This is a discussion topic for the following Common Issue:
You can discuss the problem and its solutions here, but please note that debugging and technical feedback should primarily go to the issue trackers (e.g. Bugzilla) linked in the Common Issue, because that’s the place that developers watch, not here.
If there are any updates/changes/amendments for the Common Issue description, which you believe should be performed, please post it here.
Please see the Common Issue for solution/workarounds:
The Discuss pop-up says I need to start a new topic and not reply here if the solution isn’t working for me, but the OP is clear that this is the discussion topic for the listed issue. I will defer to the OP rather than the site pop-up.
Thanks for the post, I appreciate that even these old GPUs are getting some support. The fix looks like it won’t be quite so simple for me with my GeForce GTX 1050 Ti Mobile though. When I run the swap command I get the following, even if I run in a tty terminal session. Do I need to run in a live session or something? Sorry I’m not great at this stuff.
Running transaction
Transaction failed: Rpm transaction failed.
- file /usr/lib64/libcuda.so from install of xorg-x11-drv-nvidia-580xx-cuda-libs-3:580.159.03-1.fc44.x86_64 conflicts with file from package xorg-x11-drv-nvidia-cuda-libs-3:595.71.05-1.fc44.x86_64
- file /usr/lib64/libcuda.so.1 from install of xorg-x11-drv-nvidia-580xx-cuda-libs-3:580.159.03-1.fc44.x86_64 conflicts with file from package xorg-x11-drv-nvidia-cuda-libs-3:595.71.05-1.fc44.x86_64
[etc., I’ve truncated the further output as it didn’t seem to add much, just continued conflicts with various libs-3:595.71.05-1.fc44 packages.]
A bit disappointing they’re (NVIDIA) dropping support for these so soon. They’re not even a decade old yet and are still solid budget GPUs.
Strange. Can you try removing the current packages and installing the 580xx ones separately?
sudo dnf remove xorg-x11-drv-nvidia akmod-nvidia
sudo dnf install xorg-x11-drv-nvidia-580xx akmod-nvidia-580xx
Wait a bit for the driver kmod to be built and reboot.
@leigh123linux any idea why the swap option is not working? I saw it work for others before.
Still officially supported, on the 580xx LTSB aka legacy driver, for a couple years.
Actually Pascal gen cards were released exactly 10 years ago (May 27, 2016).
Is there any better recommendation than “wait a bit”? For example running pgrep -a something until something disappears?
Yes: pgrep -a akmods covers all the cases:
Thanks. Can you please prepare an exact text snippet that I can include in the common issue description? I can write something myself blindly, but I can’t verify that it actually works.
Yes - watch -n 2 tail /var/cache/nvidia/.last.log.
You want to see something like:
Running transaction
[1/3] Verify package files 100% | 35.0 B/s | 1.0 B | 00m00s
[2/3] Prepare transaction 100% | 1.0 B/s | 1.0 B | 00m01s
[3/3] Installing kmod-nvidia-7.0.10-200 100% | 1.1 MiB/s | 10.5 MiB | 00m10s
Warning: skipped OpenPGP checks for 1 package from repository: @commandline
Complete!
2026/05/27 09:45:29 akmods: Successful.
Somethink like replacing:
Wait for akmods to finish building the driver kmod and reboot.
by
Wait for akmods to finish building the driver kmod, for example with:
watch pgrep -a akmodsinterrupt this watch with ^C when it shows no akmods process, then reboot.
In this case, it should be /var/cache/akmods/nvidia-580xx/.last.log
Also, if akmods build for 2 kernels, you should watch 2 times this log.
I did the above.
Then I did the above (and it took longer than I thought it would as it had to build for multiple kernels, so that was helpful! Normally I do just “wait a bit,” but I’ll have to keep pgrep in mind going forward).
And it worked! Thank you all, I appreciate it! I’m not sure why the swap command didn’t work. I could certainly paste the full output of the swap command that didn’t work, but I think what I’d pasted above was the relevant bit.
I did end up also installing xorg-x11-drv-nvidia-580xx-cuda so I get the nvidia-smi command that I think I’d had before, but now everything’s working again!
I’ll have to defer to your info - I’m still using the 595 driver, but I do have to switch my wifes machine over to the 580 range, as she’s rocking my old GTX 980 or something like that,
A simple way that works for me is open system monitor and and watch the cpu as while the drivers are building the cores will pretty much max out.Once the driver has built the cores will drop back to normal.This also works with a new kernel as the drivers will be built to the new kernel.I have been doing that for years and never had any issue when rebooting after an update with nvidia or the kernel.
Used to do this, - watched the compilation complete and when I rebooted the modules were built again. When I looked at the failure log, it failed to install the package as the rpm database was in use.
Now, I double-check that the compilation finished AND the rpm work didn’t soil the bed for any reason; .last.log is the easiest way I’ve found to ensure both complete successfully.
I’ve updated the description (relying on processes rather than watching the log file, which might get confusing). Thanks.
Just be aware - pgrep check would have completed and the drivers would still not be installed:
Total size of inbound packages is 8 MiB. Need to download 0 B.
After this operation, 36 MiB extra will be used (install 36 MiB, remove 0 B).
Running transaction
Transaction failed: Failed to obtain rpm transaction lock. Another transaction is in progress.
Warning: skipped OpenPGP checks for 1 package from repository: @commandline
2026/02/20 11:40:51 akmods: Could not install newly built RPMs. You can find them and the logfile in
2026/02/20 11:40:51 akmods: /var/cache/akmods/nvidia/580.119.02-1-for-6.18.12-200.fc43.x86_64.failed.log
The instruction was supposed to cover finishing the transaction, not inspecting it fully. But sure, I can add a hint for the log as well. But what path is it then? A couple lines above it says /var/cache/akmods/nvidia-580xx/.last.log while your output contains /var/cache/akmods/nvidia/580....failed.log
This log output is from 580 drivers when 580 drivers where the latest thing, 20th Feb 2026.
The path for the logfile for me back then was (and still is) /var/cache/akmods/nvidia/.
However if the user is specifically installing the 580xx package, then according to @francismontagnac the path would be /var/cache/akmods/nvidia-580xx/, presumably because it’s a separate package.