How to Install Nvidia Drivers on Debian 12 or 11

Installing NVIDIA drivers on Debian is essential for users who need to harness the full power of their NVIDIA GPUs, whether for gaming, graphics rendering, or CUDA-based computational tasks. Proper installation of these drivers ensures optimal performance and stability of your system, making it crucial to choose the right installation method based on your specific requirements.

On Debian 12 or 11, NVIDIA drivers can be installed using two primary methods. The first method involves using Debian’s default repository, which provides a stable and straightforward installation suitable for most users. For those who require the latest features, updates, and CUDA support, the second method is to add NVIDIA’s CUDA APT repository mirror, which offers access to the latest driver builds. This guide will walk you through both installation methods, ensuring your NVIDIA drivers are set up according to your system’s needs.

Pre-Installation Steps Before Nvidia Drivers Installation

Update Debian System Packages Before Nvidia Drivers Installation

Before proceeding with the tutorial, ensuring your system is up-to-date with all existing packages is good.

sudo apt update

Proceed to upgrade any outdated packages using the following command.

sudo apt upgrade

Previous Nvidia Installations From Debian (Situational)

When Nvidia drivers over the standard packages, it can result in unexpected issues, especially if you install the Nvidia drivers from Debian’s repository, then import the Nvidia Cuda repository and try to re-install or upgrade directly over those packages. If you use one method, stick to it, and if you want to swap, ensure you remove the previous installation to avoid conflicts.

Below, I will briefly review some options for removing the previous Nvidia Drivers.

Example 1: Remove Previous Nvidia Drivers Installed via APT

First, use a blanket command to remove any traces of Nvidia installation on your system:

sudo apt autoremove nvidia* --purge

Example 2: Remove Previous Nvidia Drivers Installed via RUNFILE

The second method is for anyone who installed the Nvidia drivers using the “.run” file. To remove the runfile type of installation, use the following command:

sudo /usr/bin/nvidia-uninstall

Example 3: Remove Previous Nvidia Cuda Drivers Installed via RUNFILE

Lastly, remove the CUDA toolkit runfile installation using the following command.

sudo /usr/local/cuda-X.Y/bin/cuda-uninstall

These few methods may remove the Nvidia Drivers.

Enable Contrib and Non-Free Repositories

The first task is enabling the “contrib” and “non-free” repositories in your Debian repositories. The command changes from Debian 12 Bookworm onwards. Ensure you use the correct command matching your Debian distribution version.

First, make sure you have installed the following on your Debian system:

sudo apt install software-properties-common -y

Debian 12 Bookworm and onwards:

sudo add-apt-repository contrib non-free-firmware

Debian 11 Bullseye downwards:

sudo add-apt-repository contrib non-free

Once added, update your APT cache with the following command:

sudo apt update

Method 1: Install Nvidia Drivers via DEFAULT APT Repository

Install Linux Kernel Headers

Before installing any Nvidia drivers, you must install the proper kernel headers for the NVIDIA driver to build with. You can install either 64bit or 32bit, and you can find this out by running the lscpu command:

lscpu | grep CPU

Example output:

CPU op-mode(s):                  32-bit, 64-bit
CPU(s):                          2
On-line CPU(s) list:             0,1
CPU family:                      23
CPU MHz:                         3900.025
NUMA node0 CPU(s):               0,1

From the output, the system is 64bit, so you need to install the following headers:

sudo apt install linux-headers-amd64

32-bit Debian systems, use the following command instead:

Note: DO NOT USE THE BELOW COMMANDS IF YOU ARE ON A 64-BIT SYSTEM.

non-PAE kernel:

sudo apt install linux-headers-686

PAE kernel:

sudo apt install linux-headers-686-pae

Install Nvidia Drivers via Debian’s Default Repository

Next, you can install the “nvidia-detect,” which will automatically suggest the best package for your graphics card.

To do this, use the following command:

sudo apt install nvidia-detect

With Nvidia-detect installed, execute the command to identify your current Nvidia Graphics Card:

nvidia-detect

The output reveals that the example machine features a GeForce GTX 1650 card and recommends installing the nvidia-driver package. However, this remains a suggestion.

Next, install the recommended package:

sudo apt install nvidia-driver linux-image-amd64

You will see a note about the free Nouveau graphics driver conflicting with the installed new driver. As per the message, you need to reboot to correct this; for now, press the “ENTER KEY” to proceed.

Reboot and Confirm Installation

Once installed, reboot your system:

sudo reboot

Situational: Install Nvidia Legacy Drivers (For Older Cards)

If your Nvidia Graphics card is old from 400 Series downwards, you must install the legacy drivers. The process is the same, just with a new install command:

sudo apt install nvidia-legacy-390xx-driver firmware-misc-nonfree

Once complete, do not forget to reboot your system.

sudo reboot now

You can technically install the 340 series, but it no longer has support and contains several major security flaws. Although you might consider replacing 390xx with 340xx, Debian advises against it.

Method 2: Install Nvidia Drivers via PPA

The second method lets you install the latest drivers for your desktop or server from the Nvidia Cuda APT repository. I suggest you use these drivers because you’ll get updates immediately after their release. This approach benefits security and performance since Nvidia frequently releases updates. Now, you also have the option to install the open-source version.

