Fixing Windows MSVC Compatibility #10

yuanluxu · 2020-01-24T19:45:04Z

🐛 Bugs / Unexpected behaviors

Fixed Windows MSVC build compatibility issues. All tests passed.

If you are using pre-compiled pytorch 1.4 and torchvision 0.5. You should make the following revisions to the pytorch source codes also to successfully compile with visual studio 2019 (MSVC 19.16.27034) and CUDA 10.1.

Change python/Lib/site-packages/torch/include/csrc/jit/script/module.h

L466, 476, 493, 506, 536
-static constexpr *
+static const *

Change python/Lib/site-packages/torch/include/csrc/jit/argument_spec.h

L190
-static constexpr size_t DEPTH_LIMIT = 128;
+static const size_t DEPTH_LIMIT = 128;

Change python/Lib/site-packages/torch/include/pybind11/cast.h

L1449
-explicit operator type&() { return *(this->value); }
+explicit operator type& () { return *((type*)(this->value)); }

If you are using CUDA 10.2 with pre-compiled pytorch 1.4 and torchvision 0.5, you need to check
Pytorch Issue [(https://github.com/pytorch/pytorch/issues/33203)] for patching the pytorch source code in addition.

Instructions To Reproduce the Issue:

Please include the following (depending on what the issue is):

Any changes you made (git diff) or code you wrote

https://github.com/facebookresearch/pytorch3d/pull/9

The exact command(s) you ran:
What you observed (including the full logs):

<put logs here>

Please also simplify the steps as much as possible so they do not require additional resources to
run, such as a private dataset.

The text was updated successfully, but these errors were encountered:

srxdev0619 · 2020-02-10T09:02:38Z

Hi,

I get the following errors while trying to install Pytorch3d

c:/users/shahrukh/anaconda3/lib/site-packages/torch/include\c10/util/BFloat16.h(48): warning: calling a __host__ function from a __host__ __device__ function is not allowed

C:\Users\ShahRukh\Research\pytorch3d\pytorch3d\csrc\rasterize_meshes\geometry_utils.cuh(231): error: identifier "kEpsilon" is undefined in device code

C:/Users/ShahRukh/Research/pytorch3d/pytorch3d/csrc/rasterize_meshes/rasterize_meshes.cu(391): error: identifier "kEpsilon" is undefined in device code

C:/Users/ShahRukh/Research/pytorch3d/pytorch3d/csrc/rasterize_meshes/rasterize_meshes.cu(427): error: identifier "kEpsilon" is undefined in device code

C:/Users/ShahRukh/Research/pytorch3d/pytorch3d/csrc/rasterize_meshes/rasterize_meshes.cu(427): error: calling a __host__ function("pow<float, double, void> ") from a __global__ function("RasterizeMeshesBackwardCudaKernel") is not allowed

C:/Users/ShahRukh/Research/pytorch3d/pytorch3d/csrc/rasterize_meshes/rasterize_meshes.cu(427): error: identifier "pow<float, double, void> " is undefined in device code

C:/Users/ShahRukh/Research/pytorch3d/pytorch3d/csrc/rasterize_meshes/rasterize_meshes.cu(133): error: identifier "kEpsilon" is undefined in device code

C:/Users/ShahRukh/Research/pytorch3d/pytorch3d/csrc/rasterize_meshes/rasterize_meshes.cu(133): error: identifier "kEpsilon" is undefined in device code

C:/Users/ShahRukh/Research/pytorch3d/pytorch3d/csrc/rasterize_meshes/rasterize_meshes.cu(141): error: identifier "kEpsilon" is undefined in device code
8 errors detected in the compilation of "C:/Users/ShahRukh/AppData/Local/Temp/tmpxft_000026f0_00000000-10_rasterize_meshes.cpp1.ii".
    rasterize_meshes.cu
    error: command 'C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.1\\bin\\nvcc.exe' failed with exit status 1
    ----------------------------------------
ERROR: Command errored out with exit status 1: 'c:\users\shahrukh\anaconda3\python.exe' -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'C:\\Users\\ShahRukh\\Research\\pytorch3d\\setup.py'"'"'; __file__='"'"'C:\\Users\\ShahRukh\\Research\\pytorch3d\\setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' develop --no-deps Check the logs for full command output.

