How to install nvidia proprietary driver?

Right now, I have the open driver and I crash regularly after the resume with the error:

kernel: NVRM: RmHandleDNotifierEvent: RmHandleDNotifierEvent: Failed to handle ACPI D-Notifier event, status=0x11

And it looks like I’m not the only one. I have found on the nvidia forums, that this happens to a lot of other people with Dells (mine is). The temporary solution until nvidia fixes it is to switch to the proprietary driver.

I will be reinstalling my system. I have looked at: https://sndirsch.github.io/nvidia/2025/07/16/nvidia-drivers.html and the cuda repo but only talks about installing the open driver.

Which driver do I select for G07 proprietary or do I need to download the .run from nvidia?

Thanks

There are no rpm packed proprietary kernel modules for the G07 series. So you need the run file when you believe that the proprietary kernel module will solve anything.

The cuda repository is using the 610 series driver, you don’t indicate what one your using or your GPU…

Please post the output from inxi -GSaz

Hi,
I am a long time Opensuse user , a question

Is there any reason Slowroll repo does not get current nvidia driver let’s say 615 series or other fairly current ones?

The idea of a rolling release is to be quite close to current. I is because nvidia does not allow current kernels that arrive to slowroll quite fast and nvidia lags behind in making the drivers compatible with current kernels due to testing?

Slowroll ships the same latest (WHQL in Windows terms) drivers like Tumbleweed or Leap.
Latest stable upstream driver is 595.99.02 and this driver is available in openSUSE products.

This discussion was already held several times. For openSUSE products, the maintainer decided to use the stable (also rolling and WHQL in Windows terms) driver branch and not an experimental (SLB) one. Else you would get issues with stable openSUSE products like Leap.

If you want the experimental (SLB) drivers, use the CUDA repo or the run file.

1 Like

@elfroggio if you will use the .run driver, just an FYI.
I tried the new 615.71.09.run and broke the suspend mode.
systemctl suspend didn’t work, so I am back to using 610.57.04.run driver.
This is on the latest tumbleweed kernel.

@conram suspend is working fine on GNOME with the 615.71.09 series packages from the cuda repository and 7.2.3 kernel :wink:

1 Like

Is the CUDA method reliable? I switch between the run file and meta package but I prefer being on the bleeding edge for NVIDIA (I do use the run file inside distroboxes though when I’ve been testing things out).

If you don’t need 32bit, yes. Make sure to add the dkms package from the repo as well.

Oh there’s no 32 bit support? I didn’t realise, I’m using Steam but I could try the RT3 client with WOW64 and see if it works.

Correct, not from the cuda repository…

Ah that’s a shame. Maybe I’ll stick with the .run.

It’s working on the new tumbleweed kernel that was release today, kernel:7.2.4-1
suspend now works here.

You can install the i586 version of Steam from the games tools repo - https://download.opensuse.org/repositories/games:/tools/

Seems to works fine with wow64 for all games I’ve checked using Nvidia 615.71.09-1 driver from cuda repo.

Could possibly alternatively install flatpak version of Steam which people seem to complain about if I recall. It’d take some tweaking so I didn’t do that for now.

I’m not sure how that would help? sorry!

Oh yeah I think I’d prefer to give that a miss, I know Steam don’t verify the flatpak for some reason, I’d go Distrobox over that myself as Distrobox has worked great for me. But at that point I think I’d be better off just using the .run file for all the hassle :laughing:

Oh there’s no 32 bit support? I didn’t realise, I’m using Steam

It might help if you want to run Steam with the Nvidia driver from the Cuda repo without 32 bit libraries. That’s what I’m doing on my x86_64 machine because I don’t want to manually install the .run file every time it updates. Seems to work well on my system. Not noticing any issues with performance in games.

I’d go Distrobox over that myself

I used to do similar stuff with Docker and yes it is a hassle sometimes. Not sure if you can have a mismatch in the Nvidia kernel mod between the container and the host system but maybe it just needs to be the same driver version. Probably better going with the .run file as you say if you can’t get Steam to install with the Cuda driver or you just don’t think that’s a good idea.

Isn’t the i586 version of Steam 32bit only though?

Maybe I don’t understand distrobox correctly but there is no Kernel mod on the distrobox side, it uses the host kernel, so you just need the userspace 32bit libs that match the 64 ones on the host machine, as far as I know.

Yes, it is but you can install it anyway whereas the 64 bit version from ‘repo-non-oss’ will give you dependency issues that aren’t resolvable, at least didn’t appear to be on my system. Both older 32 bit games and newer ones seem to run fine for me with the i586 Steam client installed.

there is no Kernel mod on the distrobox side

Not sure how Distrobox handles it since it can use Docker or Podman to create containers and my only experience was with Docker. I was assuming you would need to install the 32 bit Nvidia libraries from a repo other than Cuda in the container and maybe that’s fine. It’s been a while since I dealt with it.

At least the i586 version is compiled on i586 architecture so it may be missing some optimisations but it should run 64 bit games without issue as far as I understand. I don’t notice any performance degradation.