How to Install NVIDIA Drivers on Linux

The world of high-performance computing, from advanced simulations to machine learning model training and demanding graphical workloads, often hinges on the power of NVIDIA GPUs. For Linux users, harnessing this power effectively requires the correct installation and configuration of NVIDIA’s proprietary drivers. While the open-source Nouveau drivers offer basic functionality, they generally fall short of the performance and feature set required for professional applications. This guide will walk you through the process of installing NVIDIA drivers on a Linux system, ensuring optimal performance and stability for your GPU.

Table of Contents

Understanding Driver Installation Methods

Installing NVIDIA drivers on Linux is not a monolithic process. Several methods exist, each with its own advantages and complexities. The best approach often depends on your Linux distribution, your comfort level with the command line, and whether you prefer automated or manual installations. Understanding these distinctions is crucial before diving into the installation itself.

Distribution-Specific Package Managers

Most major Linux distributions provide a streamlined way to install proprietary NVIDIA drivers through their native package managers. This is often the easiest and most recommended method for users who prioritize simplicity and ongoing system updates. These packages are typically tested and curated by the distribution maintainers, offering a higher degree of compatibility with the rest of your system.

Popular Distribution Examples:

Ubuntu/Debian-based Systems: These distributions commonly utilize the apt package manager. NVIDIA drivers are usually available in repositories like multiverse or through dedicated PPA (Personal Package Archive) repositories.
Fedora/RHEL-based Systems: For these systems, the dnf or yum package managers are used. Drivers might be available in third-party repositories such as RPM Fusion.
Arch Linux: Arch users can leverage the Arch User Repository (AUR) for up-to-date NVIDIA drivers, or install them directly from the official repositories if available.

The advantage of using distribution-provided packages is that they often handle kernel module rebuilding automatically after kernel updates, preventing common issues that can arise when the kernel and driver modules are out of sync.

NVIDIA’s Official `.run` Installer

NVIDIA also provides a self-contained installer script (a .run file) directly from their website. This method offers the most up-to-date drivers, often including the very latest features and performance optimizations. It’s particularly useful for users who need the absolute newest driver version or are running a less common Linux distribution that might not have readily available driver packages.

However, this method requires more manual intervention and carries a higher risk of compatibility issues, especially after system or kernel updates. The installer compiles kernel modules from source, and if the kernel changes significantly, these modules will need to be recompiled. This often necessitates manually running the installer again or using tools designed to manage kernel module recompilation.

Containerized or Virtualized Environments

For specific use cases, such as running machine learning frameworks or particular applications, users might opt for containerization technologies like Docker or NVIDIA’s own CUDA Toolkit installations. These methods can abstract away some of the complexities of direct driver installation, but they require a foundational understanding of how containers interact with host hardware. For the purpose of this guide, we will focus on direct driver installation on the host system.

Preparing Your System for Installation

Before you begin the installation process, it’s essential to prepare your Linux system to minimize potential conflicts and ensure a smooth experience. This preparation phase involves backing up critical data, identifying your NVIDIA hardware, and potentially disabling other graphics drivers.

Backing Up Your System

While driver installations are generally safe, unforeseen issues can arise. It’s always prudent to have a recent backup of your important data. Consider using tools like rsync, Timeshift, or your distribution’s built-in backup utilities. This will provide peace of mind and a recovery path should anything go wrong.

Identifying Your NVIDIA Hardware

Knowing the specific NVIDIA GPU model you have is crucial. This information helps you select the correct driver version and verify compatibility.

Methods to Identify Your GPU:

Command Line:
- Open a terminal.
- Run the command: lspci | grep -i nvidia
- This will display information about your NVIDIA graphics card, including its model.
System Information Tools:
- Many desktop environments have graphical system information tools that can display hardware details. Look for “System Settings” or “About This Computer.”
NVIDIA Settings (if already installed):
- If you previously had NVIDIA drivers installed, you might be able to launch nvidia-settings from the terminal or application menu to view hardware details.

Disabling or Uninstalling Existing Graphics Drivers

If you are currently using the open-source Nouveau drivers or a previous version of the NVIDIA proprietary drivers, it’s often best to remove them before installing a new version. This prevents conflicts.

