p></td>
<td><p>Perform upsampling on the data.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.upsampling3d" title="tvm.topi.nn.upsampling3d"><code class="xref py py-obj docutils literal notranslate"><span class="pre">upsampling3d</span></code></a>(data, scale_d, scale_h, scale_w)</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.upsampling3d" title="tvm.topi.nn.upsampling3d"><code class="xref py py-obj docutils literal notranslate"><span class="pre">upsampling3d</span></code></a>(data, scale_d, scale_h, scale_w)</p></td>
<td><p>Perform upsampling on the data.</p></td>
</tr>
-<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.winograd_transform_matrices" title="tvm.topi.nn.winograd_transform_matrices"><code class="xref py py-obj docutils literal notranslate"><span class="pre">winograd_transform_matrices</span></code></a>(tile_size, ...)</p></td>
+<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.winograd_transform_matrices" title="tvm.topi.nn.winograd_transform_matrices"><code class="xref py py-obj docutils literal notranslate"><span class="pre">winograd_transform_matrices</span></code></a>(tile_size, ...)</p></td>
<td><p>Compute the A, B, and G transform matrices for <cite>tile_size</cite> as a <cite>tvm.Expr</cite>.</p></td>
</tr>
-<tr class="row-even"><td><p><a class="reference internal" href="#tvm.topi.nn.layer_norm" title="tvm.topi.nn.layer_norm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">layer_norm</span></code></a>(data, gamma, beta, axis[, epsilon])</p></td>
+<tr class="row-odd"><td><p><a class="reference internal" href="#tvm.topi.nn.layer_norm" title="tvm.topi.nn.layer_norm"><code class="xref py py-obj docutils literal notranslate"><span class="pre">layer_norm</span></code></a>(data, gamma, beta, axis[, epsilon])</p></td>
<td><p>Layer normalization operator.</p></td>
</tr>
</tbody>
@@ -7007,6 +7010,23 @@ quantized.</p>
</div>
</dd></dl>
+<dl class="py function">
+<dt class="sig sig-object py" id="tvm.topi.nn.qnn_conv2d_alter_layout">
+<span class="sig-prename descclassname"><span class="pre">tvm.topi.nn.</span></span><span class="sig-name descname"><span class="pre">qnn_conv2d_alter_layout</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">_attrs</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">_inputs</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">_tinfos</span></span></em>, <em class="sig-param"><span class="n"> [...]
+<dd><p>Change qnn.conv2D layout.
+Not to change by default</p>
+<dl class="field-list simple">
+<dt class="field-odd">Parameters</dt>
+<dd class="field-odd"><ul class="simple">
+<li><p><strong>attrs</strong> (<a class="reference internal" href="ir.html#tvm.ir.Attrs" title="tvm.ir.Attrs"><em>tvm.ir.Attrs</em></a>) – Attributes of current convolution</p></li>
+<li><p><strong>inputs</strong> (<em>tvm.relay.Expr</em>) – Grouped input symbols</p></li>
+<li><p><strong>tinfos</strong> (<a class="reference external" href="https://docs.python.org/3/library/stdtypes.html#list" title="(in Python v3.11)"><em>list</em></a>) – Input shape and dtype</p></li>
+<li><p><strong>out_type</strong> (<a class="reference external" href="https://docs.python.org/3/library/functions.html#type" title="(in Python v3.11)"><em>type</em></a>) – The output type</p></li>
+</ul>
+</dd>
+</dl>
+</dd></dl>
+
<dl class="py function">
<dt class="sig sig-object py" id="tvm.topi.nn.qnn_requantize_alter_layout">
<span class="sig-prename descclassname"><span class="pre">tvm.topi.nn.</span></span><span class="sig-name descname"><span class="pre">qnn_requantize_alter_layout</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">_attrs</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">_inputs</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">_tinfos</span></span></em>, <em class="sig-param"><span class= [...]
diff --git a/docs/reference/api/typedoc/classes/bytestreamreader.html b/docs/reference/api/typedoc/classes/bytestreamreader.html
index 7c1554d94b..a5ad5698e1 100644
--- a/docs/reference/api/typedoc/classes/bytestreamreader.html
+++ b/docs/reference/api/typedoc/classes/bytestreamreader.html
@@ -119,7 +119,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -141,7 +141,7 @@
<div class="tsd-signature tsd-kind-icon">bytes<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Uint8Array</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/rpc_server.ts#L43">rpc_server.ts:43</a></li>
</ul>
</aside>
</section>
@@ -151,7 +151,7 @@
<div class="tsd-signature tsd-kind-icon">offset<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span><span class="tsd-signature-symbol"> = 0</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/rpc_server.ts#L42">rpc_server.ts:42</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/rpc_server.ts#L42">rpc_server.ts:42</a></li>
</ul>
</aside>
</section>
@@ -168,7 +168,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/rpc_server.ts#L63">rpc_server.ts:63</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/rpc_server.ts#L63">rpc_server.ts:63</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">Uint8Array</span></h4>
@@ -185,7 +185,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/rpc_server.ts#L49">rpc_server.ts:49</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/rpc_server.ts#L49">rpc_server.ts:49</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">number</span></h4>
@@ -202,7 +202,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/rpc_server.ts#L57">rpc_server.ts:57</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/rpc_server.ts#L57">rpc_server.ts:57</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">number</span></h4>
diff --git a/docs/reference/api/typedoc/classes/cachedcallstack.html b/docs/reference/api/typedoc/classes/cachedcallstack.html
index c2d3374818..38f8bed985 100644
--- a/docs/reference/api/typedoc/classes/cachedcallstack.html
+++ b/docs/reference/api/typedoc/classes/cachedcallstack.html
@@ -144,7 +144,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L223">memory.ts:223</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L223">memory.ts:223</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -172,7 +172,7 @@
<div class="tsd-signature tsd-kind-icon">temp<wbr>Args<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Array</span><span class="tsd-signature-symbol"><</span><a href="../interfaces/disposable.html" class="tsd-signature-type">Disposable</a><span class="tsd-signature-symbol">></span><span class="tsd-signature-symbol"> = []</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L208">memory.ts:208</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L208">memory.ts:208</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -194,7 +194,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L312">memory.ts:312</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L312">memory.ts:312</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -226,7 +226,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L284">memory.ts:284</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L284">memory.ts:284</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -262,7 +262,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L388">memory.ts:388</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L388">memory.ts:388</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -300,7 +300,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L376">memory.ts:376</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L376">memory.ts:376</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -340,7 +340,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L267">memory.ts:267</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L267">memory.ts:267</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -373,7 +373,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L243">memory.ts:243</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L243">memory.ts:243</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">void</span></h4>
@@ -390,7 +390,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L321">memory.ts:321</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L321">memory.ts:321</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -422,7 +422,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L252">memory.ts:252</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L252">memory.ts:252</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -444,7 +444,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L359">memory.ts:359</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L359">memory.ts:359</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -470,7 +470,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L342">memory.ts:342</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L342">memory.ts:342</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -496,7 +496,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L350">memory.ts:350</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L350">memory.ts:350</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -522,7 +522,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L326">memory.ts:326</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L326">memory.ts:326</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -548,7 +548,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L363">memory.ts:363</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L363">memory.ts:363</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -574,7 +574,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L346">memory.ts:346</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L346">memory.ts:346</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -600,7 +600,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/memory.ts#L334">memory.ts:334</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/memory.ts#L334">memory.ts:334</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
diff --git a/docs/reference/api/typedoc/classes/dldatatype.html b/docs/reference/api/typedoc/classes/dldatatype.html
index 817f4758e4..7fa329ef02 100644
--- a/docs/reference/api/typedoc/classes/dldatatype.html
+++ b/docs/reference/api/typedoc/classes/dldatatype.html
@@ -119,7 +119,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L262">runtime.ts:262</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L262">runtime.ts:262</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -147,7 +147,7 @@
<div class="tsd-signature tsd-kind-icon">bits<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L260">runtime.ts:260</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L260">runtime.ts:260</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -162,7 +162,7 @@
