Set of scripts to build PyTorch wheels for sm_75 (Nvidia Tegra T4) and aarch64 (arm64).
I wanted to make sure built torch wheels are compatible sith SwarmUI so I used SwarmUI's docker image as a base, but, it will probably work with any Ubuntu based image aswell.
-
Docker
-
NVIDIA Docker
-
NVIDIA GPU with compute capability 7.5 (Tegra T4)
-
Ubuntu 22.04
-
At least 575 driver version
-
CUDA 11.8
RUN wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/sbsa/cuda-keyring_1.1-1_all.deb
RUN dpkg -i cuda-keyring_1.1-1_all.deb
RUN apt update
RUN apt install cuda-toolkit-11-8 -y
RUN echo 'export PATH=/usr/local/cuda/bin:$PATH' | tee /etc/profile.d/cuda.sh
RUN echo 'export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH' | tee -a /etc/profile.d/cuda.shAI is glorified stats, deal with it...