[CI] Add CUDA 13.0 inductor CI benchmarks#165029
[CI] Add CUDA 13.0 inductor CI benchmarks#165029tinglvv wants to merge 31 commits intopytorch:mainfrom
Conversation
bdhirsh
added
the
triaged
|
@tinglvv You can try this for cuSPARSE deprecation warnings. I have not tested it so might not work but something like this should resolve the issue. diff --git a/cmake/Modules/FindCUDAToolkit.cmake b/cmake/Modules/FindCUDAToolkit.cmake
index ec9ae530aa..0e0460451a 100644
--- a/cmake/Modules/FindCUDAToolkit.cmake
+++ b/cmake/Modules/FindCUDAToolkit.cmake
@@ -905,6 +905,7 @@ if(CUDAToolkit_FOUND)
endif()
endif()
set_property(TARGET CUDA::${lib_name} PROPERTY IMPORTED_LOCATION "${CUDA_${lib_name}_LIBRARY}")
+ set_property(TARGET CUDA::${lib_name} PROPERTY SYSTEM TRUE)
foreach(dep ${arg_DEPS})
if(TARGET CUDA::${dep})
set_property(TARGET CUDA::${lib_name} APPEND PROPERTY |
|
From cusparse doc, https://docs.nvidia.com/cuda/pdf/CUSPARSE_Library.pdf, suggested passing |
|
Previous flag worked in resolving the deprecation warning. New build error |
|
Disabled fbgemm for CUDA 13 inductor build due to #165029 (comment). CUDA 13 support for FBGEMM is WIP pytorch/FBGEMM#4783 |
tinglvv
changed the title
|
Further debug shows that the job calls into install_torchrec_and_fbgemm function from Line 1771 in d049ed2 |
|
Should not disable torchrec since it will lead to failure in: previous fbgemm failure is gone now. |
|
@pytorchbot rebase |
|
@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here |
|
Successfully rebased |
pytorchmergebot
force-pushed
the
cu13-ci-test
branch
from
13421de to
10763ab
Compare
|
@pytorchbot rebase |
|
@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here |
|
@pytorchbot rebase -b main |
|
@pytorchbot started a rebase job onto refs/remotes/origin/main. Check the current status here |
This reverts commit bf7ca1e.
…ther than driver issue
|
Successfully rebased |
pytorchmergebot
force-pushed
the
cu13-ci-test
branch
from
feaa12b to
82d2791
Compare
|
@pytorchbot merge -i "failures are not related" |
|
❌ 🤖 pytorchbot command failed: Try |
|
@pytorchbot merge -i |
Adding CUDA 13.0 to the inductor bencharks as it is the latest support CUDA version Pull Request resolved: pytorch#165029 Approved by: https://github.com/atalman
Adding CUDA 13.0 to the inductor bencharks as it is the latest support CUDA version Pull Request resolved: #165029 Approved by: https://github.com/atalman
7 participants
Adding CUDA 13.0 to the inductor bencharks as it is the latest support CUDA version