<div class="tsd-signature tsd-kind-icon">code<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L258">runtime.ts:258</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L258">runtime.ts:258</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -177,7 +177,7 @@
<div class="tsd-signature tsd-kind-icon">lanes<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L262">runtime.ts:262</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L262">runtime.ts:262</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -199,7 +199,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L279">runtime.ts:279</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L279">runtime.ts:279</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">number</span></h4>
@@ -216,7 +216,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L270">runtime.ts:270</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L270">runtime.ts:270</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">string</span></h4>
diff --git a/docs/reference/api/typedoc/classes/dldevice.html b/docs/reference/api/typedoc/classes/dldevice.html
index ca8a196155..eb3a7d0b0c 100644
--- a/docs/reference/api/typedoc/classes/dldevice.html
+++ b/docs/reference/api/typedoc/classes/dldevice.html
@@ -118,7 +118,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L202">runtime.ts:202</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L202">runtime.ts:202</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -146,7 +146,7 @@
<div class="tsd-signature tsd-kind-icon">device<wbr>Id<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L200">runtime.ts:200</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L200">runtime.ts:200</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -161,7 +161,7 @@
<div class="tsd-signature tsd-kind-icon">device<wbr>Type<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">number</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L198">runtime.ts:198</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L198">runtime.ts:198</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -183,7 +183,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L223">runtime.ts:223</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L223">runtime.ts:223</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -205,7 +205,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L230">runtime.ts:230</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L230">runtime.ts:230</a></li>
</ul>
</aside>
<h4 class="tsd-returns-title">Returns <span class="tsd-signature-type">string</span></h4>
diff --git a/docs/reference/api/typedoc/classes/environment.html b/docs/reference/api/typedoc/classes/environment.html
index edfdf11916..e581a97c91 100644
--- a/docs/reference/api/typedoc/classes/environment.html
+++ b/docs/reference/api/typedoc/classes/environment.html
@@ -125,7 +125,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/environment.ts#L86">environment.ts:86</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/environment.ts#L86">environment.ts:86</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -169,7 +169,7 @@
<aside class="tsd-sources">
<p>Implementation of <a href="../interfaces/libraryprovider.html">LibraryProvider</a>.<a href="../interfaces/libraryprovider.html#imports">imports</a></p>
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/environment.ts#L70">environment.ts:70</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/environment.ts#L70">environment.ts:70</a></li>
</ul>
</aside>
</section>
@@ -179,7 +179,7 @@
<div class="tsd-signature tsd-kind-icon">logger<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-symbol">(</span>msg<span class="tsd-signature-symbol">: </span><span class="tsd-signature-type">string</span><span class="tsd-signature-symbol">)</span><span class="tsd-signature-symbol"> => </span><span class="tsd-signature-type">void</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/environment.ts#L69">environment.ts:69</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/environment.ts#L69">environment.ts:69</a></li>
</ul>
</aside>
<div class="tsd-type-declaration">
@@ -210,7 +210,7 @@
<div class="tsd-signature tsd-kind-icon">packedCFunc<wbr>Table<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Array</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">ctypes.FTVMWasmPackedCFunc</span><span class="tsd-signature-symbol"> | </span><span class="tsd-signature-type">undefined</span><span class="tsd-signature-symbol">></span><span class="tsd-signature-symbol"> = [undefined,]</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/environment.ts#L78">environment.ts:78</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/environment.ts#L78">environment.ts:78</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -228,7 +228,7 @@
<div class="tsd-signature tsd-kind-icon">packedCFunc<wbr>Table<wbr>Free<wbr>Id<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Array</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">number</span><span class="tsd-signature-symbol">></span><span class="tsd-signature-symbol"> = []</span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/environment.ts#L84">environment.ts:84</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/environment.ts#L84">environment.ts:84</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
@@ -250,7 +250,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/environment.ts#L105">environment.ts:105</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/environment.ts#L105">environment.ts:105</a></li>
</ul>
</aside>
<div class="tsd-comment tsd-typography">
diff --git a/docs/reference/api/typedoc/classes/ffilibrary.html b/docs/reference/api/typedoc/classes/ffilibrary.html
index a6039496bc..5cd23bb9a3 100644
--- a/docs/reference/api/typedoc/classes/ffilibrary.html
+++ b/docs/reference/api/typedoc/classes/ffilibrary.html
@@ -131,7 +131,7 @@
<li class="tsd-description">
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L49">runtime.ts:49</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L49">runtime.ts:49</a></li>
</ul>
</aside>
<h4 class="tsd-parameters-title">Parameters</h4>
@@ -156,7 +156,7 @@
<div class="tsd-signature tsd-kind-icon">exports<span class="tsd-signature-symbol">:</span> <span class="tsd-signature-type">Record</span><span class="tsd-signature-symbol"><</span><span class="tsd-signature-type">string</span><span class="tsd-signature-symbol">, </span><span class="tsd-signature-type">Function</span><span class="tsd-signature-symbol">></span></div>
<aside class="tsd-sources">
<ul>
- <li>Defined in <a href="https://github.com/apache/tvm/blob/77b6f0eec/web/src/runtime.ts#L46">runtime.ts:46</a></li>
+ <li>Defined in <a href="https://github.com/apache/tvm/blob/15e185d92/web/src/runtime.ts#L46">runtime.ts:46</a></li>
</ul>
</aside>
</section>
@@ -166,7 +166,7 @@
<div class="tsd-signature tsd-kind-icon">memory<span class="tsd-signature-symbol">:</span> <a href="memory.html" class="tsd-signature-type">Memory</a></div>
... 2525 lines suppressed ...