You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tvm.apache.org by ma...@apache.org on 2022/04/20 23:20:16 UTC
[tvm] branch ci-gpu-update updated (1c76aa4c33 -> 64d2f9416d)
This is an automated email from the ASF dual-hosted git repository.
masahi pushed a change to branch ci-gpu-update
in repository https://gitbox.apache.org/repos/asf/tvm.git
omit 1c76aa4c33 [CI] Update GPU image
add 01e0e2d0e1 [CI] Update GPU image (#10992)
add ef163a5791 [Hexagon] Remove HexagonBuffer external constructor and support (#10978)
add 814e856851 sort axes (#10985)
add 11b8cd3ca1 [ONNX] Add imports for BERT contrib operators (#10949)
add dbfab5c10d [Metaschedule] Make custom schedule_rule registration optional (#10975)
add 1bfb9cac93 [COMMUNITY] @yzh119 -> Reviewer (#10993)
add ce8f83e3c5 [hexagon] 'add_hvx' test to explore HVX usage. (#10604)
add e370ed4597 [Hexagon] Less aggressive adb state clean up (#10909)
add 985fc933f4 [Hexagon] Handle TCP server binding to unknown port (#10945)
add 52f52c83de [BYOC][ACL] Fix list is not supported as an input node (#10801)
add 96616b7af8 [Hexagon] Add top-level CMakeLists.txt for apps/hexagon_launcher (#11006)
add 1cf0c0a5bf [CUDNN] Add partitioning support for fused conv2d+bias+act (#10997)
add 324bf4cac5 Add driazati to triagers. (#11004)
add e84b3590dc [Relay] Refactor inline composites transformation (#10995)
add 529da9bd56 [TensorRT][BYOC] Minor refactoring to handle constants in pattern-based ops for TRT (#10994)
add 3d63b2d944 [CI] Update CI Images to include `pytest-lazy-fixture` (#10999)
add a34731b7fc [ROCM] DP4A intrinsic support for TE/TIR (#11009)
add b94119692e [TIR] Ignore Allocate/AllocateConst in BufferAllocationLocator (#10998)
add 0e1a2a2ff8 Fix broken CI when git-merge needs to create a commit. (#11007)
add 29774bddd8 [RELAY][FRONTEND] Initial OneFlow frontend support. (#8790)
add a9d86e61b6 [Metaschedule] Support tuning on rocm and vulkan target (#11017)
add 715f24d238 [Metaschedule] Enable continuing tuning after schedule application failure (#10937)
add b99a096825 [Metaschedule, Refactor] Move MultiLevelTilingNode decl to a header (#11020)
add f238900e6b Add Havisha to triagers and alphabetize. (#11005)
add 8bfe3bbb3c [Arith] Updated arith::DetectIterMap to keep extent=1 components (#10980)
add 37db213a84 [QNNParam] Refactor the implmentation of QNNParam (#11011)
add 365fcc832d Use TVM log instead of hexagon_print (#11024)
add 8aafe5b109 [OpenCL] Fix type casting error (#11021)
add 351f31b51c [Runtime][PipelineExecutor]Add forwarding queue logic for set input. (#10990)
add fafabc96c1 [VirtualMachine] Zero copy in set_input when input is DLTensor (#11003)
add f92d351e11 [COMMUNITY] @wrongtest -> Committer (#11028)
add 7d9b7bbd50 [ci] Always assume num executors == 1 (#11014)
add 69ba7ed640 [BugFix][TIR] Fix narrower dtype of loop vars in CreatePrimFunc (#11030)
add 8d868f6bf3 [BugFix][TIR] Fix rfactor when RF block becomes spatial (#11031)
add 9c2df39376 [BugFix][TIR] Error check: Inline Block with Init Stmt (#11033)
add 9f3da1cbae [Frontend][Paddle] Fix pool2d op (#11029)
add 0b9bcf0e7a [MetaSchedule][Refactor] Introduce TuneConfig (#10986)
add 240fac437c Fix typo in tutorial doc (#10974)
add 28aac0e48f [Hexagon] Deprecate USE_HEXAGON_DEVICE, introduce USE_HEXAGON (#11025)
