Hi Forum!
I decided to build pytorch on GTX660 graphics card (don’t ask why, I don’t know).
I got the idea from the article.
What I used:
NVIDIA Driver
474.30-desktop-win10-win11-64bit-international-dch-whql
cuda 10.1
cudnn v8.0.5.39
BuildTools 14.29.30133
magma_2.5.4_cuda101_release
mkl_2020.2.254
I reinstalled the system, put CUDA 10.1, put all the other components, but during the build process an error occurs, first it was received for pytorch 1.8.1, then for 1.8.0
In the process, I had to install the latest version of the pybind11 library
[4499/4612] Building CXX object caffe2\torch\CMakeFiles\torch_python.dir\csrc\Stream.cpp.obj
FAILED: caffe2/torch/CMakeFiles/torch_python.dir/csrc/Stream.cpp.obj
C:\Users\PromiX\scoop\shims\sccache.exe C:\PROGRA~2\MICROS~3\2019\BUILDT~1\VC\Tools\MSVC\1429~1.301\bin\Hostx64\x64\cl.exe /nologo /TP -DBUILDING_TESTS -DFMT_HEADER_ONLY=1 -DIDEEP_USE_MKL -DMAGMA_V2 -DMINIZ_DISABLE_ZIP_READER_CRC32_CHECKS -DONNXIFI_ENABLE_EXT=1 -DONNX_ML=1 -DONNX_NAMESPACE=onnx_torch -DTHP_BUILD_MAIN_LIB -DTH_BLAS_MKL -DUSE_C10D -DUSE_C10D_GLOO -DUSE_CUDA -DUSE_CUDNN -DUSE_DISTRIBUTED -DUSE_EXTERNAL_MZCRC -DUSE_NUMPY -DWIN32_LEAN_AND_MEAN -D_CRT_SECURE_NO_DEPRECATE=1 -D_OPENMP_NOFORCE_MANIFEST -Dtorch_python_EXPORTS -ID:\pytorch\build\aten\src -ID:\pytorch\aten\src -ID:\pytorch\build -ID:\pytorch -ID:\pytorch\cmake\..\third_party\benchmark\include -ID:\pytorch\build\caffe2\contrib\aten -ID:\pytorch\third_party\onnx -ID:\pytorch\build\third_party\onnx -ID:\pytorch\third_party\foxi -ID:\pytorch\build\third_party\foxi -ID:\pytorch\torch\.. -ID:\pytorch\torch\..\aten\src -ID:\pytorch\torch\..\aten\src\TH -ID:\pytorch\build\caffe2\aten\src -ID:\pytorch\build\third_party -ID:\pytorch\torch\..\third_party\valgrind-headers -ID:\pytorch\torch\..\third_party\gloo -ID:\pytorch\torch\..\third_party\onnx -ID:\pytorch\torch\csrc -ID:\pytorch\torch\csrc\api\include -ID:\pytorch\torch\lib -ID:\pytorch\torch\lib\libshm_windows -ID:\pytorch\torch\csrc\api -ID:\pytorch\c10\.. -ID:\pytorch\build\third_party\ideep\mkl-dnn\include -ID:\pytorch\third_party\ideep\mkl-dnn\src\..\include -ID:\pytorch\c10\cuda\..\.. -ID:\pytorch\third_party\fmt\include -ID:\pytorch\torch\lib\c10d\.. -external:ID:\pytorch\build\third_party\gloo -external:ID:\pytorch\cmake\..\third_party\gloo -external:ID:\pytorch\cmake\..\third_party\googletest\googlemock\include -external:ID:\pytorch\cmake\..\third_party\googletest\googletest\include -external:ID:\pytorch\third_party\protobuf\src -external:I"D:\Pytorch requirements\mkl\include" -external:ID:\pytorch\third_party\XNNPACK\include -external:ID:\pytorch\third_party -external:ID:\pytorch\cmake\..\third_party\eigen -external:ID:\condaEnv\include -external:ID:\condaEnv\Lib\site-packages\numpy\core\include -external:ID:\pytorch\cmake\..\third_party\pybind11\include -external:ID:\pytorch\cmake\..\third_party\cub -external:I"D:\Pytorch requirements\magma\include" -external:I"C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\include" -external:ID:\pytorch\third_party\ideep\mkl-dnn\include -external:ID:\pytorch\third_party\ideep\include -external:I"C:\Program Files\NVIDIA Corporation\NvToolsExt\include" -external:W0 /DWIN32 /D_WINDOWS /GR /EHsc /w /bigobj -DUSE_PTHREADPOOL -openmp:experimental -DNDEBUG -DUSE_FBGEMM -DUSE_XNNPACK -DHAVE_AVX_CPU_DEFINITION -DHAVE_AVX2_CPU_DEFINITION /MD /O2 /Ob2 /DNDEBUG /w /bigobj -DNDEBUG -DCAFFE2_USE_GLOO -DCUDA_HAS_FP16=1 -DUSE_GCC_GET_CPUID -DUSE_AVX -DUSE_AVX2 -DTH_HAVE_THREAD /EHsc /DNOMINMAX /wd4267 /wd4251 /wd4522 /wd4838 /wd4305 /wd4244 /wd4190 /wd4101 /wd4996 /wd4275 /bigobj -std:c++14 /showIncludes /Focaffe2\torch\CMakeFiles\torch_python.dir\csrc\Stream.cpp.obj /Fdcaffe2\torch\CMakeFiles\torch_python.dir\ /FS -c D:\pytorch\torch\csrc\Stream.cpp
D:\pytorch\torch\csrc\Stream.cpp(108): error C2106: =: left operand must be a left-handed value
ninja: build stopped: subcommand failed.
Traceback (most recent call last):
File "D:\pytorch\setup.py", line 818, in <module>
build_deps()
File "D:\pytorch\setup.py", line 315, in build_deps
build_caffe2(version=version,
File "D:\pytorch\tools\build_pytorch_libs.py", line 58, in build_caffe2
cmake.build(my_env)
File "D:\pytorch\tools\setup_helpers\cmake.py", line 345, in build
self.run(build_args, my_env)
File "D:\pytorch\tools\setup_helpers\cmake.py", line 140, in run
check_call(command, cwd=self.build_dir, env=env)
File "D:\condaEnv\Lib\subprocess.py", line 413, in check_call
raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['cmake', '--build', '.', '--target', 'install', '--config', 'Release', '--', '-j', '1']' returned non-zero exit status 1.
FULL log