Last active
November 6, 2021 14:46
-
-
Save maruel/e99622298891cc856044e8c158a83fdd to your computer and use it in GitHub Desktop.
Script to install CUDA and libcudnn on Ubuntu 20.04
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
#/bin/bash | |
# Copyright 2021 Marc-Antoine Ruel. No rights reserved. | |
# Official version at https://gist.github.com/maruel/e99622298891cc856044e8c158a83fdd | |
# | |
# Used the following as reference: | |
# - https://developer.nvidia.com/cuda-downloads?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=20.04&target_type=deb_network | |
# - https://docs.nvidia.com/deeplearning/cudnn/install-guide/index.html | |
# - https://youtu.be/BXFGEGO8UY0 | |
# | |
# I recommend looking at all 3 references before moving on. | |
set -eu | |
# First, install the Ubuntu native "server" nvidia drivers via the GUI application. | |
# At the time of writing it is nvidia-driver-470. After running the following I | |
# got nvidia-driver-495 so YMMV. Also note that for recent graphic cards you | |
# generally need a recent (read: buggy) driver. | |
# Want something recent? lolsob. | |
OS='ubuntu2004' | |
# The following code tries to pin too much. Reconsider not pinning at all. | |
cudnn_version='8.3.0.*' | |
cuda_version='cuda11.5' | |
# Should we pin? Not sure. | |
wget https://developer.download.nvidia.com/compute/cuda/repos/${OS}/x86_64/cuda-${OS}.pin | |
sudo mv cuda-${OS}.pin /etc/apt/preferences.d/cuda-repository-pin-600 | |
# What's really important: | |
sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/${OS}/x86_64/7fa2af80.pub | |
sudo add-apt-repository "deb https://developer.download.nvidia.com/compute/cuda/repos/${OS}/x86_64/ /" | |
sudo apt-get update | |
# Notice how cuda and libcunn8 use different ways to pin. Folks at nvidia should talk to each other. | |
sudo apt-get -y install cuda | |
sudo apt-get install libcudnn8=${cudnn_version}-1+${cuda_version} | |
sudo apt-get install libcudnn8-dev=${cudnn_version}-1+${cuda_version} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment