how can I install this pack on conda environment??

[Lurnco]

how can I install this pack on conda environment?? It jumped ERROR: Microsoft Visual C++ 14.0 or greater is required. but I had tied: " conda install libpython m2w64-toolchain -c msys2 " and isntalled "Microsoft Visual C++ Build Tools" as some blogs answer to solve it. what should I do???

[ghostplant]

Is your environment Windows OS?

[ghostplant]

On Linux, you can install Tutel from source using conda's python ./setup.py command. On Windows, current Tutel isn't support that. We'll soon do an investigation.

[ghostplant]

@Lurnco We merged a PR that work for Windows environment. Can you let us know whether it works for you?

[Lurnco]

Yeah,my system is win10.if I insatll it in Windows ,its default path is Windows system python without ERROR I want install tutel it into my Anaconda environment(3.9) ,but it will jump ERROR. So,I can't install tutel on my Annconda environmet right now?

@Lurnco We merged a PR that work for Windows environment. Can you let us know whether it works for you?

[ghostplant]

Have you try using python3.exe -m pip install --user --upgrade git+https://github.com/microsoft/tutel@main? It should work well on Windows. If there is ERROR, can you share the detail of log file?

[Lurnco]

using: ### python3.exe -m pip install --user --upgrade git+https://github.com/microsoft/tutel@main Pop-up error： fatal: unable to access 'https://github.com/microsoft/tutel/': OpenSSL SSL_connect: SSL_ERROR_SYSCALL in connection to github.com:443

Another way executing: ### python3 setup.py install error：OSError: CUDA_HOME environment variable is not set. Please set it to your CUDA install root.

Is CUDA environment necessary?I haven't CUDA environment,which maybe causes of the problem.

[ghostplant]

@Lurnco CUDA environment is not needed, but it will only install CPU version. The error you show in the first case is not related to CUDA. It is likely your network has firewall that blocks you from accessing github over HTTPS.

For the second way, I think the error is not from Tutel, but from an old Pytorch version which only allows CUDA_HOME to be set when installing extensions. Can you try setting a CUDA_HOME variable with dummy values (e.g. C:) to avoid that?

[2385708835]

I also haven't CUDA environment set a CUDA_HOME:torch/utils/cpp_extention.py _find_cuda_home() cuda_home="C:" error: RuntimeError: Error compiling objects for extension

[ghostplant]

I also haven't CUDA environment set a CUDA_HOME:torch/utils/cpp_extention.py _find_cuda_home() cuda_home="C:" error: RuntimeError: Error compiling objects for extension

If you don't have CUDA environment, can you install Tutel successfully and run application using --device cpu?

[2385708835]

i have GPU environment now ,i try using python3.exe -m pip install --user --upgrade git+https://github.com/microsoft/tutel@main on Windows,there is complete log.

E:\Swin3Project\Swin3) E:\Swin3Project\Swin3>python.exe -m pip install --user --upgrade git+https://github.com/microsoft/tutel@main Collecting git+https://github.com/microsoft/tutel@main Cloning https://github.com/microsoft/tutel (to revision main) to c:\users\acer\appdata\local\temp\pip-req-build-5dyo0_c8 Running command git clone --filter=blob:none --quiet https://github.com/microsoft/tutel 'C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8' Resolved https://github.com/microsoft/tutel to commit 17f4aab9b69cf50dcddd2b985907126379af1568 Preparing metadata (setup.py) ... done Building wheels for collected packages: tutel Building wheel for tutel (setup.py) ... error error: subprocess-exited-with-error

