AlmaLinux users with NVIDIA GPUs now have a smoother experience with driver and CUDA updates after a new collaboration between the AlmaLinux project and NVIDIA.
The AlmaLinux team announced that starting with CUDA version 13.2, NVIDIA has added official support for enterprise Linux distributions, including AlmaLinux. Users with NVIDIA hardware on AlmaLinux can now use NVIDIA’s officially supported configurations for CUDA and drivers.
AlmaLinux has also introduced a new approach to delivering NVIDIA software. CUDA components and driver userspace packages will now be distributed through an official AlmaLinux repository, simplifying installation and keeping the stack synchronized.
Previously, when NVIDIA released new drivers, AlmaLinux sometimes needed extra time to package and publish updates. This occasionally caused brief version mismatches between kernel drivers and userspace CUDA components.
With the new setup, NVIDIA-built packages are redistributed directly through the AlmaLinux infrastructure, enabling driver and CUDA updates to arrive simultaneously.
Under this arrangement, the CUDA and userspace RPM packages remain signed by NVIDIA but are served from the nvidia.repo.almalinux.org repository instead of the AlmaLinux mirror network. AlmaLinux continues to build and sign the open-source NVIDIA kernel modules it ships.
For users who have already installed AlmaLinux’s open source NVIDIA drivers, the transition requires no manual changes. The next system update will automatically switch to the new repository configuration and update packages accordingly.
Users who want to confirm which NVIDIA packages they are currently using can check with rpm -q almalinux-release-nvidia-driver.
For example, on AlmaLinux 9 and 10 systems, the command should return packages such as:
- almalinux-release-nvidia-driver-9-4.el9.x86_64
- almalinux-release-nvidia-driver-10-4.el10.x86_64
Documentation for installing the NVIDIA driver and CUDA stack on AlmaLinux is available on the project’s wiki.
For more information, see Alma’s announcement.
CUDA is NVIDIA’s parallel computing platform that lets software use a GPU for general-purpose computing. Applications can offload compute-intensive tasks to the GPU rather than running them solely on the CPU. This is widely used in AI and machine learning.
