We got a report that our 580.76.05 packages were crashing X11 for an end-user with a large number of systems. We pulled it and rebased to 570.181 and were able to pinpoint the problem.
I’m in the process of patching all of our NVIDIA packages. Expect in the next few days to see:
Thank you very much. Last time you made driver 580.76.05, I have been using it without problems for almost a month. True, I have Wayland and Gnome40.
I changed my 580.76.05 to 580.82.07 and it works as usual (I have Wayland).
During the update I got: egl-wayland
and there was a failure: nvidia-persisteneed.service
Thanks.
Thanks for the update.
I had to revert the bundling of the newer NVIDIA egl-wayland because:
As for the nvidia-persistenced service error, the updated service is probably being restarted before the updated driver is running. After a reboot is the service is running ok?
systemctl status nvidia-persistenced
I installed the .run for version 580.82.07, and to make it work with wayland I added this to my grub config modprobe.blacklist=nouveau,nova_core nvidia_drm.modeset=1 I’m on rl10 but I believe it should work in rl9. give it a try and let me know.
I find the .run performs better in general.
Ok. All works.
nvidia-persistenced.service - NVIDIA Persistence Daemon
Loaded: loaded (/usr/lib/systemd/system/nvidia-persistenced.service; enabled; preset: enabled)
Active: active (running) since Mon 2025-09-08 07:19:10 +04; 10min ago
Process: 1023 ExecStart=/usr/bin/nvidia-persistenced --user root (code=exited, status=0/SUCCESS)
FYI we have RPM’s for EL10 in elrepo-testing repo for now.
I’ve been monitoring the NVIDIA forum for 580 driver feedback. Still lots of problems reported by gamers, so I don’t want to introduce that to large-scale production environments.
sorry.
nothing provides kernel >= 5.14.0-570.42.2.el9_6 needed by kmod-nvidia-580.82.09-1.1.el9_6.elrepo.x86_64
uname -r: 5.14.0-570.39.1.el9_6.x86_64
You will have to wait until that package is released by the Rocky team since that kernel hasn’t appeared yet. Should be soon.
Yes, I understand. This driver is from the el-repo testing branch, and I wanted to test it on the current kernel.
rpm -qa | grep nvidia
nvidia-x11-drv-libs-580.82.07-1.el9_6.elrepo.x86_64
kmod-nvidia-580.82.07-1.1.el9_6.elrepo.x86_64
nvidia-x11-drv-580.82.07-1.el9_6.elrepo.x86_64
It looks like from the info from DNF that the module is only allowed to work on 5.14.0-570.42.2.el9_6 and higher. So doesn’t look like the kmod will work against earlier kernels.
You probably will have to contact and ask elrepo about it, if they will build for earlier kernels or not.
ELRepo tends to build modules with “kABI-something”, that is a module that should work with every 5.14.0-570*.el9_6 kernel. Previous build (of same module) was for 5.14.0-503*.el9_5 kernel, and next build for 5.14.0-???*.el9_7 kernels (if the version of driver is still built for el9_7).
A more strict kernel version dependency could be … @tqhoang should be able to explain why.
Could be restricted at the minute due to that maybe?
I thought it was because it was only in elrepo-testing. I installed the previous driver from there.
Probably the reason it’s still in testing is because it’s not stable yet, and so in terms of being able to use it, limited into the kernel it’s being built for.
No real sense building it for every kernel, when it’s unstable. Build for one, when stability improves, then build for all kernels. Less work and effort required.
My other computer has Fedora 42 and rpm-fusion drivers installed.
It just updated to 580.82.09, and it’s started acting weird. It’s no longer smooth in Wayland, and the NVIDIA settings are showing some weird behavior. Before that, I had 580.82.07 drivers, and everything was fine.
Rocky 9 seems to be fairly delayed with that kernel for some reason. You can see that it was built back on 9/16 and was supposed to be published.
I had already been running that kernel under RHEL 9 and the “other” EL9 had also published it on 9/15.
For the interim, you’ll have to boot to the previous kernel (-570.37.1.el9_6 ???). Hopefully it’ll all resolve itself in a few days.
I’ll wait for the new kernel and hope this driver on RL9 won’t work as badly as on F42.
P.S. I’m a bit confused. Yesterday I turned off my computer with F42, and now I’ve noticed everything is working fine. I don’t understand anything anymore, and the picture is fine. I rebooted the computer twice yesterday. How is this possible?
I haven’t seen where the nvidia-settings is bugged like that under RHEL 9, but I do notice issues with monitor detection after I’m away for a while (using a KVM). Sometimes toggling to a console (CTRL+ALT+F2…) and back to terminal one (CTRL+ALT+F1) helps.
Also I noticed that the updated Rocky 9 kernel is finally available.
Yes, the kernel arrived. I decided to test it.
I don’t even want to think about how that 570 driver (not a test) updated to the new kernel 