For Nouveau Drivers:

Nouveau drivers are usually part of the kernel. The most common approach to prevent them from loading is to blacklist them.

Create or edit a file in /etc/modprobe.d/ (e.g., sudo nano /etc/modprobe.d/blacklist-nouveau.conf).
Add the following lines to the file:
blacklist nouveau options nouveau modeset=0
Save the file and update your initramfs:
- On Debian/Ubuntu: sudo update-initramfs -u
- On Fedora/RHEL: sudo dracut -f
- On Arch Linux: sudo mkinitcpio -P
Reboot your system.

For Previous NVIDIA Drivers:

If you are uninstalling a previous NVIDIA proprietary driver installation, the method will depend on how it was installed.

If installed via Package Manager: Use your distribution’s package manager to uninstall the NVIDIA packages. For example, on Ubuntu:
bash sudo apt-get --purge remove "*nvidia*" sudo apt autoremove
If installed via .run installer: Navigate to the directory where you ran the installer and execute it with the --uninstall flag:
bash sudo /usr/bin/nvidia-uninstall
(The exact path might vary, but this is a common location.)

Ensuring a Clean Environment: The Command Line Interface

While graphical installers exist for some distributions, a command-line interface (CLI) approach is often the most robust and universally applicable method for installing NVIDIA drivers, especially when dealing with advanced configurations or troubleshooting. It also allows for a cleaner installation process by ensuring that the graphical environment is not actively using the drivers being replaced.

Before proceeding with the installation using either the package manager or the .run file, it’s highly recommended to switch to a text-only console. This prevents conflicts with the currently running display server (Xorg or Wayland).

Switching to a Text Console:

Press Ctrl + Alt + F1 through Ctrl + Alt + F6 to switch to a virtual terminal (TTY).
Log in with your username and password.
You will need to stop your display manager to ensure a clean installation. The command to do this varies by distribution:
- GDM (GNOME Display Manager): sudo systemctl stop gdm
- LightDM: sudo systemctl stop lightdm
- SDDM (KDE Display Manager): sudo systemctl stop sddm
- Other Display Managers: Consult your distribution’s documentation.

Once you have completed the preparation steps and are in a text console with the display manager stopped, you are ready to proceed with the actual driver installation.

Installing Drivers via Distribution Package Managers

This is the most common and recommended method for most Linux users. Each distribution has its own way of managing packages, but the general principle is to leverage the built-in tools to install the proprietary NVIDIA drivers.

Ubuntu/Debian-Based Systems (apt)

Ubuntu and its derivatives offer a convenient way to install NVIDIA drivers through the “Additional Drivers” utility or directly via apt.

Method 1: Using “Additional Drivers” (GUI)

Open “Software & Updates.”
Navigate to the “Additional Drivers” tab.
The system will scan for available proprietary drivers.
Select the NVIDIA driver that is recommended or the latest tested version.
Click “Apply Changes.”
The system will download and install the drivers.
Reboot your system after the installation is complete.

Method 2: Using the Command Line (apt)

Update your package list:
bash sudo apt update
Search for available NVIDIA drivers:
bash sudo apt search nvidia-driver
This command will list various NVIDIA driver packages. Look for packages named nvidia-driver-XXX where XXX is a version number (e.g., nvidia-driver-535). There will also be a meta-package like nvidia-driver-latest or nvidia-headless-XXX for server environments.
Install the recommended driver: A simple way to get the recommended driver is to use the ubuntu-drivers command:
bash sudo ubuntu-drivers autoinstall
Alternatively, if you know the specific driver version you want to install (e.g., nvidia-driver-535):
bash sudo apt install nvidia-driver-535
Reboot your system:
bash sudo reboot

Using PPAs for Newer Drivers:

If the drivers in the default repositories are not recent enough, you can add a PPA. For NVIDIA drivers on Ubuntu, the “graphics-drivers PPA” is a popular choice.

sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt update
sudo apt install nvidia-driver-535 # Or the latest version available from the PPA

Fedora/RHEL-Based Systems (dnf/yum)

Fedora and its enterprise derivatives like CentOS Stream and RHEL typically use the dnf (or yum on older systems) package manager. NVIDIA drivers are usually not included in the default repositories due to licensing. You’ll need to enable third-party repositories. The RPM Fusion repository is a common choice.