add 1e524d0670 [ci] Migrate all test steps to macros (#10968)
add 1542286826 [ci] Add branch protections to .asf.yaml (#10964)
add c2eb51311c [COMMUNITY] New Committer -- lhutton1 (#11049)
add f6171914a7 [Hexagon] Adjust RPC read buffer size from python (#11022)
add 0cd4fa688c Register PackedFuncObj with type registry. (#11039)
add 4178617fcf [AOT] Support LLVM backend with C++ runtime (#10753)
add 71067799ef [TE][TIR] Enable CreatePrimFunc to properly handle 'layout_free_placeholder' (#11054)
add fe948da88e Upgrad oneflow version (#11052)
add e2dd0f8251 Let remote RPCModule get function recursively (#11053)
add 94f28b29c5 [TVMC] Add `--config` argument for config files (#11012)
add 81073962ef Revert "[OpenCL] Fix type casting error (#11021)" (#11035)
add 312b552b11 Add FlattenAtrousConv transformation (#10996)
add 557fc6c107 Hotfix CI (black check not caught by PR CI) (#11056)
add 24d5539c43 [Hexagon] Pass stack size to simulator (#11046)
add a945586868 [Hexagon] Refactor test scripts (#11026)
add 5987982fae Attempt to prevent concurrent update in Map (#9842)
add 3b5189563c Add ekalda to reviewers. (#11061)
add 1efd7df370 [ci] Remove TensorCore node name (#11048)
add 0bf7f3313a [OpenCL] Fix type casting (#11038)
add b78e44e42d [TOPI] Fix nn.lrn result dtype on fp16 (#11032)
add 2025e368c4 [CMSIS-NN] Add Arm(R) Cortex(R)-M55 CPU and CMSIS-NN demo (#11013)
add 3823b39b8a [TIR] Utility function to decide loop mapping for auto tensorization (#11050)
add 8cf0c3e992 Fix While Node StructuralEqual and StructuralHash issue (#11073)
add 0b957802b0 [microNPU] Cascader performance model bugfixes (#10510)
add 970f868272 [TVMC] Add configuration json files to the Python package (#11063)
add 5b586605da Better version handling for Arduino (#11043)
add 58b7a5a268 [QNN] Support input scale and zp of 1-element vector in qnn.conv2d_transpose (#10952)
add 7612b22cf5 [Frontend][ONNX]support Pool2D layout is CHW (#11034)
new 64d2f9416d [CI] Update GPU image for oneflow v0.7
This update added new revisions after undoing existing revisions.
That is to say, some revisions that were in the old version of the
branch are not in the new version. This situation occurs
when a user --force pushes a change and generates a repository
containing something like this:
* -- * -- B -- O -- O -- O (1c76aa4c33)
\
N -- N -- N refs/heads/ci-gpu-update (64d2f9416d)
You should already have received notification emails for all of the O
revisions, and so the following emails describe only the N revisions
from the common base, B.
Any revisions marked "omit" are not gone; other references still
refer to them. Any revisions marked "discard" are gone forever.
The 1 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "add" were already present in the repository and have only
been added to this reference.
Summary of changes:
.asf.yaml | 13 +
CMakeLists.txt | 25 +-
CONTRIBUTORS.md | 4 +
Jenkinsfile | 199 ++-
apps/hexagon_api/CMakeLists.txt | 4 +
apps/hexagon_launcher/CMakeLists.txt | 78 +
apps/hexagon_launcher/cmake/android/CMakeLists.txt | 9 +-
apps/hexagon_launcher/cmake/hexagon/CMakeLists.txt | 7 +-
.../template_project/microtvm_api_server.py | 104 +-
.../tests/test_arduino_microtvm_api_server.py | 82 +-
apps/microtvm/{ethosu => cmsisnn}/.gitignore | 1 -
apps/microtvm/{ethosu => cmsisnn}/Makefile | 49 +-
apps/microtvm/{ethosu => cmsisnn}/README.md | 34 +-
.../{ethosu => cmsisnn}/arm-none-eabi-gcc.cmake | 0
apps/microtvm/{ethosu => cmsisnn}/convert_image.py | 11 +-
apps/microtvm/{ethosu => cmsisnn}/corstone300.ld | 8 +-
.../{ethosu => cmsisnn}/include/crt_config.h | 0
.../{ethosu => cmsisnn}/include/tvm_runtime.h | 0
apps/microtvm/{ethosu => cmsisnn}/requirements.txt | 0
apps/microtvm/{ethosu => cmsisnn}/run_demo.sh | 56 +-
.../{ethosu => cmsisnn}/src/demo_bare_metal.c | 40 +-
cmake/config.cmake | 13 +-
cmake/modules/Hexagon.cmake | 90 +-
cmake/modules/LibInfo.cmake | 2 +-
configs/host/default.json | 7 +
configs/test/compile_config_test.json | 9 +
configs/test/tune_config_test.json | 6 +
docker/install/ubuntu_install_oneflow.sh | 4 +-
gallery/tutorial/tvmc_python.py | 12 +-
include/tvm/arith/iter_affine_map.h | 4 +-
include/tvm/runtime/container/map.h | 37 +-
include/tvm/runtime/metadata.h | 2 +
include/tvm/runtime/metadata_base.h | 31 +-
include/tvm/runtime/ndarray.h | 18 +
include/tvm/tir/stmt.h | 6 +-
include/tvm/topi/nn/local_response_norm.h | 6 +-
jenkins/Jenkinsfile.j2 | 211 +--
python/setup.py | 7 +
python/tvm/arith/iter_affine_map.py | 17 +-
.../tvm/contrib/ethosu/cascader/device_config.py | 49 +-
python/tvm/contrib/hexagon/build.py | 114 +-
python/tvm/contrib/hexagon/session.py | 5 +-
python/tvm/contrib/pipeline_executor.py | 5 +-
python/tvm/driver/tvmc/autotuner.py | 5 +-
python/tvm/driver/tvmc/compiler.py | 5 +-
python/tvm/driver/tvmc/config_options.py | 195 +++
python/tvm/driver/tvmc/main.py | 14 +-
python/tvm/driver/tvmc/micro.py | 5 +-
python/tvm/driver/tvmc/runner.py | 5 +-
python/tvm/driver/tvmc/target.py | 2 +-
python/tvm/meta_schedule/__init__.py | 9 +-
.../search_strategy/evolutionary_search.py | 14 +-
.../meta_schedule/task_scheduler/round_robin.py | 5 +
.../testing/tune_relay_meta_schedule.py | 56 +-
.../meta_schedule/testing/tune_te_meta_schedule.py | 4 +-
python/tvm/meta_schedule/tune.py | 575 +++---
python/tvm/relay/frontend/__init__.py | 1 +
python/tvm/relay/frontend/oneflow.py | 1821 ++++++++++++++++++++
python/tvm/relay/frontend/onnx.py | 241 ++-
python/tvm/relay/frontend/paddlepaddle.py | 12 +-
python/tvm/relay/frontend/qnn_torch.py | 9 +-
python/tvm/relay/op/contrib/arm_compute_lib.py | 37 +-
python/tvm/relay/op/contrib/cudnn.py | 79 +-
python/tvm/relay/op/contrib/tensorrt.py | 38 +-
python/tvm/relay/op/strategy/cuda.py | 8 +-
python/tvm/relay/op/strategy/rocm.py | 172 +-
python/tvm/relay/qnn/op/legalizations.py | 52 +-
python/tvm/relay/transform/transform.py | 27 +
python/tvm/runtime/vm.py | 4 +
python/tvm/script/tir/special_stmt.py | 21 +-
python/tvm/testing/tir.py | 13 +
python/tvm/tir/schedule/analysis.py | 33 +-
python/tvm/tir/stmt_functor.py | 12 +
python/tvm/tir/tensor_intrin/__init__.py | 2 +
python/tvm/tir/tensor_intrin/dot_product_common.py | 55 +
python/tvm/tir/tensor_intrin/rocm.py | 47 +
python/tvm/topi/cuda/batch_matmul.py | 7 +-
python/tvm/topi/cuda/conv2d_alter_op.py | 12 +-
python/tvm/topi/cuda/conv2d_int8.py | 4 +-
python/tvm/topi/cuda/dense.py | 5 +-
python/tvm/topi/cuda/tensor_intrin.py | 23 +-
python/tvm/topi/rocm/dense.py | 79 +-
