You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This guide applies for the latest stable release of your distribution.
Consult this section to find out the correct architecture for your nVidia GPU.
The CUDA package is optional unless you want CUDA, NVDEC & NVENC support. It's typically used for AI/ML, video editing, 3D rendering, cryptography, etc.
nVidia recommends using the open kernel modules (OKM) for Turing & later GPUs.
Blackwell & later GPUs only support the open kernel modules driver.
For Fermi & Kepler GPUs:
The Nouveau drivers are generally recommended, as nVidia no longer supports these GPUs with the proprietary drivers. Keep in mind that Nouveau has significantly lower 3D performance & lacks some features compared to the proprietary drivers.
If you choose to install the proprietary drivers, you're limited to kernel 6.1 LTS for Fermi & kernel 6.6 LTS for Kepler, as these are the last Linux kernels supported by these drivers. The proprietary drivers may also have issues such as incompatibility with newer kernels, potential security vulnerabilities & no Wayland support.
The important part is AAxxx(M), AA is the architecture letter (e.g, GBxxx, GKxxx, etc.), xxx is the tier (e.g, GK110, AD107M, etc.) & (M) indicated if it's a mobile GPU.
If you're using a kernel other than the standard linux kernel, replace the linux-headers package with the header package that matches your kernel (e.g., linux-lts-headers for the LTS kernel, linux-zen-headers for the Zen kernel, etc.).
If Secure Boot is enabled, consulte the Secure Boot guide before installing.
Driver Installation
Warning
After you choose the correct driver for your system & install it, you'll need to wait atleast 5-15 minutes before you reboot so the driver builds in the background. You can check progress using this command:
They're 2 ways of installing nVidia drivers on Debian.
Debian-packages drivers: nVidia drivers provided by Debian which are stable & well-tested. However, they are much older than the upstream releases & may not support the latest GPUs, features & improvements. Recommended for stability. Best for older or current GPUs & a system that just works without breaking.
nVidia-packaged drivers: nVidia drivers provided directly by NVIDIA which are the latest upstream releases. They support the newest GPUs, features & improvements. However, they are not as thoroughly tested by Debian which carry a stability risk. Recommended for performance. Best for brand-new GPUs, gaming & the latest AI/ML (CUDA) features.
Debian-packages drivers
Prerequisites
Update your system:
sudo apt update && sudo apt upgrade -y
reboot
Enable required Debian repository components:
sudo nano /etc/apt/sources.list
Add the components contrib, non-free & non-free-firmware (Order of components doesn't matter)
It should look something like this: (Example for Debian 13; steps should be similar)
deb http://deb.debian.org/debian/ trixie main contrib non-free non-free-firmware
deb-src http://deb.debian.org/debian/ trixie main contrib non-free non-free-firmware
deb http://security.debian.org/debian-security/ trixie-security main contrib non-free non-free-firmware
deb-src http://security.debian.org/debian-security/ trixie-security main contrib non-free non-free-firmware
deb http://deb.debian.org/debian/ trixie-updates main contrib non-free non-free-firmware
deb-src http://deb.debian.org/debian/ trixie-updates main contrib non-free non-free-firmware
Note
If you use -backports, make sure these components are enabled there as well.
Save (Ctrl + S) & exit (Ctrl + X) then run sudo apt update
Install Kernel Headers:
sudo apt install linux-headers-generic
32-bit support:
This is needed for 32-bit apps (such as many Windows games through Wine or Steam), so it's recommended to enable the i386 architecture before installing the driver to ensure the required 32-bit libs are installed automatically.
sudo dpkg --add-architecture i386
sudo apt update
Caution
If your system uses any of the features below, complete the required steps before installing the drivers:
Secure Boot (Based on my testing, it appears that the driver handles Secure Boot automatically, so you may not need to follow that step)
Same prerequisites as the Debian-packaged drivers, with a few extra steps.
Network Repository Enablement:
Important
If you're using a Maxwell, Pascal, or Volta GPU, you'll need to use the Debian 12 (debian12) repository, as it contains the 580 driver required by these GPUs. You'll also need follow these additional steps (Use the apt sources (/etc/apt/sources.list) method).
if you're using an ARM CPU, replace x86_64 with sbsa.
# Replace ${distro} with the Debian version you're running (e.g. debian13)
wget https://developer.download.nvidia.com/compute/cuda/repos/${distro}/x86_64/cuda-keyring_1.1-1_all.deb
sudo dpkg -i cuda-keyring_1.1-1_all.deb
sudo apt update
If Secure Boot is enabled, consulte the Secure Boot guide before installing.
Driver Installation
Warning
After you choose the correct driver for your system & install it, you'll need to wait atleast 5-15 minutes before you reboot so the driver builds in the background. You can check progress using this command:
The proprietary nvidia[version] driver are not supported on the musl-libc variant of Void Linux. You'll need to use the open-source nouveau driver instead.
For these distros, the official documentation already provides detailed nVidia driver installation instructions. Follow the linked guide for your system.
