How to use NVIDIA GPUs on Linux
Follow this wiki step by step to set up NVIDIA GPU on Linux (using Ubuntu 22.04 as an example)
1 Install a Suitable NVIDIA Driver¶
Method 1: Manually download and install¶
This driver can be manually downloaded and installed from the NVIDIA official website.
Method 2: Command line¶
Update your package list¶
sudo apt update
List available drivers:¶
sudo apt-get install ubuntu-drivers-common
ubuntu-drivers devices
Install the recommended or the specific version NVIDIA driver¶
sudo ubuntu-drivers autoinstall
sudo apt install nvidia-driver-532
Reboot your computer¶
sudo reboot
2 Install CUDA Toolkit¶
Visit the CUDA Toolkits Downloads pages on the NVIDIA official website.
Then choose the options according to your situation.
NVIDIA provide three ways to install the CUDA Toolkit with specific version.
Here is the runfile(local) I recommend because I am sure you are as lazy as me.
Installation of other development tools
While you run the command sudo sh cuda_xxxxxx_linux.run, you need several development tools, such as gcc, g++, make, etc.
If you haven't installed them yet, you can use sudo apt install build-essential to install almost of them.
After the installation is complete, to ensure that nvcc and other CUDA commands can be called from anywhere, you need to add the CUDA Toolkit's binary directory to your PATH environment variable, as well as configure the LD_LIBRARY_PATH for the library files.
You can add the following lines to your .bashrc or .profile file:
export PATH=/usr/local/cuda-12.4/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda-12.4/lib64:$LD_LIBRARY_PATH
Finally, use the following command to test the environment configuration:
nvcc -V