Enabling RPM Fusion:

Install RPM Fusion repositories:
- For Fedora:
  bash sudo dnf install https://mirrors.rpmfusion.org/free/fedora/rpmfusion-free-release-$(rpm -E %fedora).noarch.rpm https://mirrors.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-$(rpm -E %fedora).noarch.rpm
- For RHEL/CentOS Stream (consult RPM Fusion documentation for specific versions, as it can vary).
Update your package list:
```
sudo dnf update
```
Install NVIDIA drivers:
- Search for available NVIDIA packages:
  bash sudo dnf search nvidia-driver
- Install the driver (e.g., for a common driver version):
  bash sudo dnf install akmod-nvidia xorg-x11-drv-nvidia-cuda
  akmod-nvidia installs the driver and provides kernel module updates. xorg-x11-drv-nvidia-cuda provides CUDA support.
Reboot your system:
bash sudo reboot

Arch Linux (pacman and AUR)

Arch Linux users have excellent access to up-to-date drivers through official repositories and the Arch User Repository (AUR).

Method 1: Official Repositories

Update your system:
bash sudo pacman -Syu
Install the NVIDIA driver package:
bash sudo pacman -S nvidia
This installs the latest stable NVIDIA driver. For users who prefer the latest features, nvidia-lts is available for the Linux-LTS kernel.
Reboot your system:
bash sudo reboot

Method 2: Using the AUR (e.g., with yay)

For bleeding-edge drivers or specific versions not in the official repos, the AUR is your best bet.

Install an AUR helper (if you don’t have one, yay is popular):
bash sudo pacman -S --needed git base-devel git clone https://aur.archlinux.org/yay.git cd yay makepkg -si
Install the NVIDIA driver from AUR:
bash yay -S nvidia-dkms # For dynamic kernel module support # or yay -S nvidia-open-dkms # For open kernel modules
nvidia-dkms is generally preferred as it handles kernel module recompilation automatically after kernel updates.
Reboot your system:
bash sudo reboot

Installing Drivers via NVIDIA’s `.run` Installer

This method involves downloading the driver installer directly from NVIDIA’s website. It’s more manual but ensures you have the absolute latest version of the driver.

Downloading the NVIDIA Driver

Visit the official NVIDIA driver download page: https://www.nvidia.com/Download/index.aspx
Select your GPU type, series, product, operating system (Linux 64-bit), and language.
Click “Search.”
On the search results page, download the latest recommended driver. The filename will typically look like NVIDIA-Linux-x86_64-XXX.XX.run (where XXX.XX is the version number).
Save the .run file to a convenient location, such as your Downloads folder.

Executing the Installer

Ensure you are in a text console (as described in the preparation section). Stop your display manager.
Navigate to the download directory in your terminal:
bash cd ~/Downloads # Or wherever you saved the file
Make the installer executable:
bash chmod +x NVIDIA-Linux-x86_64-XXX.XX.run
Replace NVIDIA-Linux-x86_64-XXX.XX.run with the actual filename.
Run the installer with root privileges:
bash sudo ./NVIDIA-Linux-x86_64-XXX.XX.run
Follow the on-screen prompts:
- The installer will likely prompt you to accept the license agreement.
- It will check for existing installations and ask if you want to continue.
- It will usually ask if you want to install the NVIDIA xorg configuration file. It’s generally recommended to let the installer manage this.
- The installer will compile kernel modules. If you have development tools (like gcc and make) and kernel headers installed, this should proceed smoothly. If not, the installer will likely prompt you to install them or fail.
- You may be asked about 32-bit compatibility libraries. Install them if you use any 32-bit applications that require them.
- When prompted about registering kernel module sources with DKMS, it’s usually best to say “Yes” if DKMS is installed. This helps with automatic recompilation after kernel updates.
Reboot your system:
bash sudo reboot

Handling Kernel Updates with the `.run` Installer

One of the main drawbacks of the .run installer is that kernel updates can break the NVIDIA driver installation. If the NVIDIA kernel modules are compiled against an older kernel version, they won’t work with the new one.