× python setup.py bdist_wheel did not run successfully. │ exit code: 1 ╰─> [98 lines of output] running bdist_wheel running build running build_py creating build creating build\lib.win-amd64-cpython-39 creating build\lib.win-amd64-cpython-39\tutel copying tutel\jit.py -> build\lib.win-amd64-cpython-39\tutel copying tutel\moe.py -> build\lib.win-amd64-cpython-39\tutel copying tutel\net.py -> build\lib.win-amd64-cpython-39\tutel copying tutel\system.py -> build\lib.win-amd64-cpython-39\tutel copying tutel__init.py -> build\lib.win-amd64-cpython-39\tutel creating build\lib.win-amd64-cpython-39\tutel\checkpoint copying tutel\checkpoint\gather.py -> build\lib.win-amd64-cpython-39\tutel\checkpoint copying tutel\checkpoint\scatter.py -> build\lib.win-amd64-cpython-39\tutel\checkpoint copying tutel\checkpoint__init.py -> build\lib.win-amd64-cpython-39\tutel\checkpoint creating build\lib.win-amd64-cpython-39\tutel\custom copying tutel\custom__init.py -> build\lib.win-amd64-cpython-39\tutel\custom creating build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_amp.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_ddp.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_ddp_tutel.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_deepspeed.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_from_scratch.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\moe_cifar10.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\moe_mnist.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples__init.py -> build\lib.win-amd64-cpython-39\tutel\examples creating build\lib.win-amd64-cpython-39\tutel\experts copying tutel\experts\ffn.py -> build\lib.win-amd64-cpython-39\tutel\experts copying tutel\experts\init__.py -> build\lib.win-amd64-cpython-39\tutel\experts creating build\lib.win-amd64-cpython-39\tutel\gates copying tutel\gates\cosine_top.py -> build\lib.win-amd64-cpython-39\tutel\gates copying tutel\gates\top.py -> build\lib.win-amd64-cpython-39\tutel\gates copying tutel\gates\init__.py -> build\lib.win-amd64-cpython-39\tutel\gates creating build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\communicate.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\fast_dispatch.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\jit_compiler.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\losses.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\moe_layer.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\overlap.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\init.py -> build\lib.win-amd64-cpython-39\tutel\impls creating build\lib.win-amd64-cpython-39\tutel\jit_kernels copying tutel\jit_kernels\gating.py -> build\lib.win-amd64-cpython-39\tutel\jit_kernels copying tutel\jit_kernels\sparse.py -> build\lib.win-amd64-cpython-39\tutel\jit_kernels copying tutel\jit_kernels\init__.py -> build\lib.win-amd64-cpython-39\tutel\jit_kernels creating build\lib.win-amd64-cpython-39\tutel\launcher copying tutel\launcher\execl.py -> build\lib.win-amd64-cpython-39\tutel\launcher copying tutel\launcher\run.py -> build\lib.win-amd64-cpython-39\tutel\launcher copying tutel\launcher\init.py -> build\lib.win-amd64-cpython-39\tutel\launcher creating build\lib.win-amd64-cpython-39\tutel\parted copying tutel\parted\patterns.py -> build\lib.win-amd64-cpython-39\tutel\parted copying tutel\parted\solver.py -> build\lib.win-amd64-cpython-39\tutel\parted copying tutel\parted\spmdx.py -> build\lib.win-amd64-cpython-39\tutel\parted copying tutel\parted\init.py -> build\lib.win-amd64-cpython-39\tutel\parted creating build\lib.win-amd64-cpython-39\tutel\parted\backend copying tutel\parted\backend\init.py -> build\lib.win-amd64-cpython-39\tutel\parted\backend creating build\lib.win-amd64-cpython-39\tutel\parted\backend\torch copying tutel\parted\backend\torch\config.py -> build\lib.win-amd64-cpython-39\tutel\parted\backend\torch copying tutel\parted\backend\torch\executor.py -> build\lib.win-amd64-cpython-39\tutel\parted\backend\torch copying tutel\parted\backend\torch\init__.py -> build\lib.win-amd64-cpython-39\tutel\parted\backend\torch running build_ext building 'tutel_custom_kernel' extension creating C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39 creating C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release creating C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\tutel creating C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\tutel\custom Emitting ninja build file C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\build.ninja... Compiling objects... