Qlustar 14 on NVIDIA DGX-1, nvidia-580
Hi, On our NVIDIA DGX-1 Tesla V100 servers we are running Qlustar 14 (latest 14.1.14 update) with Qlustar image with configured nvidia module. I tried to upgrade to nvidia-580, but kernel module doesn't load. From one side it says it is not compatible with V100, but it is compatible. Our Nvidia-noble qlustar image have core,lustre-client,ofed,slurm,nvidia modules. I tried in place of nvidia module to use nvidia-580. The same with nvidia-570 and nvidia-595 (not supported on Tesla V100). Regards Rolandas
Hi Rolandas, unfortunately we can't test the full lineup of Nvidia GPUs. The driver's you used previously should still work, we didn't change them. Just added the new branch versions 580 + 595. Best, Roland On 10/6/26 13:49, Rolandas Naujikas via Qlustar General wrote:
Hi,
On our NVIDIA DGX-1 Tesla V100 servers we are running Qlustar 14 (latest 14.1.14 update) with Qlustar image with configured nvidia module.
I tried to upgrade to nvidia-580, but kernel module doesn't load. From one side it says it is not compatible with V100, but it is compatible.
Our Nvidia-noble qlustar image have core,lustre-client,ofed,slurm,nvidia modules.
I tried in place of nvidia module to use nvidia-580. The same with nvidia-570 and nvidia-595 (not supported on Tesla V100).
Regards Rolandas
Hi Roland, For nvidia 580.178.04 we have choice between open and proprietaries linux kernel modules. Open version supports only newer (Turing and later) GPUs. For older GPUs (like our Tesla V100) we have to use proprietary kernel modules. In Qlustar 14 nvidia-580 is distributing only open version of kernel modules. This is why it don't work with our Tesla V100 GPUs: NVRM: The NVIDIA GPU 0000:86:00.0 (PCI ID: 10de:1db5) NVRM: installed in this system is not supported by open NVRM: nvidia.ko because it does not include the required GPU NVRM: System Processor (GSP). NVRM: Please see the 'Open Linux Kernel Modules' and 'GSP NVRM: Firmware' sections in the driver README, available on NVRM: the Linux graphics driver download page at NVRM: www.nvidia.com. Regards Rolandas On 2026-10-06 21:12, Roland Fehrenbacher via Qlustar General wrote:
Hi Rolandas,
unfortunately we can't test the full lineup of Nvidia GPUs. The driver's you used previously should still work, we didn't change them. Just added the new branch versions 580 + 595.
Best,
Roland
On 10/6/26 13:49, Rolandas Naujikas via Qlustar General wrote:
Hi,
On our NVIDIA DGX-1 Tesla V100 servers we are running Qlustar 14 (latest 14.1.14 update) with Qlustar image with configured nvidia module.
I tried to upgrade to nvidia-580, but kernel module doesn't load. From one side it says it is not compatible with V100, but it is compatible.
Our Nvidia-noble qlustar image have core,lustre- client,ofed,slurm,nvidia modules.
I tried in place of nvidia module to use nvidia-580. The same with nvidia-570 and nvidia-595 (not supported on Tesla V100).
Regards Rolandas
Qlustar General mailing list -- qlustar-general@qlustar.org To unsubscribe send an email to qlustar-general-leave@qlustar.org
participants (2)
-
Roland Fehrenbacher -
Rolandas Naujikas