Ah that’s why, makes sense. The sources have now appeared on Github, and I could compile it too.
Indeed it is not completely fixed, but it depends on settings (what’s your screen and settings?).
I use an LG 32gs95ue 3440x1440 monitor, connected via DP to a 5080 from Palit. OS is Archlinux with Plasma 6.
Waking up works only up to 144 Hz. At 240 Hz screen does not wake up ; in that case the TTL switching trick is still needed. Adaptive Sync being on of off (VRR) does not make a difference - the key difference is the refresh rate.
I join my bug report dump to this post.
nvidia-bug-report.log.gz (975.4 KB)
@amrits I’m happy to report that the XID crash running Cyberpunk 2077 with Path Tracing enabled no longer occurs for me since updating to NVIDIA driver 580.105.08. So may have been a unique issue to 580.95.05
Bug #5626879, unless others are still seeing this issue, likely is good to close out
@Dekatria you had the same issue as me
1 Like
I’m still seeing flickering on Resident Evil 4 (2023) with the latest (580.105.08)
Can anyone confirm this bug and that it isn’t just me as I’ve been pointing it out since the start of the 580 branch.
With 580.95.05, Xorg’s log was reporting 600 MHz for a monitor connected through hdmi on a GTX 1070:
[ 10.560] (–) NVIDIA(GPU-0): Asustek Computer Inc VG245 (DFP-3): 600.0 MHz maximum pixel clock
[ 10.560] (–) NVIDIA(GPU-0): EDID maximum pixel clock : 250.0 MHz
With 580.105.08, log now shows it dropped to 165 MHz:
[ 2972.907] (–) NVIDIA(GPU-0): Asustek Computer Inc VG245 (DFP-3): 165.0 MHz maximum pixel clock
[ 2972.907] (–) NVIDIA(GPU-0): EDID maximum pixel clock : 250.0 MHz
And a 220.6 MHz mode which used to be fine now isn’t:
[ 2972.948] (WW) NVIDIA(GPU-0): Validating Mode “1920x1080_75”:
[ 2972.948] (WW) NVIDIA(GPU-0): Mode Source: EDID
[ 2972.948] (WW) NVIDIA(GPU-0): 1920 x 1080 @ 75 Hz
[ 2972.948] (WW) NVIDIA(GPU-0): Pixel Clock : 220.64 MHz
[ 2972.948] (WW) NVIDIA(GPU-0): HRes, HSyncStart : 1920, 2056
[ 2972.948] (WW) NVIDIA(GPU-0): HSyncEnd, HTotal : 2264, 2608
[ 2972.948] (WW) NVIDIA(GPU-0): VRes, VSyncStart : 1080, 1081
[ 2972.948] (WW) NVIDIA(GPU-0): VSyncEnd, VTotal : 1084, 1128
[ 2972.948] (WW) NVIDIA(GPU-0): Sync Polarity : -H +V
[ 2972.948] (WW) NVIDIA(GPU-0): Mode is rejected: PixelClock (220.6 MHz) too high for
[ 2972.948] (WW) NVIDIA(GPU-0): Display Device (Max: 165.0 MHz).
[ 2972.948] (WW) NVIDIA(GPU-0): GPU extended capability check failed.
[ 2972.948] (WW) NVIDIA(GPU-0): Mode “1920x1080_75” is invalid.
Forces me back to using 60fps, what’s going on?
Edit: and sorry for the lack of nvidia-bug-report.log, will not be sharing this – can answer specific questions or test potential workarounds if needed though (fwiw tried to drop most ModeValidation checks but it ultimately instead fails with Unable to construct hardware-specific mode timings which I guess is not surprising, validation isn’t the issue but rather that 165 MHz)
Latest release is missing aarch64, i.e., https://download.nvidia.com/XFree86/Linux-aarch64/580.105.08/ does not exist yet.
This is blocking rollout of the update on NixOS.
Same issue here.. Driver 580.105.08 only shows 1920x1080, instead of 2560x1080.
GeForce RTX 3070, HDMI/DVI connect, Fedora 43, Kernel 6.17.7 (kernel.org), KDE/Plasma 6.5.1.
Reverting to 580.95.05 fixes the problem.
1 Like
If I were to take a guess, it’s probably the same that I also described above where it seems to think it has a lesser connection to the monitor, and it ends up thinking it’s unable to display these larger resolutions or with higher framerate.
