I use conda env and tensorflow-gpu==2.4.0 for my GPU.
The tricky part is installing Pytorch 3d. Installing it first will save much trouble.
I installed pytorch3d-0.7.1 to avoid compiling errors.
Corresponding torch version and CUDA toolkit:
pip --default-timeout=1000 install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 -f https://download.pytorch.org/whl/torch_stable.html
For compiling pytorch3d, I also installed CUDA 11.1 besides my original version of CUDA on Windows.
VS 2019 might also be necessary for compiling. More details, please check: https://stackoverflow.com/questions/62304087/installing-pytorch3d-fails-with-anaconda-and-pip-on-windows-10
Please refer to issues for other errors, which might be caused by package versions.
I use conda env and tensorflow-gpu==2.4.0 for my GPU. The tricky part is installing Pytorch 3d. Installing it first will save much trouble. I installed pytorch3d-0.7.1 to avoid compiling errors. Corresponding torch version and CUDA toolkit:
pip --default-timeout=1000 install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 -f https://download.pytorch.org/whl/torch_stable.html
For compiling pytorch3d, I also installed CUDA 11.1 besides my original version of CUDA on Windows. VS 2019 might also be necessary for compiling. More details, please check: https://stackoverflow.com/questions/62304087/installing-pytorch3d-fails-with-anaconda-and-pip-on-windows-10 Please refer to issues for other errors, which might be caused by package versions.