Black Screen
nvidia-bug-report.log_v580-76-05_CachyOS_2025-08-15.gz (645.9 KB)
log_v580-76-05_CachyOS_2025-08-15_01.txt (39.4 KB)
Oddly enough, my setup is wayland-only and everything is working fine so far (regular desktop use, no gaming).
The graphics still crashes for me when trying to enable HDR: Graphics driver hangs when enabling HDR
LFC is still broken: Wildly fluctuating refresh rate and flicker at low framerates , Gsync compatible VRR - Low Framerate Compensation (LFC) transition not seamless, unlike on Windows
Looks like certain applications are freezing upon rendering the first frame. Here is an issue on the Zed repo.
Much love to the developers, but man, can ya’ll get some of that sweet AI money to help with these drivers? I can only imagine the spaghetti code they must consist of, considering every release fixes 5 things but breaks 10.
Hello, I’m getting repeated crashes in Elite Dangerous since 575, now on 580.76.05.
Aug 15 21:06:17 cachyos-x8664 kernel: NVRM: GPU at PCI:0000:01:00: GPU-9e6dc988-ab46-8294-6c4a-5f8fafa84c15
Aug 15 21:06:17 cachyos-x8664 kernel: NVRM: Xid (PCI:0000:01:00): 13, pid=103289, name=RunLoop, Graphics Exception: channel 0x00000063, Class 0000c9c0, Offset 00000000, Data 00000000
Aug 15 21:09:40 cachyos-x8664 kernel: NVRM: Xid (PCI:0000:01:00): 13, pid=105491, name=RunLoop, Graphics Exception: channel 0x0000005f, Class 0000c9c0, Offset 00000000, Data 00000000
Aug 15 21:24:31 cachyos-x8664 kernel: NVRM: Xid (PCI:0000:01:00): 13, pid=109293, name=RunLoop, Graphics Exception: channel 0x00000060, Class 0000c9c0, Offset 00000000, Data 00000000
Aug 15 21:27:42 cachyos-x8664 kernel: NVRM: Xid (PCI:0000:01:00): 13, pid=110749, name=RunLoop, Graphics Exception: channel 0x0000005f, Class 0000c9c0, Offset 00000000, Data 00000000
Aug 15 21:33:30 cachyos-x8664 kernel: NVRM: Xid (PCI:0000:01:00): 13, pid=112684, name=RunLoop, Graphics Exception: channel 0x0000005f, Class 0000c9c0, Offset 00000000, Data 00000000
Aug 15 21:36:20 cachyos-x8664 kernel: NVRM: Xid (PCI:0000:01:00): 13, pid=114056, name=RunLoop, Graphics Exception: channel 0x00000060, Class 0000c9c0, Offset 00000000, Data 00000000
Aug 15 21:48:39 cachyos-x8664 kernel: NVRM: GPU at PCI:0000:01:00: GPU-9e6dc988-ab46-8294-6c4a-5f8fafa84c15
Aug 15 21:48:39 cachyos-x8664 kernel: NVRM: Xid (PCI:0000:01:00): 13, pid=4560, name=RunLoop, Graphics Exception: channel 0x00000053, Class 0000c9c0, Offset 00000000, Data 00000000
Bug Report:
nvidia-bug-report.log.gz (1.5 MB)
580.76.05 is a broken driver.
575.64.05 works.
- The symptoms of the breakage depends on the graphics card.
- All tests are on Wayland, GNOME 48.4 (Mutter & GNOME-Shell –version), Fedora 42. Two different computers. Both use AMD motherboards/CPUs.
- Results:
- RTX 2070 Super (1440p, 120Hz) = Every application immediately freezes when trying to close their windows.
- RTX 3090 (4K, 120Hz VRR) = Files (Nautilus) always freezes when trying to close its window, and other applications randomly freeze either when closing or during usage. They pretty much always freeze whenever they attempt to trigger a Nautilus-based file picker dialog.
- In both cases, you get a “Force Quit / Wait”. Picking “Force Quit” (at least on Files/Nautilus) crashes Mutter or Shell, gives you a black screen for a moment, and then throws you to the login screen.
- I’ve always seen “SIGSEGV” (segfault) as the crash reason for all of the applications, seen via Fedora’s “Problem Reporting” application.
My bet is on something related to GPU memory usage after free, since the 580.x drivers did a lot of changes related to VRAM memory handling.
I appreciate all the great changes you are making in the 580 driver and look forward to a version where the bug has been squashed. :)
I downgraded to 575.64.05 and everything is rock solid again on both machines. :)
PS: I also saw austin58 above mentioning that “wp_fifo_manager_v1” protocol support was added in 580.x, and may be buggy too (or may even be the root cause).
PPS: Apparently X11 works, but I haven’t confirmed it personally.
I have an 1060. After updating to the 580 driver on ArchBTW my two hdmi>dsub adapters lost their ability to set 1080p/60hz mode for my displays. Did anything related to EDID change? Reverting fixes the problem. Using a xorg override also fixes the problem, but this is really not a great solution..
Adding GSK_RENDERER=ngl to /etc/environment works around the gnome vulkan apps hang on wayland session
That was a lot of typing for a known issue.