I tried the demo and didn’t observe anything with a 4060. I had to use Proton 9-7 otherwise the game would just crash using experimental. Maybe showing a video of what it looks like and where would help?
1 Like
It sounds like DSC related code is broken. FWIW, 4k 144hz seems to work fine.
Black screens are still happening, by the way. VRR flickering still happens too.
Hi @BlueGoliath
Could you please help to capture nvidia bug report whenever you hit either of below issues.
1) Random display related errors with Xid 56
2) HDMI FRL link training failed errors
3) [drm:nv_drm_gem_alloc_nvkms_memory_ioctl [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000c00] Failed to allocate NVKMS memory for GEM object
@vrachatte @aplattner @amrits this is still not fixed in the latest release 580.105.08
this application is still hitting a FPE signal 8 when computing using the 580.x drivers. it does work properly with 575 and older drivers.
please clarify if you mean “fixed” as in you deployed the fix into some driver, or “fixed” means it’s fixed internally and waiting to be merged into the master code still. as of right now, the issue is NOT fixed for the user.
internal bug #5497358, issue when running the Einstein@home O4MDG/O3AS application with r580 drivers.
2 Likes
Greetings, and thanks for keeping us up to date.
It is possible to check the status of internal tracking issue 5556719? It is discussed in posts Fd leak with explicit sync and kde plasma - #30 by renari and following.
It results in sync_file descriptors being leaked every time a KDE taskbar tooltip appears, resulting in a crash and loss of all applications in normal work use for me. It may only happen with nVidia explicit sync enabled.
NVIDIA-Linux-aarch64-580.105.08.run is missing from Index of /XFree86/Linux-aarch64
leigh@mpd-pc:~/Desktop/packaging/xorg-x11-drv-nvidia$ spectool -g xorg-x11-drv-nvidia.spec
Downloading: https://us.download.nvidia.com/XFree86/Linux-x86_64/580.105.08/NVIDIA-Linux-x86_64-580.105.08.run
File '/home/leigh/Desktop/packaging/xorg-x11-drv-nvidia/NVIDIA-Linux-x86_64-580.105.08.run' already present.
Downloading: https://us.download.nvidia.com/XFree86/aarch64/580.105.08/NVIDIA-Linux-aarch64-580.105.08.run
Download failed:
404 Client Error: Not Found for url: https://us.download.nvidia.com/XFree86/aarch64/580.105.08/NVIDIA-Linux-aarch64-580.105.08.run
1 Like
I see something strange: when playing a Youtube video in full screen in Firefox, then I get sometimes a black screen for 1 second before the screen reappears. The screen (LG 32gs95ue 3440x1440 monitor, connected via DP to a 5080 from Palit) really went to some off mode, since the HDR logo reappears on the top, as if there was some mode switch for a second. Anyone seeing this?
This also happens with mpv, but then no image happens.
Edit: I only see the problem in 120 Hz or 144 Hz. If I go back to 240 Hz then it works fine! But then I lose the wake up capability…
Edit 2: I just switch off HDR in KDE, then switched it back on and now mpv seems to work fine. I am puzzled.
Edit 3: ok this problem only happens after waking up the screen. Then mpv full screen is black, until changing image settings or restarting. Could be a bug in Plasma/Wayland actually.
Edit 4: it is caused by the adaptive sync in Plasma. I disabled it completely for the time being. I never noticed it before upgrading to the new driver 580.105.08 though, but it may be a coincidence.
I usually don’t see any of those when the screen goes black even if the driver recovers. I do have both 2 and 3 errors in dmesg on the latest driver.
It seems like The Finals triggers this the most but it’s so random. I’ve gone a week without it happening and sometimes multiple times in a day.
Funnily if I use steam screen recording the output file doesn’t have any flickering hence mobile phone footage.
I’ve tried multiple versions of proton all the same. I can’t change any graphical settings as you can see in the video. There doesn’t appear to be a way to do it in the text files as Steam puts back the old ones. Even deleting the the proton/wine prefix didn’t help.
No flickering though on the 575 series drivers
Yes, I am also see this bug with 580.105.08 drivers, I have also tried proton-ge wine-wayland, but it still flickering
nvidia-bug-report.log.gz (2.1 MB)
Definitely didn’t see that with the demo on my system. Does it happen with the demo too? Do you have VRR enabled?