python/tvm/topi/utils.py | 7 +
src/arith/iter_affine_map.cc | 15 +-
src/contrib/ethosu/cascader/parts/ethosu.cc | 24 +-
.../schedule_rule/multi_level_tiling.cc | 205 +--
.../schedule_rule/multi_level_tiling.h | 187 ++
.../space_generator/post_order_apply.cc | 22 +-
src/meta_schedule/task_scheduler/task_scheduler.cc | 1 +
src/meta_schedule/utils.h | 12 +-
src/printer/tir_text_printer.cc | 11 +
src/relay/backend/aot_executor_codegen.cc | 276 +--
.../backend/contrib/arm_compute_lib/codegen.cc | 26 +
src/relay/qnn/op/convolution_transpose.cc | 14 +-
src/relay/qnn/utils.h | 6 +
src/relay/transforms/flatten_atrous_conv.cc | 195 +++
src/relay/transforms/inline_composites.cc | 79 +-
src/runtime/contrib/arm_compute_lib/acl_runtime.cc | 82 +-
src/runtime/contrib/arm_compute_lib/acl_utils.cc | 16 +-
src/runtime/contrib/arm_compute_lib/acl_utils.h | 9 +-
src/runtime/contrib/cudnn/conv_forward.cc | 62 +
src/runtime/contrib/cudnn/cudnn_utils.cc | 4 +
src/runtime/contrib/cudnn/cudnn_utils.h | 2 +
src/runtime/contrib/json/json_runtime.h | 1 +
src/runtime/hexagon/README.md | 70 +-
src/runtime/hexagon/android/hexagon_device_api.cc | 2 +-
src/runtime/hexagon/hexagon/hexagon_buffer.cc | 33 +-
src/runtime/hexagon/hexagon/hexagon_buffer.h | 12 -
src/runtime/hexagon/hexagon/hexagon_common.cc | 9 +-
.../hexagon/hexagon/hexagon_device_api_v2.cc | 9 +-
src/runtime/hexagon/rpc/android/session.cc | 15 +-
src/runtime/hexagon/rpc/hexagon/rpc_server.cc | 72 +-
src/runtime/hexagon/rpc/hexagon_rpc.idl | 1 +
src/runtime/hexagon/rpc/simulator/session.cc | 14 +-
src/runtime/metadata.cc | 15 +-
src/runtime/ndarray.cc | 31 +
.../utils.cc => runtime/packed_func.cc} | 20 +-
src/runtime/pipeline/pipeline_executor.cc | 20 +-
src/runtime/pipeline/pipeline_executor.h | 9 +-
src/runtime/pipeline/pipeline_scheduler.cc | 15 +-
src/runtime/pipeline/pipeline_scheduler.h | 9 +-
src/runtime/pipeline/pipeline_struct.h | 296 ++--
src/runtime/rpc/rpc_module.cc | 2 +-
src/runtime/vm/vm.cc | 22 +-
src/support/libinfo.cc | 6 +-
src/target/llvm/codegen_cpu.cc | 362 +++-
src/target/llvm/codegen_cpu.h | 10 +-
src/target/llvm/codegen_llvm.cc | 85 +-
src/target/llvm/codegen_llvm.h | 42 +-
src/target/llvm/llvm_common.cc | 7 +
src/target/llvm/llvm_common.h | 2 +
src/target/llvm/llvm_module.cc | 41 +-
src/target/llvm/llvm_module.h | 3 +
src/target/metadata.h | 11 +-
src/target/metadata_module.cc | 6 +
src/target/metadata_utils.cc | 155 ++
src/target/metadata_utils.h | 141 ++
src/target/source/codegen_c_host.cc | 28 +-
src/target/source/codegen_opencl.cc | 38 +
src/target/source/codegen_opencl.h | 4 +
src/target/source/source_module.cc | 232 ++-
src/target/target_kind.cc | 6 +
src/te/operation/create_primfunc.cc | 50 +-
src/tir/ir/expr.cc | 2 +-
src/tir/ir/index_map.cc | 4 +-
src/tir/ir/stmt_functor.cc | 4 +
src/tir/schedule/analysis.h | 33 +
src/tir/schedule/analysis/analysis.cc | 167 +-
src/tir/schedule/primitive/compute_inline.cc | 26 +
src/tir/schedule/primitive/reduction.cc | 34 +-
src/tir/transforms/legalize_packed_calls.cc | 78 +-
src/tir/transforms/lower_tvm_builtin.cc | 103 +-
.../plan_update_buffer_allocation_location.cc | 33 +-
.../convert_pool_allocations_to_offsets.cc | 11 +-
tests/cpp/aot_metadata_test.cc | 183 +-
tests/cpp/container_test.cc | 15 +
