Torch 2.0.1 for Nvidia Jetson AGX Orin Dev Kit 64GB

iqbal.rahmoon · August 1, 2023, 10:56pm

I need to install torch 2.0.1 for NVidia AGX Orin 64GB dev kit but not able to find it. Can anyone share the link please. If I am install using pip install torch then torch.cuda.is_available() return false but when installing from here https://developer.download.nvidia.com/compute/redist/jp/v51/pytorch/torch-2.0.0a0+8aa34602.nv23.03-cp38-cp38-linux_aarch64.whl I can get the cuda device but then I am not able to install torchvision and torchaudio packages. Hope to get response soon. thanks

kayccc · August 2, 2023, 12:24am

Your topic was posted in the wrong category. I am moving this to the Jetson AGX Orin forum for visibility.

AastaLLL · August 2, 2023, 3:36am

Hi,

Do you meet errors when installing torchvision and torchaudio ?
Could you share the error log with us?

Or you can try our l4t-pytorch container.
It has PyTorch, torchvision and torchaudio installed.

Thanks.

dusty_nv · August 2, 2023, 12:15pm

BTW torch 2.0.1 wheel is found here if you require it:

https://developer.download.nvidia.com/compute/redist/jp/v512/pytorch/torch-2.1.0a0+41361538.nv23.06-cp38-cp38-linux_aarch64.whl

iqbal.rahmoon · August 2, 2023, 8:21pm

Thank you

iqbal.rahmoon · August 2, 2023, 8:22pm

I meant there isn’t any error except that I am not able to find the compatible version of torchvision and torchaudio. see below
ERROR: torchvision 0.15.2 has requirement torch==2.0.1, but you’ll have torch 2.0.0a0+8aa34602.nv23.3 which is incompatible.
ERROR: torchaudio 2.0.2 has requirement torch==2.0.1, but you’ll have torch 2.0.0a0+8aa34602.nv23.3 which is incompatible.

dusty_nv · August 2, 2023, 8:25pm

@iqbal.rahmoon I believe these message imply that you are trying to install torchvision/torchaudio via pip and their wheels on PyPi, which won’t work anyways because those PyPi wheels are built without CUDA support. Instead, see the Installation section of this post to build torchvision from source, or see the l4t-pytorch dockerfile to see how I build them:

github.com

dusty-nv/jetson-containers/blob/74afe69b91ed8256e1cd344a6c52301a36e631bc/Dockerfile.pytorch#L61


      
          ARG PYTORCH_URL
          ARG PYTORCH_WHL
          
          RUN wget --quiet --show-progress --progress=bar:force:noscroll --no-check-certificate ${PYTORCH_URL} -O ${PYTORCH_WHL} && \
              pip3 install --no-cache-dir --verbose ${PYTORCH_WHL} && \
              rm ${PYTORCH_WHL}
          
          RUN python3 -c 'import torch; print(f"PyTorch version: {torch.__version__}"); print(f"CUDA available:  {torch.cuda.is_available()}"); print(f"cuDNN version:   {torch.backends.cudnn.version()}"); print(torch.__config__.show());'
          
          
          #
          # torchvision
          #
          ARG TORCHVISION_VERSION
          ARG TORCH_CUDA_ARCH_LIST="5.3;6.2;7.2;8.7"
          
          RUN printenv && echo "torchvision version = $TORCHVISION_VERSION" && echo "TORCH_CUDA_ARCH_LIST = $TORCH_CUDA_ARCH_LIST"
          
          RUN apt-get update && \
              apt-get install -y --no-install-recommends \
          		  git \

system · August 23, 2023, 3:28am

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.