Identify Your Nvidia Graphics Card on Debian

If you have an older NVIDIA Graphics card, identify it to ensure support. However, if you own a brand-new card, you can skip this part because your card undoubtedly has support.

Now, let’s identify your graphics card module:

lspci | grep -e VGA

Example output:

03:00.0 VGA compatible controller: NVIDIA Corporation TU117 [GeForce GTX 1650] (rev a1)

Import Nvidia GPG Key

First, check if you have already installed the following packages by using this command:

sudo apt install dirmngr ca-certificates software-properties-common apt-transport-https dkms curl -y

Next, import the GPG key, and ensure you match the import to your distribution version.

Note: Please comment or message me via the contact form if the GPG import fails. Debian can occasionally change its GPG keys. This is rare, but it does occur; if it does, I’ll update the command.

Debian 12 Bookworm Import GPG Key for Nvidia:

curl -fSsL https://developer.download.nvidia.com/compute/cuda/repos/debian12/x86_64/3bf863cc.pub | sudo gpg --dearmor | sudo tee /usr/share/keyrings/nvidia-drivers.gpg > /dev/null 2>&1

Debian 11 Bullseye Import GPG Key for Nvidia:

curl -fSsL https://developer.download.nvidia.com/compute/cuda/repos/debian11/x86_64/3bf863cc.pub | sudo gpg --dearmor | sudo tee /usr/share/keyrings/nvidia-drivers.gpg > /dev/null 2>&1

Import Nvidia APT Repository

Second, import the repository for your Debian system, and again, ensure you import the correct one to match your Debian distribution version.

Debian 12 Bookworm Import Nvidia CUDA APT Repository:

echo 'deb [signed-by=/usr/share/keyrings/nvidia-drivers.gpg] https://developer.download.nvidia.com/compute/cuda/repos/debian12/x86_64/ /' | sudo tee /etc/apt/sources.list.d/nvidia-drivers.list

Debian 11 Bullseye Import Nvidia CUDA APT Repository:

echo 'deb [signed-by=/usr/share/keyrings/nvidia-drivers.gpg] https://developer.download.nvidia.com/compute/cuda/repos/debian11/x86_64/ /' | sudo tee /etc/apt/sources.list.d/nvidia-drivers.list

Install Nvidia Drivers – Proprietary or Open-Source Options

Update your sources list to reflect the newly added repository.

sudo apt update

Next, install the latest NVIDIA drivers; below are examples with and without CUDA and proprietary and open-source versions.

Proprietary Nvidia Drivers

Install NVIDIA Drivers Without Cuda Support

sudo apt install nvidia-driver nvidia-smi nvidia-settings

Install NVIDIA Drivers With Cuda Support (Proprietary):

sudo apt install nvidia-driver cuda nvidia-smi nvidia-settings

Open-Source Nvidia Drivers

Install NVIDIA Drivers Without Cuda Support (Opensource):

sudo apt install nvidia-driver nvidia-kernel-open-dkms nvidia-smi nvidia-settings

Install NVIDIA Drivers With Cuda Support (Opensource):

sudo apt install nvidia-driver nvidia-kernel-open-dkms cuda nvidia-smi nvidia-settings

Note: The download is large and can take a while; given this is a general release, it is recommended to install all the drivers to avoid any issues or inconsistencies on your Debian desktop.

One good thing about the Nvidia repository installation is that if you forgot to remove the previous one, you will see a prompt for users with pre-existing installations before the installation continues. Select Yes to remove any existing drivers. Failure to do this may cause system instability. Once done, the installation will be complete.

Reboot

During the final phase of the installation, most users will typically see a message like the image below:

To activate the installation, reboot your system. You can use the following command after completing the installation:

sudo reboot

Situational: Install 32-bit Support for Nvidia Drivers

You can easily enable and install 32-bit support first. Then, install the 64-bit drivers mentioned earlier and follow these steps.

First, activate the 32-bit architecture using this command:

sudo dpkg --add-architecture i386

Update the APT-CACHE to reflect the changes to the architecture:

sudo apt update

Install 32-bit support, and remove the Cuda package “libcuda1-i386” for those users not requiring it.

sudo apt install libcuda1-i386 nvidia-driver-libs-i386

Reboot your PC:

sudo reboot

Confirm Nvidia Drivers Installation

Verify Nvidia Drivers Utilizing via CLI Command

First, verify the installation by running NVIDIA-SMI, as the manual installation steps showed beforehand.

nvidia-smi

If successful, you should see something similar too:

Verify Nvidia Drivers Utilizing via GUI

Alternatively, you can open the GUI for desktop users using the following command:

nvidia-settings

Or open the application icon located in the following path:

Activities > Show Applications > Nvidia X Server

Here, you should see a panel with all your Nvidia Driver settings available and system information. At this stage, for most, you have successfully installed the Nvidia Drivers of your choice on Debian.

Below are additional screenshots of the Nvidia Settings GUI Example Screenshots for users new to Linux and utilizing Nvidia Drivers:

Conclusion and Recap

With NVIDIA drivers installed on your Debian system, either through the default repository or the CUDA APT repository, you can ensure that your GPU is fully optimized for performance. The Debian repository provides a stable and easy-to-manage installation, while the CUDA repository offers access to the latest features and enhancements. Regularly updating your drivers, especially when using the CUDA repository, will help maintain optimal performance and compatibility, allowing you to make the most of your NVIDIA hardware on Debian.