tests/cpp/runtime/hexagon_buffer.cc | 34 -
.../contrib/test_arm_compute_lib/infrastructure.py | 28 +-
.../test_arm_compute_lib/test_concatenate.py | 151 ++
tests/python/contrib/test_cudnn.py | 51 +-
.../cascader/test_ethosu_block_config.py | 4 +-
.../test_ethosu/cascader/test_ethosu_part.py | 2 +
.../cascader/test_ethosu_part_performance.py | 1 +
.../contrib/test_hexagon/benchmark_hexagon.py | 335 ++++
tests/python/contrib/test_hexagon/conftest.py | 6 +-
.../test_hexagon/test_2d_physical_buffers.py | 3 -
.../contrib/test_hexagon/test_cache_read_write.py | 3 -
tests/python/contrib/test_hexagon/test_launcher.md | 2 +-
tests/python/contrib/test_hexagon/test_launcher.py | 25 +-
tests/python/driver/tvmc/test_command_line.py | 36 +
tests/python/driver/tvmc/test_parse_config_file.py | 155 ++
tests/python/frontend/oneflow/test_forward.py | 723 ++++++++
tests/python/frontend/onnx/test_forward.py | 223 ++-
tests/python/relay/aot/test_c_device_api.py | 23 +-
tests/python/relay/aot/test_cpp_aot.py | 47 +-
tests/python/relay/aot/test_crt_aot.py | 2 +-
tests/python/relay/test_op_level1.py | 38 +
tests/python/relay/test_op_level10.py | 35 +
tests/python/relay/test_op_level2.py | 58 +-
tests/python/relay/test_op_qnn_conv2_transpose.py | 25 +
.../python/relay/test_pass_flatten_atrous_conv.py | 427 +++++
tests/python/relay/test_pipeline_executor.py | 21 +-
tests/python/topi/python/test_topi_conv2d_int8.py | 13 +-
tests/python/topi/python/test_topi_dense.py | 1 -
tests/python/topi/python/test_topi_lrn.py | 8 +-
.../unittest/test_aot_legalize_packed_call.py | 88 +-
.../test_meta_schedule_measure_callback.py | 4 +-
.../test_meta_schedule_post_order_apply.py | 4 +-
.../unittest/test_meta_schedule_search_strategy.py | 2 +
.../unittest/test_meta_schedule_task_scheduler.py | 2 +
.../unittest/test_meta_schedule_tune_relay.py | 72 +-
.../python/unittest/test_meta_schedule_tune_te.py | 5 +-
.../python/unittest/test_meta_schedule_tune_tir.py | 14 +-
.../python/unittest/test_target_codegen_opencl.py | 46 +
tests/python/unittest/test_te_create_primfunc.py | 59 +-
.../python/unittest/test_tir_schedule_analysis.py | 183 +-
.../unittest/test_tir_schedule_compute_inline.py | 44 +
tests/python/unittest/test_tir_schedule_rfactor.py | 63 +-
.../python/unittest/test_tir_schedule_tensorize.py | 50 +
.../unittest/test_tir_structural_equal_hash.py | 10 +
.../test_tir_transform_extract_constants.py | 2 +
...ransform_convert_pool_allocations_to_offsets.py | 99 +-
tests/python/unittest/test_transform_layout.py | 30 +
.../python/unittest/test_tvmscript_error_report.py | 22 +
.../python/unittest/test_tvmscript_syntax_sugar.py | 17 +
tests/scripts/task_build.py | 5 +-
tests/scripts/task_config_build_hexagon.sh | 3 +-
tests/scripts/task_demo_microtvm.sh | 4 +
tests/scripts/task_python_frontend.sh | 3 +
tests/scripts/task_python_hexagon.sh | 28 +-
210 files changed, 9679 insertions(+), 2601 deletions(-)
create mode 100644 apps/hexagon_launcher/CMakeLists.txt
copy apps/microtvm/{ethosu => cmsisnn}/.gitignore (67%)
copy apps/microtvm/{ethosu => cmsisnn}/Makefile (67%)
copy apps/microtvm/{ethosu => cmsisnn}/README.md (71%)
copy apps/microtvm/{ethosu => cmsisnn}/arm-none-eabi-gcc.cmake (100%)
copy apps/microtvm/{ethosu => cmsisnn}/convert_image.py (86%)
