My monitor often cannot turn on again after screen goes off due to inactivity on Linux. Manual DisplayPort switch helps, but this is annoying as hell and wears out ports. Is that it? I use exactly DisplayPort with DSC.
Problem persists on all drivers I tried 550, 555, 560, 565.
The latest stable driver 560.35.03 was released in August 21 (3 months ago) and there is no new driver since on Linux as the 565.57.01 is a beta driver.
I have a similar issue with a MSI QHD 144Hz monitor and I found the issue is caused by the monitor which randomly downgrades the DisplayPort link. On Windows it fallbacks to 120Hz YUV 420. I have to turn off and turn on again the monitor to work around the issue.
Edit: I donāt have this issue with my other monitors.
I guess in my case it could be also PCI riser cable instability, but on Windows I donāt notice any issues.
560.x is not a stable driver, itās a ānew feature branchā thatās (as it says) for testing new features, and these are typically short lived and die quick.
The current latest stable (or production branch as nvidia calls it), is 550.135 which was just released barely yesterday and it includes the fix for kernel 6.12.0.
Edit:
535.x is also production branch and still getting releases (can be useful if hit by regressions in 550 for now).
But of course, many major wayland improvements are in 560+, so if you want wayland youāll likely want the 565 beta for now and not the ancient 560 NFB.
If donāt want betas and wayland is not working out w/ 550, then Iād recommend to stick to Xorg for now and wait.
560.x is not a stable driver, itās a ānew feature branchā thatās (as it says) for testing new features, and these are typically short lived and die quick.
I donāt agree with this statement as a short lived driver doesnāt mean that it is not stable. If you look at kernel.org, Linux 6.11.x kernel are short lived but 6.11.9 is clearly marked as stable.
550.x and 535.x drivers are the equivalents of LTS kernels such as 6.6.x, 6.1.x, ā¦
How kernel.org does thing has little to do with how NVIDIA does it.
NVIDIA does not list 560.x as a āstableā driver and has never ārecommendedā it either. If you search for drivers, youāll see 3 tabs, Production/Recommended, NFB, and Beta. 560 is listed under NFB.
What is a New Feature Branch Driver?
This driver provides early adopters and bleeding edge developers access to the latest driver features before they are integrated into the Production Branches.
Formerly known as Linux Short Lived Branch (SLB).
Regardless of how you see it, the 560 branch is dead, meaning either need to go back to the production branch or the newer beta.
If 560.x is not stable, does that mean every 560.x Windows users are using a non stable driver?
Edit: You might have right and if thatās the case I think the Nvidia branch schema is a bit confusing since some stable Linux distros such as Fedora are using Nvidia drivers from new feature branch.
@nagnibeda.sn good news on the external-monitor-plugged-to-HDMI front: the nice folks responsible for RPMFusion packages have released a patched akmod-nvidia RPM that fixes the problem with latest 6.12 kernel from Rawhide:
āÆ uname -r
6.12.0-65.fc42.x86_64
āÆ modinfo -F version nvidia
565.57.01
āÆ rpm -q akmod-nvidia
akmod-nvidia-565.57.01-2.fc41.x86_64
This is specific to Fedora, but at least proves a fix is available and should be widely available pretty soon.
Now I will finally be able to test suspend/resume on kernel 6.12, but itās definitely a step in the right direction, weāre making progress! \o/ Thanks to all who contributed.
fwiw, in Gentoo, we currently do it like this:
And currently āhopingā that 565.x will become the next production branch after itās out of beta so we can make it more widely available in the stable branch with less risks of exposing users to major WIP regressions that they didnāt voluntary opt-in (it may instead become the next NFB though, donāt know NVIDIAās plans for this beta).
As for Windows, drivers use different versioning/branches and I wouldnāt compare it with the linux model.
Anyhow, this is kind of of off-topic for this thread so I donāt plan to discuss this further here.
Agreed. Itās easy to get angry at a multi-billion dollar company, and we (users) are in our right to demand proper support for the hardware they sell, but IMHO we have to acknowledge Nvidia has been stepping up their game when it comes to Linux. Besides old-timers from Nvidia staff that have been here sinceā¦ forever?, I see a lot of new Nvidia devs participating on theses threads, bugs being addressed and so on. And, as @luisalvaradox has stated, they are asking for feedback, and they are listening.
Linux is a very complex ecosystem, with all the distros, flavors, kernel versions, the still ongoing evolution of Wayland, the eternal opensource vs. proprietary struggle, and so on. But the simple fact that many bugs being reported here are related to non-native AAA games running on Linux shows the landscape has changed.
Having said that, I also feel frustrated when ābasicā (from the daily usage standpoint) features, like plugging an external monitor or suspending your laptop, break. Sometimes these problems takes weeks or months to be fixed. But maybe thatās because they are complex issues, and have been caused by external factors like changes in the kernel. So, I try my best to report objectively, provide all the info I can, and engage in productive conversations.
Iām on Manjaro and Iām using kernel 6.12 and 565 beta driver, I donāt know why you have this problem on your distro
What exactly patch did you use and with what patch command? Also you see, same driver version. Maybe left overs of same non patched driver remains and nvidia doesnāt clean up after itself. I do not know force commands to rebuild all
Iāve done absolutely nothing to make it work, Iāve just updated the system and installed the new kernel 6.12 the day after it was released as stable, and it worked.
On Manjaro there is a nice GUI that automatically installs linux612 & linux612-nvidia, so maybe thatās what youāre missing
The closed version can be patched, using nvidia-all in my case. Adding/tweaking the patch noted above, replacing nvidia-open with nvidia-dkms, and adding the patch/md5sum to the pkgbuild. Now running 6.12 with the closed module.
Seems can not be used with latest beta. And as I understand I can not use open driver too. Cause my gpu is NVidia Quadro P3000 mobile. Newer version of gpu needed for open driver.
I donāt understand why NVidia canāt just take themselves current beta patch it and upload changing .01 in the end to .02 (565.71.02) for example. And everyone would be happy.
Forza Horizon 4 crashes when pressing continue and loading into the game.:
nvidia-bug-report from repro state:
nvidia-bug-report.log.gz (592.6 KB)
steam launch options:
export DISABLE_GAMESCOPE_WSI=1 MANGOHUD=1 OBS_VKCAPTURE=1; gamemoderun gamescope -f -W 2560 -H 1440 -r 170 -- %command%
gamescope has to be used on this game to work around black bars on right and bottom, that gamescope wsi env var must also be used to avoid a freeze in the bootup with gamescope
disabling gamescope wsi wouldnāt even be needed if it wasnāt for a different nvidia bug: Forza Horizon 4 (1293830) Ā· Issue #4667 Ā· ValveSoftware/Proton Ā· GitHub
Hello everyone, Iām on Nobara kernel 6.11.8-200, drivers 565.57.01
hardware: r7 3700x, 32g, Rtx 3080 10gb.
I have 2 monitors, since a few weeks ago I get weird freezes, both or one of the screens totally freeze but the computer its clearly still running (with audio and everything), every time I have to restart.
There is any way to fix or limit it? Sometimes it happens once per day, sometimes all I do is restarting.
There is a log right after one of the million freezes:
Hope it gets fixed soon.
I know that fbdev being enabled was causing similar issues, but I thought that was fixed in the last beta series. With that said it is still experimental so you could try disabling it.
Greetings distinguished gentleman and gentlewomen,
TK-Glitch just released a patched version, for the 6.12 kernel.
IwanIV
