Installing Client GPUs

This document explains how to choose the appropriate version of Ubuntu Desktop for your client GPU and install the necessary drivers from the latest rolling release.

Selecting the right operating system version

Intel actively collaborates with various upstream projects to enable the best possible experience using the software and drivers provided by Linux operating system distributions. For the best support of Intel graphics products on Linux, we recommend using the kernel, drivers, and packages supplied by your operating system vendor. As new hardware platforms enter the market, updating to the latest version of the operating system may be necessary to enable these platforms.

Selecting the right version of Ubuntu Desktop to support a specific platform can be confusing. To simplify this process, we recommend using the hardware table to locate your PCI device ID and identify the corresponding kernel version required for your platform. This will help you determine the minimum required kernel version and select the proper operating system version:

  • For hardware devices that require at least kernel version 6.12, use Ubuntu 24.10 (Oracular) or newer.

  • For hardware devices that require at least kernel version 6.8, use Ubuntu 24.04 (Noble) or newer.

  • For hardware devices that require at least kernel version 5.15, use Ubuntu 22.04 (Jammy) or newer.

For general information on which kernel is provided by each Ubuntu Desktop release, refer to Ubuntu Kernel Lifecycle.

Installing Client GPUs on Ubuntu Desktop

For Ubuntu 24.10 and 24.04, we offer the intel-graphics Personal Package Archive (PPA). This PPA provides early access to the latest packages, along with additional tools and features, such as EU debugging. Follow these steps to install the intel-graphics PPA and the required compute and media packages.

Note

Using Ubuntu 24.04 with Lunar Lake or Battlemage requires your system to be running the hardware enablement (HWE) kernel. By default, Ubuntu Desktop 24.04 tracks the HWE stack. However, if your system is instead using the general availability (GA) kernel, you must switch to the HWE kernel before proceeding with the Client GPU installation.

  1. Refresh the local package index and install the package for managing software repositories.

    sudo apt-get update
sudo apt-get install -y software-properties-common

  2. Add the intel-graphics PPA.

    sudo add-apt-repository -y ppa:kobuk-team/intel-graphics

  3. Install the compute-related packages.

    sudo apt-get install -y libze-intel-gpu1 libze1 intel-metrics-discovery intel-opencl-icd clinfo intel-gsc

  4. Install the media-related packages.

    sudo apt-get install -y intel-media-va-driver-non-free libmfx1 libmfx-gen1 libvpl2 libvpl-tools libva-glx2 va-driver-all vainfo

The commands listed above install all the essential packages needed for most users, aiming to minimize the installation of unnecessary packages. However, if you plan to use PyTorch, install libze-dev and intel-ocloc additionally:

sudo apt-get install -y libze-dev intel-ocloc

If you wish to enable hardware ray tracing support, install libze-intel-gpu-raytracing additionally:

sudo apt-get install -y libze-intel-gpu-raytracing

The Ubuntu 22.04 repositories do not contain compute packages for various Intel graphics products. To install these packages, you can use Intel’s dedicated package repository.

  1. Install the Intel graphics GPG public key.

    wget -qO - https://repositories.intel.com/gpu/intel-graphics.key | \
  sudo gpg --yes --dearmor --output /usr/share/keyrings/intel-graphics.gpg

  2. Configure the repositories.intel.com package repository.

    echo "deb [arch=amd64,i386 signed-by=/usr/share/keyrings/intel-graphics.gpg] https://repositories.intel.com/gpu/ubuntu jammy unified" | \
  sudo tee /etc/apt/sources.list.d/intel-gpu-jammy.list

  3. Update the package repository metadata.

    sudo apt update

  4. Install the compute-related packages.

    sudo apt-get install -y libze-intel-gpu1 libze1 intel-opencl-icd clinfo

The commands listed above install all the essential packages needed for most users, aiming to minimize the installation of unnecessary packages. However, if you plan to use PyTorch, install libze-dev and intel-ocloc additionally:

sudo apt-get install -y libze-dev intel-ocloc

If you wish to enable hardware ray tracing support, install intel-level-zero-gpu-raytracing additionally:

sudo apt-get install -y intel-level-zero-gpu-raytracing

Verifying installation

To verify that the kernel and compute drivers are installed and functional, run clinfo:

clinfo | grep "Device Name"

You should see the Intel graphics product device names listed. If they do not appear, ensure you have permissions to access /dev/dri/renderD*. This typically requires your user to be in the render group:

sudo gpasswd -a ${USER} render
newgrp render

Alternatively, you can run the clinfo command as root.