copy apps/microtvm/{ethosu => cmsisnn}/corstone300.ld (98%)
copy apps/microtvm/{ethosu => cmsisnn}/include/crt_config.h (100%)
copy apps/microtvm/{ethosu => cmsisnn}/include/tvm_runtime.h (100%)
copy apps/microtvm/{ethosu => cmsisnn}/requirements.txt (100%)
copy apps/microtvm/{ethosu => cmsisnn}/run_demo.sh (67%)
copy apps/microtvm/{ethosu => cmsisnn}/src/demo_bare_metal.c (56%)
create mode 100644 configs/host/default.json
create mode 100644 configs/test/compile_config_test.json
create mode 100644 configs/test/tune_config_test.json
create mode 100644 python/tvm/driver/tvmc/config_options.py
create mode 100644 python/tvm/relay/frontend/oneflow.py
create mode 100644 python/tvm/tir/tensor_intrin/dot_product_common.py
create mode 100644 python/tvm/tir/tensor_intrin/rocm.py
create mode 100644 src/meta_schedule/schedule_rule/multi_level_tiling.h
create mode 100644 src/relay/transforms/flatten_atrous_conv.cc
copy src/{auto_scheduler/utils.cc => runtime/packed_func.cc} (76%)
mode change 100755 => 100644
create mode 100644 src/target/metadata_utils.cc
create mode 100644 src/target/metadata_utils.h
create mode 100644 tests/python/contrib/test_arm_compute_lib/test_concatenate.py
create mode 100644 tests/python/contrib/test_hexagon/benchmark_hexagon.py
create mode 100644 tests/python/driver/tvmc/test_parse_config_file.py
create mode 100644 tests/python/frontend/oneflow/test_forward.py
create mode 100644 tests/python/relay/test_pass_flatten_atrous_conv.py
[tvm] 01/01: [CI] Update GPU image for oneflow v0.7
Posted by ma...@apache.org.
This is an automated email from the ASF dual-hosted git repository.
masahi pushed a commit to branch ci-gpu-update
in repository https://gitbox.apache.org/repos/asf/tvm.git
commit 64d2f9416ddc48b66c148c5987170b8cc2841e72
Author: Masahiro Masuda <ma...@gmail.com>
AuthorDate: Thu Apr 21 08:19:01 2022 +0900
[CI] Update GPU image for oneflow v0.7
---
Jenkinsfile | 4 ++--
jenkins/Jenkinsfile.j2 | 2 +-
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/Jenkinsfile b/Jenkinsfile
index 96ec10725b..c10a32409d 100755
--- a/Jenkinsfile
+++ b/Jenkinsfile
@@ -45,12 +45,12 @@
// 'python3 jenkins/generate.py'
// Note: This timestamp is here to ensure that updates to the Jenkinsfile are
// always rebased on main before merging:
-// Generated at 2022-04-19T10:04:53.134656
+// Generated at 2022-04-21T08:18:57.400427
import org.jenkinsci.plugins.pipeline.modeldefinition.Utils
// NOTE: these lines are scanned by docker/dev_common.sh. Please update the regex as needed. -->
ci_lint = 'tlcpack/ci-lint:v0.71'
-ci_gpu = 'tlcpack/ci-gpu:v0.85'
+ci_gpu = 'tlcpack/ci-gpu:v0.86'
ci_cpu = 'tlcpack/ci-cpu:v0.84'
ci_wasm = 'tlcpack/ci-wasm:v0.73'
ci_i386 = 'tlcpack/ci-i386:v0.77'
diff --git a/jenkins/Jenkinsfile.j2 b/jenkins/Jenkinsfile.j2
index 1dd9c318ef..17759ae84b 100644
--- a/jenkins/Jenkinsfile.j2
+++ b/jenkins/Jenkinsfile.j2
@@ -52,7 +52,7 @@ import org.jenkinsci.plugins.pipeline.modeldefinition.Utils
// NOTE: these lines are scanned by docker/dev_common.sh. Please update the regex as needed. -->
ci_lint = 'tlcpack/ci-lint:v0.71'
-ci_gpu = 'tlcpack/ci-gpu:v0.85'
+ci_gpu = 'tlcpack/ci-gpu:v0.86'
ci_cpu = 'tlcpack/ci-cpu:v0.84'
ci_wasm = 'tlcpack/ci-wasm:v0.73'
ci_i386 = 'tlcpack/ci-i386:v0.77'