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) 1.10.2 "d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\BIN\x86_amd64\link.exe" /nologo /INCREMENTAL:NO /LTCG /DLL /MANIFEST:EMBED,ID=2 /MANIFESTUAC:NO /LIBPATH:/usr/local/cuda/lib64/stubs /LIBPATH:E:\Swin3Project\Swin3\lib\site-packages\torch\lib "/LIBPATH:C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA\v11.7\lib/x64" /LIBPATH:E:\Swin3Project\Swin3\libs /LIBPATH:E:\Swin3Project\Swin3 /LIBPATH:E:\Swin3Project\Swin3\PCbuild\amd64 "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\LIB\amd64" "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\ATLMFC\LIB\amd64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\ucrt\x64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\um\x64" cuda.lib nvrtc.lib nccl.lib c10.lib torch.lib torch_cpu.lib torch_python.lib cudart.lib c10_cuda.lib torch_cuda_cu.lib torch_cuda_cpp.lib /EXPORT:PyInit_tutel_custom_kernel C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj /OUT:build\lib.win-amd64-cpython-39\tutel_custom_kernel.cp39-win_amd64.pyd /IMPLIB:C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom\tutel_custom_kernel.cp39-win_amd64.lib LINK : fatal error LNK1181: 无法打开输入文件“nccl.lib” Try installing without NCCL extension.. running bdist_wheel running build running build_py running build_ext building 'tutel_custom_kernel' extension Emitting ninja build file C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\build.ninja... Compiling objects... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) 1.10.2 "d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\BIN\x86_amd64\link.exe" /nologo /INCREMENTAL:NO /LTCG /DLL /MANIFEST:EMBED,ID=2 /MANIFESTUAC:NO /LIBPATH:/usr/local/cuda/lib64/stubs /LIBPATH:E:\Swin3Project\Swin3\lib\site-packages\torch\lib "/LIBPATH:C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA\v11.7\lib/x64" /LIBPATH:E:\Swin3Project\Swin3\libs /LIBPATH:E:\Swin3Project\Swin3 /LIBPATH:E:\Swin3Project\Swin3\PCbuild\amd64 "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\LIB\amd64" "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\ATLMFC\LIB\amd64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\ucrt\x64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\um\x64" cuda.lib nvrtc.lib c10.lib torch.lib torch_cpu.lib torch_python.lib cudart.lib c10_cuda.lib torch_cuda_cu.lib torch_cuda_cpp.lib /EXPORT:PyInit_tutel_custom_kernel C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj /OUT:build\lib.win-amd64-cpython-39\tutel_custom_kernel.cp39-win_amd64.pyd /IMPLIB:C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom\tutel_custom_kernel.cp39-win_amd64.lib LINK : fatal error LNK1181: 无法打开输入文件“C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj” Try installing without CUDA extension.. running bdist_wheel running build running build_py running build_ext building 'tutel_custom_kernel' extension Emitting ninja build file C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\build.ninja... Compiling objects... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) 1.10.2 "d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\BIN\x86_amd64\link.exe" /nologo /INCREMENTAL:NO /LTCG /DLL /MANIFEST:EMBED,ID=2 /MANIFESTUAC:NO /LIBPATH:/usr/local/cuda/lib64/stubs /LIBPATH:E:\Swin3Project\Swin3\lib\site-packages\torch\lib /LIBPATH:E:\Swin3Project\Swin3\libs /LIBPATH:E:\Swin3Project\Swin3 /LIBPATH:E:\Swin3Project\Swin3\PCbuild\amd64 "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\LIB\amd64" "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\ATLMFC\LIB\amd64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\ucrt\x64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\um\x64" c10.lib torch.lib torch_cpu.lib torch_python.lib /EXPORT:PyInit_tutel_custom_kernel C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj /OUT:build\lib.win-amd64-cpython-39\tutel_custom_kernel.cp39-win_amd64.pyd /IMPLIB:C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom\tutel_custom_kernel.cp39-win_amd64.lib LINK : fatal error LNK1181: 无法打开输入文件“C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj” error: command 'd:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\BIN\x86_amd64\link.exe' failed with exit code 1181 [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip. ERROR: Failed building wheel for tutel Running setup.py clean for tutel Failed to build tutel Installing collected packages: tutel Running setup.py install for tutel ... error error: subprocess-exited-with-error

