NVIDIA driver fails to install in Fedora 34

I have no idea what happened, but my updates repo was disabled (did I ever have it enabled??). The 470 is here:

http://download1.rpmfusion.org/nonfree/fedora/updates/34/x86_64/a/

So probably I just had the rpmfusion-nonfree-updates disabled, didn’t find the new kernel-devel package, and thus failed to install the new driver.

Sorry, it was all my fault.

Hello Luis.

I have had the same problem since Nvidia updated its driver from 465 to 470. I asked for help at the Fedora’s official Matrix channel and they told me to run “nividia-bug-report.sh” (included with the package from rpmfusion) and send the log to forums.developer.nvidia.com or linux-bugs@nvidia.com:

If you visit the forum mentioned above, you’ll see that the bug has been already reported by others. There was no answer by Nvidia and I don’t think that anybody is really expecting it, considering the contemptuous treatment that Linux users suffer from on it’s behalf.

There’s nothing we can do, but wait for NVIDIA’s propietary driver version 470.74 release. According to (see the following link), it’s a bug of version 470.63.01, which is the current version at rpmfusion. Kernel parameter “nvidia-drm.modeset=1” is causing this bug.

Press “e” key a grub screen to edit kernel parameters, remove it and specify runlevel 3 in order to boot in a console, something like this:

rd.driver.blacklist=nouveau modprobe.blacklist=nouveau rhgb quiet 3

I often remove parameters “rhgb” and “quiet” as well. Log in and type

lsmod | grep nvidia

You’ll see that nvidia driver is loaded but, as soon as you attempt to start an XOrg session by typing “startx”, the driver will crash and the desktop won’t be displayed.

Regards.

The link:

https://www.gamingonlinux.com/2021/09/nvidia-47074-for-linux-is-out-fixing-up-memory-usage-for-direct3d-12-with-vkd3d-proton

You just were lucky. Tomorrow, you won’t. I know by my own experience.

:D:D:D

Just so everyone is aware, the newly released 470.74 driver is available on rpmfusion. I just installed it on my fedora 35 system.

Actually, I had the opposite problem. Nouveau is crashing, nvidia drivers 470.63.01 are working fine (so far!). That’s why I changed drivers.

Where did you get them? I don’t see them here:

http://download1.rpmfusion.org/nonfree/fedora/nvidia-driver/34/x86_64/a/
http://download1.rpmfusion.org/nonfree/fedora/nvidia-driver/35/x86_64/a/
http://download1.rpmfusion.org/nonfree/fedora/updates/34/x86_64/a/

EDIT: Sorry, just found it in the testing repo.

If you search NVidia site for installing RTX A6000 GPU, one of the latest high end devices on the market, you get information about setup on linux invoking Fedora and Red Hat with specific details for Suse. NVidia hardware is used at gigantic scale for cognitives services using exclusively linux on servers connected to SaaS services (Microsoft Azure, IBM Cloud, … Nouveau drivers are built with NVidia support for open source enthusiasts, don’t give everything but …
NVidia/PNY serve several markets with ‘gamer’ using ‘latest’. Search NVidia/PNY sites about details as support lifespan isn’t fantastic but they do advise to prefer distribution drivers specify specific driver for specif hardware, just follow …

Yes sorry, I read to fast and misunderstood your explanation.

NVidia drivers working fine, crashes eliminated.

Yet, there is a small issue: a couple of errors at boot time:

Sep 29 17:33:05 Orion systemd-udevd[623]: nvidia: Process '/usr/bin/bash -c '/usr/bin/mknod -Z -m 666 /dev/nvidiactl c 195 255'' failed with exit code 1.
Sep 29 17:33:08 Orion systemd-udevd[623]: nvidia: Process '/usr/bin/bash -c 'for i in $(cat /proc/driver/nvidia/gpus/*/information | grep Minor | cut -d \  -f 4); do /usr/bin/mknod -Z -m 666 /dev/nvidia${i} c 195 ${i}; done'' failed with exit code 1.
Sep 29 17:33:08 Orion alsactl[2122]: alsa-lib parser.c:242:(error_node) UCM is not supported for this HDA model (HDA NVidia at 0xf7080000 irq 17)

I didn’t detect any working problems, though.