The installation commands I ran were the following, in the order of execution

conda install -c pytorch pytorch torchvision cudatoolkit=10.1
conda install -c conda-forge -c fvcore fvcore

pip install scikit-image matplotlib imageio

git clone https://github.com/facebookresearch/pytorch3d.git
cd pytorch3d
pip install -e .

I'm using Visual Studio 2019 Community. It'd be great if these errors could be fixed. Thank you!

yuanluxu · 2020-02-10T18:17:32Z

Hi @srxdev0619, I believe I fixed this issue in my fork. You can clone my folk and re-compile it.
https://github.com/yuanluxu/pytorch3d

srxdev0619 · 2020-02-11T07:09:58Z

Yes! The fix works perfectly, thank you so much!

ZM-J · 2020-12-30T08:54:40Z

Hi! I followed your instructions, but still failed to compile mmcv-full.

When I set types as constexpr in Lib\site-packages\torch\include\torch\csrc\jit\api, it shows:

  [28/34] C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin\nvcc --use-local-env -Xcompiler /MD -Xcompiler /wd4819 -Xcompiler /wd4251 -Xcompiler /wd4244 -Xcomp
iler /wd4267 -Xcompiler /wd4275 -Xcompiler /wd4018 -Xcompiler /wd4190 -Xcompiler /EHsc -Xcudafe --diag_suppress=base_class_has_different_dll_interface -Xcudafe --diag_sup
press=field_without_dll_interface -Xcudafe --diag_suppress=dll_interface_conflict_none_assumed -Xcudafe --diag_suppress=dll_interface_conflict_dllexport_assumed -DMMCV_WI
TH_CUDA -IC:\Users\ADMINI~1\AppData\Local\Temp\pip-install-wzd40w8w\mmcv-full\mmcv\ops\csrc -Id:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include -Id:\sof
tware\anaconda3\envs\ipercore\lib\site-packages\torch\include\torch\csrc\api\include -Id:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include\TH -Id:\softwar
e\anaconda3\envs\ipercore\lib\site-packages\torch\include\THC "-IC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\include" -Id:\software\anaconda3\envs\ipercore\i
nclude -Id:\software\anaconda3\envs\ipercore\include "-ID:\Software\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.27.29110\ATLMFC\include" "-ID:\Software\Micros
oft Visual Studio\2019\Community\VC\Tools\MSVC\14.27.29110\include" "-IC:\Program Files (x86)\Windows Kits\NETFXSDK\4.8\include\um" "-ID:\Windows Kits\10\include\10.0.190
41.0\ucrt" "-ID:\Windows Kits\10\include\10.0.19041.0\shared" "-ID:\Windows Kits\10\include\10.0.19041.0\um" "-ID:\Windows Kits\10\include\10.0.19041.0\winrt" "-ID:\Windo
ws Kits\10\include\10.0.19041.0\cppwinrt" -c C:\Users\ADMINI~1\AppData\Local\Temp\pip-install-wzd40w8w\mmcv-full\mmcv\ops\csrc\pytorch\psamask_cuda.cu -o C:\Users\ADMINI~
1\AppData\Local\Temp\pip-install-wzd40w8w\mmcv-full\build\temp.win-amd64-3.6\Release\mmcv/ops/csrc/pytorch\psamask_cuda.obj -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_
CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_ext -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=
compute_75,code=sm_75
  FAILED: C:/Users/ADMINI~1/AppData/Local/Temp/pip-install-wzd40w8w/mmcv-full/build/temp.win-amd64-3.6/Release/mmcv/ops/csrc/pytorch/psamask_cuda.obj
  C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin\nvcc --use-local-env -Xcompiler /MD -Xcompiler /wd4819 -Xcompiler /wd4251 -Xcompiler /wd4244 -Xcompiler /wd
4267 -Xcompiler /wd4275 -Xcompiler /wd4018 -Xcompiler /wd4190 -Xcompiler /EHsc -Xcudafe --diag_suppress=base_class_has_different_dll_interface -Xcudafe --diag_suppress=fi
