Skip to content

Instantly share code, notes, and snippets.

@ocramz
Created July 7, 2023 12:20
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save ocramz/9061c1649a8d506aa833225a20589180 to your computer and use it in GitHub Desktop.
Save ocramz/9061c1649a8d506aa833225a20589180 to your computer and use it in GitHub Desktop.
Setup CUDA 11.8 drivers, miniconda and pytorch on Ubuntu 22.04 on AWS EC2
#!/bin/bash
apt update
apt -y install build-essential # install GCC
apt -y install --reinstall software-properties-common # enable add-apt-repository
apt-get -y install lshw
apt-get clean
export DISTRO_UBUNTU=ubuntu2204/x86_64
# # Ubuntu 22 - : install CUDA 11.8 (for Pytorch support)
wget https://developer.download.nvidia.com/compute/cuda/repos/${DISTRO_UBUNTU}/cuda-ubuntu2204.pin
mv cuda-ubuntu2204.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda-repo-ubuntu2204-11-8-local_11.8.0-520.61.05-1_amd64.deb
dpkg -i cuda-repo-ubuntu2204-11-8-local_11.8.0-520.61.05-1_amd64.deb
cp /var/cuda-repo-ubuntu2204-11-8-local/cuda-*-keyring.gpg /usr/share/keyrings/
apt-get update
apt-get -y install cuda
# ensure cuda binaries are on path
export PATH=/usr/local/cuda-11.8/bin${PATH:+:${PATH}}
nvcc --version
curl -O https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh -b -f
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment