On May 19, 2022 NVIDIA made a release of their Open GPU kernel modules for their newer GPU platforms (Turing and newer) with Risc-V system processor. Meanwhile we have packages available in our currently supported openSUSE/SLE distributions. If you want to use these you need to install nvidia-open-driver-G06-signed package.

Installation

Installation instructions since Leap 15.6/SLE15-SP6 and Tumbleweed:

# will install needed packages
zypper in nvidia-open-driver-G06-signed-kmp-default

Find supported Turing/Ampere/Hopper GPUs here. Check with inxi -aG. Use hwinfo --gfxcard on SLE.

Display Drivers

nvidia-video-G06, nvidia-gl-G06 and nvidia-compute-utils-G06 packages are available via NVIDIA’s openSUSE/SLE repositories, which then can be used together with NVIDIA’s Open GPU kernel modules above.

Installing Display Drivers on Leap 15.6/Tumbleweed/SLE15-SPx

# if you have not added this repository yet
# Leap 15.6
zypper addrepo https://download.nvidia.com/opensuse/leap/15.6/  nvidia
# Tumbleweed
zypper addrepo https://download.nvidia.com/opensuse/tumbleweed/  nvidia
# SLE15-SP6
zypper addrepo https://download.nvidia.com/suse/sle15sp6/  nvidia
# SLE15-SP7 (Beta)
zypper addrepo https://download.nvidia.com/suse/sle15sp7/  nvidia

# install all required packages
zypper in nvidia-video-G06 nvidia-gl-G06 nvidia-compute-utils-G06

CUDA

With that - after installing nvidia-compute-utils-G06 (contains libcuda) - you can experiment with CUDA. Install CUDA stack from NVIDIA’s webserver.

Installing CUDA on Leap 15.6/Tumbleweed/SLE15-SPx

# if you have not added this repository yet
# Leap 15.6/Tumbleweed
zypper addrepo https://developer.download.nvidia.com/compute/cuda/repos/opensuse15/x86_64/  cuda
# SLE15-SPx (x86_64)
zypper addrepo https://developer.download.nvidia.com/compute/cuda/repos/sles15/x86_64/  cuda
# SLE15-SPx (aarch64)
zypper addrepo https://developer.download.nvidia.com/compute/cuda/repos/sles15/sbsa/  cuda

# will install needed CUDA packages
zypper in cuda-12-8

Let’s have a first test for using libcuda.

/usr/local/cuda-12.8/extras/demo_suite/deviceQuery
# Unfortunately this is not available for aarch64; but there are
# other samples available below /usr/local/cuda-12.4/extras/ ...

CUDA Minimal Installation

Users, who don’t need a graphical desktop, can omit the installation of the display driver packages above and perform a CUDA Minimal Installation instead.

zypper in --no-recommends nvidia-compute-utils-G06 cuda-libraries-12-8

Feedback

If you have questions, comments and any kind of feedback regarding this topic, don’t hesitate to contact me via email. Thanks!