eld_without_dll_interface -Xcudafe --diag_suppress=dll_interface_conflict_none_assumed -Xcudafe --diag_suppress=dll_interface_conflict_dllexport_assumed -DMMCV_WITH_CUDA 
-IC:\Users\ADMINI~1\AppData\Local\Temp\pip-install-wzd40w8w\mmcv-full\mmcv\ops\csrc -Id:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include -Id:\software\an
aconda3\envs\ipercore\lib\site-packages\torch\include\torch\csrc\api\include -Id:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include\TH -Id:\software\anacon
da3\envs\ipercore\lib\site-packages\torch\include\THC "-IC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\include" -Id:\software\anaconda3\envs\ipercore\include -
Id:\software\anaconda3\envs\ipercore\include "-ID:\Software\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.27.29110\ATLMFC\include" "-ID:\Software\Microsoft Visu
al Studio\2019\Community\VC\Tools\MSVC\14.27.29110\include" "-IC:\Program Files (x86)\Windows Kits\NETFXSDK\4.8\include\um" "-ID:\Windows Kits\10\include\10.0.19041.0\ucr
t" "-ID:\Windows Kits\10\include\10.0.19041.0\shared" "-ID:\Windows Kits\10\include\10.0.19041.0\um" "-ID:\Windows Kits\10\include\10.0.19041.0\winrt" "-ID:\Windows Kits\
10\include\10.0.19041.0\cppwinrt" -c C:\Users\ADMINI~1\AppData\Local\Temp\pip-install-wzd40w8w\mmcv-full\mmcv\ops\csrc\pytorch\psamask_cuda.cu -o C:\Users\ADMINI~1\AppDat
a\Local\Temp\pip-install-wzd40w8w\mmcv-full\build\temp.win-amd64-3.6\Release\mmcv/ops/csrc/pytorch\psamask_cuda.obj -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSI
ONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_ext -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_
75,code=sm_75
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\c10/util/ThreadLocalDebugInfo.h(12): warning: modifier is ignored on an enum specifier
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\ATen/core/boxing/impl/boxing.h(100): warning: integer conversion resulted in a change of sign
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\ATen/record_function.h(13): warning: modifier is ignored on an enum specifier
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\ATen/core/op_registration/op_whitelist.h(39): warning: integer conversion resulted in a change of si
gn
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\torch/csrc/jit/ir/ir.h(1347): error: member "torch::jit::ProfileOptionalOp::Kind" may not be initial
ized
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\torch/csrc/jit/api/module.h(485): error: member "torch::jit::detail::ModulePolicy::all_slots" may no
t be initialized
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\torch/csrc/jit/api/module.h(498): error: member "torch::jit::detail::ParameterPolicy::all_slots" may
 not be initialized
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\torch/csrc/jit/api/module.h(512): error: member "torch::jit::detail::BufferPolicy::all_slots" may no
t be initialized
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\torch/csrc/jit/api/module.h(525): error: member "torch::jit::detail::AttributePolicy::all_slots" may
 not be initialized
  
  5 errors detected in the compilation of "C:/Users/ADMINI~1/AppData/Local/Temp/tmpxft_00004eac_00000000-7_psamask_cuda.cpp1.ii".
  psamask_cuda.cu

and when I set them as const, it shows:

