Open Chenxu-666 opened 1 year ago
Find any solution?
any news?
Have u ever tried to look up the dir called /CUDA/bin/nvcc? It's supposed to be in "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v.\bin" in win or in "/usr/CUDA/bin" in Linux if u installed CUDA in a default way. Torch.cpp_extention will find CUDA in auto if /CUDA/bin/nvcc exists in the default dir. So I guess u should go to torch.cpp_extension, find a val called CUDA_HOME and change the dir into where you CUDA is.
problem has been solved already Tanks anyway.
problem has been solved already Tanks anyway.
how did u solve it,I really need it[cry]
When I run sh ./make.sh command, I met with a bug -- NotImplementedError: Cuda is not availabel。Then I create a new conda environment that use python=3.9.13. In pytorch website, I installing the version of cuda is 11.8. But now it still throw this bug. Any guys can help me? I would appreciate!