Solutions:

DKMS (Dynamic Kernel Module Support): If you chose to install with DKMS support during the installation, the NVIDIA kernel modules should automatically recompile when the kernel is updated. Ensure DKMS is installed on your system (sudo apt install dkms on Debian/Ubuntu, sudo dnf install dkms on Fedora, or sudo pacman -S dkms on Arch).
Manual Reinstallation: If DKMS is not used or fails, you will need to re-run the .run installer after a kernel update.
Using Package Managers: This is why using distribution-provided packages is often preferred, as they usually manage kernel module updates more seamlessly.

Verifying the Installation and Post-Installation Steps

After rebooting, it’s crucial to verify that the NVIDIA drivers have been installed correctly and are functioning as expected. This involves checking system information and potentially using NVIDIA’s own tools.

Checking Driver Status

Method 1: Using nvidia-smi

The NVIDIA System Management Interface (nvidia-smi) is a command-line utility that provides detailed information about your NVIDIA GPUs and the drivers.

Open a terminal.
Run the command:
bash nvidia-smi
If the drivers are installed correctly, you will see output listing your GPU(s), the driver version, CUDA version, GPU utilization, memory usage, and temperature. If you see an error message, it indicates a problem with the driver installation.

Method 2: Using nvidia-settings

The NVIDIA Settings utility provides a graphical interface to configure various aspects of your NVIDIA display driver.

Open a terminal.
Run the command:
bash nvidia-settings
If the drivers are installed correctly, the NVIDIA Settings window will appear, showing detailed information about your GPU, display configuration, and various performance settings.

Method 3: Checking Kernel Modules

You can also verify that the NVIDIA kernel module is loaded.

Open a terminal.
Run the command:
bash lsmod | grep nvidia
You should see output indicating that the nvidia module (and possibly others like nvidia_uvm and nvidia_modeset) is loaded.

Configuring Display Settings

Once the drivers are installed, you may want to adjust your display settings. The nvidia-settings utility is the primary tool for this.

Resolution and Refresh Rate: Adjust the screen resolution, refresh rate, and multi-monitor setups.
Color Adjustments: Fine-tune color profiles and gamma settings.
X Server Configuration: For advanced users, direct modifications to the X server configuration can be made, though it’s often better to let nvidia-settings manage these.
Power Management: Configure GPU power saving modes.

Enabling CUDA and Other NVIDIA Technologies

For users who intend to use NVIDIA GPUs for GPGPU computing (e.g., machine learning, scientific simulations), ensuring CUDA is properly installed and configured is paramount.

CUDA Toolkit: If you installed drivers via a package manager, CUDA might be a separate package (sudo apt install nvidia-cuda-toolkit on Ubuntu). If you used the .run installer, it might have offered to install a version of the CUDA Toolkit. You can also download the CUDA Toolkit separately from NVIDIA’s website, which often includes a driver installer itself, but it’s generally recommended to use the distribution’s driver or the standalone driver and then install the CUDA Toolkit.
Environment Variables: Ensure that the CUDA binaries and libraries are in your system’s PATH and LDLIBRARYPATH. This is usually handled by installation scripts, but you might need to add lines like these to your ~/.bashrc or ~/.profile:
bash export PATH=/usr/local/cuda/bin:$PATH export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH
(Adjust the CUDA path if necessary).
Verification: Test your CUDA installation by compiling and running CUDA samples provided with the toolkit.

Troubleshooting Common Issues

Black Screen: This is often caused by conflicts with Nouveau drivers or incorrect Xorg configuration. Ensure Nouveau is blacklisted and reboot. If problems persist, try removing the NVIDIA drivers and reinstalling, possibly using a different method.
nvidia-smi command not found: The NVIDIA binaries might not be in your PATH. This can happen with the .run installer. Ensure /usr/bin (where nvidia-smi is typically installed) is in your PATH.
Driver compilation errors: Ensure you have the correct kernel headers and development tools installed for your kernel version.
- On Debian/Ubuntu: sudo apt install linux-headers-$(uname -r) build-essential
- On Fedora: sudo dnf install kernel-devel kernel-headers gcc make
- On Arch Linux: sudo pacman -S linux-headers base-devel
Performance issues: Double-check that the correct driver is loaded and that power management settings are not hindering performance.

By following these steps, you should be able to successfully install and configure NVIDIA drivers on your Linux system, unlocking the full potential of your hardware for demanding applications. Regular updates and careful management of kernel changes will ensure ongoing stability and performance.