  [31/34] C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin\nvcc --use-local-env -Xcompiler /MD -Xcompiler /wd4819 -Xcompiler /wd4251 -Xcompiler /wd4244 -Xcomp
iler /wd4267 -Xcompiler /wd4275 -Xcompiler /wd4018 -Xcompiler /wd4190 -Xcompiler /EHsc -Xcudafe --diag_suppress=base_class_has_different_dll_interface -Xcudafe --diag_sup
press=field_without_dll_interface -Xcudafe --diag_suppress=dll_interface_conflict_none_assumed -Xcudafe --diag_suppress=dll_interface_conflict_dllexport_assumed -DMMCV_WI
TH_CUDA -IC:\Users\ADMINI~1\AppData\Local\Temp\pip-install-5w_ltf2i\mmcv-full\mmcv\ops\csrc -Id:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include -Id:\sof
tware\anaconda3\envs\ipercore\lib\site-packages\torch\include\torch\csrc\api\include -Id:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include\TH -Id:\softwar
e\anaconda3\envs\ipercore\lib\site-packages\torch\include\THC "-IC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\include" -Id:\software\anaconda3\envs\ipercore\i
nclude -Id:\software\anaconda3\envs\ipercore\include "-ID:\Software\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.27.29110\ATLMFC\include" "-ID:\Software\Micros
oft Visual Studio\2019\Community\VC\Tools\MSVC\14.27.29110\include" "-IC:\Program Files (x86)\Windows Kits\NETFXSDK\4.8\include\um" "-ID:\Windows Kits\10\include\10.0.190
41.0\ucrt" "-ID:\Windows Kits\10\include\10.0.19041.0\shared" "-ID:\Windows Kits\10\include\10.0.19041.0\um" "-ID:\Windows Kits\10\include\10.0.19041.0\winrt" "-ID:\Windo
ws Kits\10\include\10.0.19041.0\cppwinrt" -c C:\Users\ADMINI~1\AppData\Local\Temp\pip-install-5w_ltf2i\mmcv-full\mmcv\ops\csrc\pytorch\psamask_cuda.cu -o C:\Users\ADMINI~
1\AppData\Local\Temp\pip-install-5w_ltf2i\mmcv-full\build\temp.win-amd64-3.6\Release\mmcv/ops/csrc/pytorch\psamask_cuda.obj -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_
CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_ext -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=
compute_75,code=sm_75
  FAILED: C:/Users/ADMINI~1/AppData/Local/Temp/pip-install-5w_ltf2i/mmcv-full/build/temp.win-amd64-3.6/Release/mmcv/ops/csrc/pytorch/psamask_cuda.obj
  C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin\nvcc --use-local-env -Xcompiler /MD -Xcompiler /wd4819 -Xcompiler /wd4251 -Xcompiler /wd4244 -Xcompiler /wd
4267 -Xcompiler /wd4275 -Xcompiler /wd4018 -Xcompiler /wd4190 -Xcompiler /EHsc -Xcudafe --diag_suppress=base_class_has_different_dll_interface -Xcudafe --diag_suppress=fi
eld_without_dll_interface -Xcudafe --diag_suppress=dll_interface_conflict_none_assumed -Xcudafe --diag_suppress=dll_interface_conflict_dllexport_assumed -DMMCV_WITH_CUDA 
-IC:\Users\ADMINI~1\AppData\Local\Temp\pip-install-5w_ltf2i\mmcv-full\mmcv\ops\csrc -Id:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include -Id:\software\an
aconda3\envs\ipercore\lib\site-packages\torch\include\torch\csrc\api\include -Id:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include\TH -Id:\software\anacon
da3\envs\ipercore\lib\site-packages\torch\include\THC "-IC:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\include" -Id:\software\anaconda3\envs\ipercore\include -
Id:\software\anaconda3\envs\ipercore\include "-ID:\Software\Microsoft Visual Studio\2019\Community\VC\Tools\MSVC\14.27.29110\ATLMFC\include" "-ID:\Software\Microsoft Visu
al Studio\2019\Community\VC\Tools\MSVC\14.27.29110\include" "-IC:\Program Files (x86)\Windows Kits\NETFXSDK\4.8\include\um" "-ID:\Windows Kits\10\include\10.0.19041.0\ucr
t" "-ID:\Windows Kits\10\include\10.0.19041.0\shared" "-ID:\Windows Kits\10\include\10.0.19041.0\um" "-ID:\Windows Kits\10\include\10.0.19041.0\winrt" "-ID:\Windows Kits\
10\include\10.0.19041.0\cppwinrt" -c C:\Users\ADMINI~1\AppData\Local\Temp\pip-install-5w_ltf2i\mmcv-full\mmcv\ops\csrc\pytorch\psamask_cuda.cu -o C:\Users\ADMINI~1\AppDat
a\Local\Temp\pip-install-5w_ltf2i\mmcv-full\build\temp.win-amd64-3.6\Release\mmcv/ops/csrc/pytorch\psamask_cuda.obj -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSI
ONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr -DTORCH_API_INCLUDE_EXTENSION_H -DTORCH_EXTENSION_NAME=_ext -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_
75,code=sm_75
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\c10/util/ThreadLocalDebugInfo.h(12): warning: modifier is ignored on an enum specifier
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\ATen/core/boxing/impl/boxing.h(100): warning: integer conversion resulted in a change of sign
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\ATen/record_function.h(13): warning: modifier is ignored on an enum specifier
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\ATen/core/op_registration/op_whitelist.h(39): warning: integer conversion resulted in a change of si
gn
  