× Running setup.py install for tutel did not run successfully. │ exit code: 1 ╰─> [100 lines of output] running install E:\Swin3Project\Swin3\lib\site-packages\setuptools\command\install.py:34: SetuptoolsDeprecationWarning: setup.py install is deprecated. Use build and pip and other standards-based tools. warnings.warn( running build running build_py creating build creating build\lib.win-amd64-cpython-39 creating build\lib.win-amd64-cpython-39\tutel copying tutel\jit.py -> build\lib.win-amd64-cpython-39\tutel copying tutel\moe.py -> build\lib.win-amd64-cpython-39\tutel copying tutel\net.py -> build\lib.win-amd64-cpython-39\tutel copying tutel\system.py -> build\lib.win-amd64-cpython-39\tutel copying tutel__init.py -> build\lib.win-amd64-cpython-39\tutel creating build\lib.win-amd64-cpython-39\tutel\checkpoint copying tutel\checkpoint\gather.py -> build\lib.win-amd64-cpython-39\tutel\checkpoint copying tutel\checkpoint\scatter.py -> build\lib.win-amd64-cpython-39\tutel\checkpoint copying tutel\checkpoint__init.py -> build\lib.win-amd64-cpython-39\tutel\checkpoint creating build\lib.win-amd64-cpython-39\tutel\custom copying tutel\custom__init.py -> build\lib.win-amd64-cpython-39\tutel\custom creating build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_amp.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_ddp.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_ddp_tutel.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_deepspeed.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\helloworld_from_scratch.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\moe_cifar10.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples\moe_mnist.py -> build\lib.win-amd64-cpython-39\tutel\examples copying tutel\examples__init.py -> build\lib.win-amd64-cpython-39\tutel\examples creating build\lib.win-amd64-cpython-39\tutel\experts copying tutel\experts\ffn.py -> build\lib.win-amd64-cpython-39\tutel\experts copying tutel\experts\init__.py -> build\lib.win-amd64-cpython-39\tutel\experts creating build\lib.win-amd64-cpython-39\tutel\gates copying tutel\gates\cosine_top.py -> build\lib.win-amd64-cpython-39\tutel\gates copying tutel\gates\top.py -> build\lib.win-amd64-cpython-39\tutel\gates copying tutel\gates\init__.py -> build\lib.win-amd64-cpython-39\tutel\gates creating build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\communicate.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\fast_dispatch.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\jit_compiler.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\losses.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\moe_layer.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\overlap.py -> build\lib.win-amd64-cpython-39\tutel\impls copying tutel\impls\init.py -> build\lib.win-amd64-cpython-39\tutel\impls creating build\lib.win-amd64-cpython-39\tutel\jit_kernels copying tutel\jit_kernels\gating.py -> build\lib.win-amd64-cpython-39\tutel\jit_kernels copying tutel\jit_kernels\sparse.py -> build\lib.win-amd64-cpython-39\tutel\jit_kernels copying tutel\jit_kernels\init__.py -> build\lib.win-amd64-cpython-39\tutel\jit_kernels creating build\lib.win-amd64-cpython-39\tutel\launcher copying tutel\launcher\execl.py -> build\lib.win-amd64-cpython-39\tutel\launcher copying tutel\launcher\run.py -> build\lib.win-amd64-cpython-39\tutel\launcher copying tutel\launcher\init.py -> build\lib.win-amd64-cpython-39\tutel\launcher creating build\lib.win-amd64-cpython-39\tutel\parted copying tutel\parted\patterns.py -> build\lib.win-amd64-cpython-39\tutel\parted copying tutel\parted\solver.py -> build\lib.win-amd64-cpython-39\tutel\parted copying tutel\parted\spmdx.py -> build\lib.win-amd64-cpython-39\tutel\parted copying tutel\parted\init.py -> build\lib.win-amd64-cpython-39\tutel\parted creating build\lib.win-amd64-cpython-39\tutel\parted\backend copying tutel\parted\backend\init.py -> build\lib.win-amd64-cpython-39\tutel\parted\backend creating build\lib.win-amd64-cpython-39\tutel\parted\backend\torch copying tutel\parted\backend\torch\config.py -> build\lib.win-amd64-cpython-39\tutel\parted\backend\torch copying tutel\parted\backend\torch\executor.py -> build\lib.win-amd64-cpython-39\tutel\parted\backend\torch copying tutel\parted\backend\torch\init__.py -> build\lib.win-amd64-cpython-39\tutel\parted\backend\torch running build_ext building 'tutel_custom_kernel' extension creating C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39 creating C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release creating C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\tutel creating C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\tutel\custom Emitting ninja build file C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\build.ninja... Compiling objects... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) 1.10.2 "d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\BIN\x86_amd64\link.exe" /nologo /INCREMENTAL:NO /LTCG /DLL /MANIFEST:EMBED,ID=2 /MANIFESTUAC:NO /LIBPATH:/usr/local/cuda/lib64/stubs /LIBPATH:E:\Swin3Project\Swin3\lib\site-packages\torch\lib "/LIBPATH:C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA\v11.7\lib/x64" /LIBPATH:E:\Swin3Project\Swin3\libs /LIBPATH:E:\Swin3Project\Swin3 /LIBPATH:E:\Swin3Project\Swin3\PCbuild\amd64 "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\LIB\amd64" "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\ATLMFC\LIB\amd64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\ucrt\x64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\um\x64" cuda.lib nvrtc.lib nccl.lib c10.lib torch.lib torch_cpu.lib torch_python.lib cudart.lib c10_cuda.lib torch_cuda_cu.lib torch_cuda_cpp.lib /EXPORT:PyInit_tutel_custom_kernel C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj /OUT:build\lib.win-amd64-cpython-39\tutel_custom_kernel.cp39-win_amd64.pyd /IMPLIB:C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom\tutel_custom_kernel.cp39-win_amd64.lib LINK : fatal error LNK1181: 无法打开输入文件“nccl.lib” Try installing without NCCL extension.. running install running build running build_py running build_ext building 'tutel_custom_kernel' extension Emitting ninja build file C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\build.ninja... Compiling objects... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) 1.10.2 "d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\BIN\x86_amd64\link.exe" /nologo /INCREMENTAL:NO /LTCG /DLL /MANIFEST:EMBED,ID=2 /MANIFESTUAC:NO /LIBPATH:/usr/local/cuda/lib64/stubs /LIBPATH:E:\Swin3Project\Swin3\lib\site-packages\torch\lib "/LIBPATH:C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA\v11.7\lib/x64" /LIBPATH:E:\Swin3Project\Swin3\libs /LIBPATH:E:\Swin3Project\Swin3 /LIBPATH:E:\Swin3Project\Swin3\PCbuild\amd64 "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\LIB\amd64" "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\ATLMFC\LIB\amd64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\ucrt\x64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\um\x64" cuda.lib nvrtc.lib c10.lib torch.lib torch_cpu.lib torch_python.lib cudart.lib c10_cuda.lib torch_cuda_cu.lib torch_cuda_cpp.lib /EXPORT:PyInit_tutel_custom_kernel C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj /OUT:build\lib.win-amd64-cpython-39\tutel_custom_kernel.cp39-win_amd64.pyd /IMPLIB:C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom\tutel_custom_kernel.cp39-win_amd64.lib LINK : fatal error LNK1181: 无法打开输入文件“C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj” Try installing without CUDA extension.. running install running build running build_py running build_ext building 'tutel_custom_kernel' extension Emitting ninja build file C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release\build.ninja... Compiling objects... Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N) 1.10.2 "d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\BIN\x86_amd64\link.exe" /nologo /INCREMENTAL:NO /LTCG /DLL /MANIFEST:EMBED,ID=2 /MANIFESTUAC:NO /LIBPATH:/usr/local/cuda/lib64/stubs /LIBPATH:E:\Swin3Project\Swin3\lib\site-packages\torch\lib /LIBPATH:E:\Swin3Project\Swin3\libs /LIBPATH:E:\Swin3Project\Swin3 /LIBPATH:E:\Swin3Project\Swin3\PCbuild\amd64 "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\LIB\amd64" "/LIBPATH:d:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\ATLMFC\LIB\amd64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\ucrt\x64" "/LIBPATH:C:\Program Files (x86)\Windows Kits\10\lib\10.0.20348.0\um\x64" c10.lib torch.lib torch_cpu.lib torch_python.lib /EXPORT:PyInit_tutel_custom_kernel C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj /OUT:build\lib.win-amd64-cpython-39\tutel_custom_kernel.cp39-win_amd64.pyd /IMPLIB:C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom\tutel_custom_kernel.cp39-win_amd64.lib LINK : fatal error LNK1181: 无法打开输入文件“C:\Users\Acer\AppData\Local\Temp\pip-req-build-5dyo0_c8\build\temp.win-amd64-cpython-39\Release./tutel/custom/custom_kernel.obj” error: command 'd:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\BIN\x86_amd64\link.exe' failed with exit code 1181 [end of output]

note: This error originates from a subprocess, and is likely not a problem with pip. error: legacy-install-failure

× Encountered error while trying to install package. ╰─> tutel

note: This is an issue with the package mentioned above, not pip. hint: See above for output from the failure.

[ghostplant]

Seems like it is an issue of Visual Studio for conda, you may refer this: https://github.com/pytorch/pytorch/issues/30379