- Remove the currently installed driver from your AlmaLinux server if installed using CUDA toolkit runfile.
/usr/local/cuda-X.Y/bin/cuda-uninstall
- If you have used the NVIDIA driver runfile for installation, use the following command for uninstallation.
/usr/bin/nvidia-uninstall
- Add EPEL repository,
dnf install epel-release
- Run the update command to let the system rebuild the repository cache. After updating, you will not face any problem while installing any package available via this repo.
dnf update
- Check the added EPEL repo.
rpm -q epel-release
- Add the CUDA repository. NVIDIA maintains its own repository here.
yum config-manager --add-repo http://developer.download.nvidia.com/compute/cuda/repos/rhel8/x86_64/cuda-rhel8.repo
- Install kernel-headers so DKMS can compile the NVIDIA kernel module.
dnf install kernel-devel
- Install CUDA.
dnf install cuda
- Reboot the server.
- After reboot, check available GPU with nvidia-smi.