  d:/software/anaconda3/envs/ipercore/lib/site-packages/torch/include\torch/csrc/jit/ir/ir.h(1347): error: member "torch::jit::ProfileOptionalOp::Kind" may not be initial
ized

So, I think that we may replace constexpr * in torch/include\torch/csrc/jit/ir/ir.h to const * as well. However, when I did this, it shows:

  d:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include\torch/csrc/jit/ir/ir.h(1347): error C2864: torch::jit::ProfileOptionalOp::Kind: Static data members with in-class initialization expressions must have an immutable constant integral type, or must be designated as "inline"
  d:\software\anaconda3\envs\ipercore\lib\site-packages\torch\include\torch/csrc/jit/ir/ir.h(1347): note: The type is “const c10::Symbol”

Could you help me?

BTW, I use Windows 10 1909, with CUDA 10.1. I use Visual Studio Community 2019 with MSVC v142 as compilers and libraries. My python version is 3.6.6 and my pytorch version is 1.7.1.

ZM-J · 2020-12-30T11:51:48Z

I think this answer has solved my problem: Just comment out the line.

nikhilaravi mentioned this issue Feb 10, 2020

Available for Windows ? #40

Closed

yuanluxu mentioned this issue Feb 13, 2020

cuda nvcc compile issue with pytorch 1.4.0 on windows #57

Closed

nikhilaravi assigned bottler Feb 14, 2020

yuanluxu changed the title ~~Windows MSVC Compatibility~~ Fixing Windows MSVC Compatibility Feb 18, 2020

yuanluxu closed this as completed Feb 21, 2020

rohitdavas mentioned this issue Jul 4, 2020

Not compiled with gpu support error while installing on system. #257

Closed

ZwwWayne mentioned this issue Sep 24, 2020

Can't install mmcv-full. Reported 'Ninja: build stopped: subcommand failed' open-mmlab/mmcv#575

Closed

efirdc mentioned this issue Nov 11, 2020

Error when compiling fused_bias_act_kernel.cu on pytorch 1.7/cuda 11 rosinality/stylegan2-pytorch#149

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fixing Windows MSVC Compatibility #10

Fixing Windows MSVC Compatibility #10

yuanluxu commented Jan 24, 2020 •

edited

Loading

srxdev0619 commented Feb 10, 2020

yuanluxu commented Feb 10, 2020

srxdev0619 commented Feb 11, 2020

ZM-J commented Dec 30, 2020 •

edited

Loading

ZM-J commented Dec 30, 2020

Fixing Windows MSVC Compatibility #10

Fixing Windows MSVC Compatibility #10

Comments

yuanluxu commented Jan 24, 2020 • edited Loading

🐛 Bugs / Unexpected behaviors

Instructions To Reproduce the Issue:

srxdev0619 commented Feb 10, 2020

yuanluxu commented Feb 10, 2020

srxdev0619 commented Feb 11, 2020

ZM-J commented Dec 30, 2020 • edited Loading

ZM-J commented Dec 30, 2020

yuanluxu commented Jan 24, 2020 •

edited

Loading

ZM-J commented Dec 30, 2020 •

edited

Loading