The first time I had this exact problem I was using proprietary nvidia driver, which I had always been since I did not even know they have a thing called open driver with comparable gaming performance.
AI chatbot helped to diagnose and find the mismatch issue of nvidia driver, and suggested me to switch to open driver. But AI was wrong to think that open driver will get rid of this problem.
ETA is about a week or something?
@moshpuke and what Nvidia GPU do you have? Can you post the output from inxi -GSaz please?
Nothing special, nvidia 4080.
Not sure that’s dependent on SUSE → Nvidia → Publish
I see it is built… https://build.opensuse.org/package/show/X11:Drivers:Video:Redesign/nvidia-open-driver-G07-signed
Well you should be fine sticking with the G06 series…
Yes, similar problems have been reported here for years. Unfortunately, nobody made efforts to troubleshoot it and find out what happens (or why it does not happen). And for any problem to be fixed this problem needs to be identified first. “It did not work” is not enough.
There where some steps done to prevent future breakage between kernel version and open kmp:
https://bugzilla.opensuse.org/show_bug.cgi?id=1254569
That bug says fixed, but even when I dup now Nvidia still has issues and I have to rollback and use the previous kernel.
for now:
sudo zypper al nvidia*G07*
This problem is fixed, we are just waiting for the package update to pass QA. So a better option is to just wait a little while. If you add the lock above, be sure to remove it when the package does update, with
sudo zypper rl nvidia*G07*
I find this false accusation mildly disrespectful both to myself, but mostly, to Stefan, the maintainer of the nvidia drivers, who tirelessly chases bugs for us, and the rest of the suse team who take care of things like integration with the kernel etc. Maybe you should come and join us on bugzilla if something goes wrong, and you can add to the considerable efforts made to troubleshoot it.
There are two mechanisms which have been recently added, which should resolve all the issues we are familiar with. Firstly, the kernel module has been made a critical part of kernel QA, so this will keep the module in sync with our kernel. Secondly, a meta package has been created, which will keep the userspace drivers in sync with the kernel module.
For kernel 6.18, there was a slip-up with the first part which has been fixed by the QA team the very first time it was noticed, so we should not see it again. For the brand new G07 drivers, a minor mistake occurred on their very first deployment, which caused them to be selected by zypper over the G06 drivers, and the fix for that was put in place within minutes of being identified (workaround at the top of this post)
I think it’s perfectly valid to just sit and chill and wait for someone else to fix it, or, to join us while we try to fix it for those people, but Stefan works really hard to get this driver working as best he can for all of us, as do I, and lots of other people at suse, and even at nvidia, and it really bothers me to see accusations to the contrary coming from people who weren’t helping us do that.
If we’re going to make accusations about people not doing the right thing to solve this: This failure with the kernel module for 6.18 which I mentioned above, had users crying about it here and on reddit for SIX DAYS and NOT ONE person bothered to report the bug so that it could be fixed.
I personally didn’t “cry” I posted about it as soon as I saw it (yesterday) to determine whether it was “just me” or not and then was informed Stefan knew about it and a fix was in the works … thank you for your attention to this issue 
We all appreciate the hard work done by opensuse team, and contributors. Tumbleweed, Leap are beautiful distros, and that’s why we try to make it work. I started this thread, as i am new to opensuse ecosystem, and bug reporting. 
But i have spent days trouble shooting , reinstalling, testing multiple kernels, and driver versions.
From my recent experience with Leap and Tumbleweed, the issue with nvidia drivers is the lack of sync between kernel, meta package, driver component updates. Once the release of all these packages is synchronised, this issue will be sorted out.
Hopefully from next year we will have Tumbleweed and nvidia drivers working smoothly.
Installed the latest 20251211 snapshot, and nvidia open driver G06 580.105.08. Everything is working fine now. Big relief.
But when i tried to update the driver, it installed components of unmatched versions, and it didn’t work. So, i had to roll back, and stay on G06 580.105.08
It shows two problems
Both need bug reports.
Just came here to post what I found out to be causing my problem:
After I had snapper rolled back to fix my dup problem, and used the machine for two days, I tried another dup and checked what had been updated this time before rebooting. I noticed that G07 version was installed instead of G06, which should be the correct package to match all other G06 nvidia packages on the system.
Suggestion, manually forbid all G07 packages at the moment to stay safe unless you know what you are doing.
Edit: forgot to say that I manually uninstalled G07 and installed G06 with the matching kernel version, then the machine reboot fine.
Seems to be the key here … no matter … check your DUP … remove G07 … install G06 for now
So, nvidia-open-driver-G06-signed-kmp-default with 580.105.08_k6.18.0_2-3.1 is working. other components are in sync.
Other components have a new verison 580.119.02-21.1 available, but kmp is still not updated.
G07 driver is not working.
As far as I know correct … I do it the “hard way” so I don’t have these problems … maybe others will chime in with their observations
csrinivas - thanks for confirming the correct match for the nvidia open driver. I was trying different combinations - without sucess - and in the end just installed the closed driver to get a working system (with nividia graphics).
I can confirm that I removed the G07 package (nvidia-open-driver-G07-signed-kmp-default) and then manually installed G06 variant and added a lock as directed by pallaswept, and everything is now working as expected.
How do we know when the package update has passed QA and the lock can safely be removed?
Thanks for everyones help.
I think it will show up in Myrlyn, and reflect in the version details.
If not, update after a month, and if it fails, roll back in snapper.
Or just wait until present driver fails, and then update.
