Skip to content

Conversation

@snnn
Copy link

@snnn snnn commented Nov 5, 2022

This change is very similar to #906. It seems that nvcc compiler, though is it based on msvc, does not support this namespace.

Sample build:
https://dev.azure.com/onnxruntime/onnxruntime/_build/results?buildId=802911&view=logs&j=6df8fe70-7b8f-505a-8ef0-8bf93da2bac7&t=38e4b5a4-9041-5949-6670-6ace666c420b&l=4492

Error message:

D:\a_work\1\b\RelWithDebInfo_deps\gsl-src\include\gsl\util(101,0): Error #2803-D: attribute namespace "gsl" is unrecognized

It happens only when compiling *.cu files.

The command that generated the above warning:

"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6\bin\nvcc.exe" -gencode=arch=compute_75,code="compute_75,compute_75" -gencode=arch=compute_75,code="sm_75,compute_75" --use-local-env -ccbin "C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\VC\Tools\MSVC\14.29.30133\bin\HostX64\x64" -x cu -ID:\a\_work\1\s\include\onnxruntime -ID:\a\_work\1\s\include\onnxruntime\core\session -ID:\a\_work\1\b\RelWithDebInfo -ID:\a\_work\1\s\onnxruntime -I"D:\a\_work\1\b\RelWithDebInfo\_deps\abseil_cpp-src" -ID:\a\_work\1\s\cmake\external\SafeInt -ID:\a\_work\1\s\cmake\external\mp11\include -I"D:\a\_work\1\b\RelWithDebInfo\_deps\gsl-src\include" -ID:\a\_work\1\s\cmake\external\pytorch_cpuinfo\include -ID:\a\_work\1\s\cmake\external\onnx -ID:\a\_work\1\b\RelWithDebInfo\external\onnx -ID:\a\_work\1\s\cmake\external\protobuf\src -ID:\a\_work\1\s\cmake\external\flatbuffers\include -ID:\a\_work\1\s\cmake\external\eigen -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6\include" -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6\extras\CUPTI\include" -I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.6\include" --keep-dir x64\RelWithDebInfo -maxrregcount=0 --machine 64 --compile -cudart shared --expt-relaxed-constexpr --Werror default-stream-launch -Xcudafe --diag_suppress=bad_friend_decl -Xcudafe --diag_suppress=unsigned_compare_with_zero -Xcudafe --diag_suppress=expr_has_no_effect --threads 1 -std=c++17 -Werror all-warnings -Xcompiler="/EHsc -Zi -Ob1 /utf-8 /sdl /wd4251 /wd4201 /wd5054 /w15038 /guard:cf /Qspectre /wd4251 /wd4201 /wd5054 /w15038 /wd4834 /wd4127" -D_WINDOWS -DNDEBUG -DEIGEN_USE_THREADS -DPLATFORM_WINDOWS -DNOGDI -DNOMINMAX -D_USE_MATH_DEFINES -D_SILENCE_ALL_CXX17_DEPRECATION_WARNINGS -DUSE_CUDA=1 -DCPUINFO_SUPPORTED_PLATFORM=1 -DONNX_NAMESPACE=onnx -DONNX_ML=1 -DONNX_USE_LITE_PROTO=1 -D__ONNX_NO_DOC_STRINGS -DWIN32_LEAN_AND_MEAN -DEIGEN_MPL2_ONLY -DEIGEN_HAS_CONSTEXPR -DEIGEN_HAS_VARIADIC_TEMPLATES -DEIGEN_HAS_CXX11_MATH -DEIGEN_HAS_CXX11_ATOMIC -DEIGEN_STRONG_INLINE=inline -DORT_RUN_EXTERNAL_ONNX_TESTS -DENABLE_CUDA_PROFILING -DWINAPI_FAMILY=100 -DWINVER=0x0601 -D_WIN32_WINNT=0x0601 -DNTDDI_VERSION=0x06010000 -D"CMAKE_INTDIR="RelWithDebInfo"" -Donnxruntime_providers_cuda_EXPORTS -D_WINDLL -D_MBCS -DWIN32 -D_WINDOWS -DCPUINFO_SUPPORTED -DEIGEN_HAS_C99_MATH -DNDEBUG -DEIGEN_USE_THREADS -DPLATFORM_WINDOWS -DNOGDI -DNOMINMAX -D_USE_MATH_DEFINES -D_SILENCE_ALL_CXX17_DEPRECATION_WARNINGS -DUSE_CUDA=1 -DCPUINFO_SUPPORTED_PLATFORM=1 -DONNX_NAMESPACE=onnx -DONNX_ML=1 -DONNX_USE_LITE_PROTO=1 -D__ONNX_NO_DOC_STRINGS -DWIN32_LEAN_AND_MEAN -DEIGEN_MPL2_ONLY -DEIGEN_HAS_CONSTEXPR -DEIGEN_HAS_VARIADIC_TEMPLATES -DEIGEN_HAS_CXX11_MATH -DEIGEN_HAS_CXX11_ATOMIC -DEIGEN_STRONG_INLINE=inline -DORT_RUN_EXTERNAL_ONNX_TESTS -DENABLE_CUDA_PROFILING -DWINAPI_FAMILY=100 -DWINVER=0x0601 -D_WIN32_WINNT=0x0601 -DNTDDI_VERSION=0x06010000 -D"CMAKE_INTDIR="RelWithDebInfo"" -Donnxruntime_providers_cuda_EXPORTS -Xcompiler "/EHsc /W3 /nologo /O2 /Fdonnxruntime_providers_cuda.dir\RelWithDebInfo\vc142.pdb /FS /Zi /MD /GR" -o onnxruntime_providers_cuda.dir\RelWithDebInfo/D_/a/_work/1/s/onnxruntime/core/providers/cuda/activation/activations_impl.cu.obj "D:\a\_work\1\s\onnxruntime\core\providers\cuda\activation\activations_impl.cu"

@dmitrykobets-msft
Copy link
Member

Hi @snnn, looks good, thanks for the fix!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants