NVIDIA Open GPU kernel modules: openSUSE/SLE packages available
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!