Went back to 575 for now. I really hope the plethora of issues in the 580 driver are fixed soon. :)
On 580.65.06, running warframe from steam on 3070ti with proton-ge 10.11 and using the wayland backend, the game coes to black screen after ~20 minutes, Hard reboot is necessary.
Thank you for alerting me that they’ve found the issue and that a fix is on the way.
I really hope that the same bug fix also fixes all the other issues/crashes/app window freezes etc that have been reported about 580.76.05. Because it’s not just about “GTK apps freezing”. (In fact, the freeze is not GNOME specific. It’s a bug in the Vulkan driver. :D)
There’s also reported bugs in the new Wayland FIFO protocol they added in 580, etc.
If you’ve other separate issues please file them using the template.
And be sure to attach an nvidia-bug-report for each problem after reproducing.
I’m trying to set up the NVIDIA 580 driver on a laptop that has a GeForce RTX 5090 Max-Q as the dGPU and an Intel iGPU (4xe Core). The system runs with Coreboot, and I can only enable the dGPU.
It’s a multiboot setup with Windows 11 and Debian 13. Under Windows 11 everything works fine, but under Debian 13 I’ve run into issues. I tried both the GPL and the proprietary driver, with and without the iGPU enabled.
The problems vary:
-
Sometimes
nvidia-smireports that it cannot communicate with the GPU. -
In other cases, I just get a blank screen if I disable the dGPU in UEFI or set the following in GRUB:
nvidia-drm.modeset=1 modprobe.blacklist=nouveau modprobe.blacklist=i915
According to NVIDIA’s documentation, some laptop and all-in-one designs with switchable (Hybrid/Optimus) graphics may not work with the Linux driver if the iGPU cannot be disabled in hardware. It seems that even though I can disable the iGPU in Coreboot, Linux still fails to initialize the dGPU.
Has anyone experienced the same problem and found a working solution? Any hints would be much appreciated.
Hard to say for sure without any sort of a log. See if anything interesting turns up in dmesg and/or attach a bug report here generated by running nvidia-bug-report.sh as root. Also note that 50 series cards are only supported by the open-source driver, so don’t bother with the proprietary one.
If you just disable the iGPU support via the kernel cmd without switching the MUX to dGPU (hard to say if and how is it possible without knowing the specific laptop model) you’ll naturally get a black screen, as the iGPU is expected to do presentation (and desktop rendering) in this mode and you’ve left the system basically disconnected from the screen. Not sure why turning the dGPU off in UEFI causes problems, though.
I have an Nvidia PRIME/Optimus laptop myself and it mostly works aside from a bug that causes the dGPU to fail and require a reboot after it’s powered on/off a few times.
Reporting BUG
Synopsis: nier: automata will freeze after error on dmesg
Brief Description: i’m consistently getting the following error with Nier automata on the Hegel boss battle:
[ 322.768358] NVRM: GPU at PCI:0000:01:00: GPU-87126259-f8bb-0493-7c15-81af3681fd09
[ 322.768363] NVRM: Xid (PCI:0000:01:00): 13, pid=7331, name=NieRAutomata.ex, Graphics Exception: channel 0x00000071, Class 0000c9c0, Offset 00000000,Data 00000000
after this error the game will freeze.
Steps to Reproduce and Frequency: use “A130: Bomb” pod program while the boss is using the lazer attack, make sure the effects cover a good percentage of the screenspace. this is very frequent, might take some attempts to repro.
System Configuration: [custom desktop, 9800x3d,asus crosshair hero x670e][fedora 42][Plasma desktop][RTX 4070 TI Super][1 gigabyte M28U , 4K 144Hz, displayport]
Linux Bug Report:
nvidia-bug-report.log.gz (1.1 MB)
Additional details:
all graphics settings on the maximum.
steam launch arguments: DXVK_HDR=1 gamemoderun gamescope -w 3840 -h 2160 -r 144 --hdr-enabled --adaptive-sync --force-grab-cursor -s 1.5 -f – %command%
proton-ge10-12
game save file:
location: $HOME/.steam/steam/steamapps/compatdata/524220/pfx/drive_c/users/steamuser/Documents/My Games/NieR_Automata
select the second save slot in-game
NieR_Automata.tar.gz (29.3 KB)
I haven’t played this game in many years so I don’t remember when this takes place, and it doesn’t seem to want to load your save file. Is this Chapter 12-02: A2’s Battle in the Desert?
correct, it’s the battle in the desert, first boss with A2
The Last of Us, I am getting ~50 FPS with the GPU usage in 60-70% range on the 580.76 driver at 4K with a 4090. When enabling AMD FSR 3.1 with frame gen, the gpu usage is 99% most of the time and getting 120fps plus. This is with Wayland and NTsync on CachyOS. There is something with the gpu usage with this driver and DLSS that is causing low gpu usage. Agressive Vblank is enabled btw.
Despite the changelog suggesting potential improvements, my wired Valve Index VR experience remains subpar. I have tested the 570, 575, and 580 beta and release drivers, experimenting with different feature flags, RMIntrLockingMode, RmEnableAggressiveVblank, and NVreg_EnableGpuFirmware using both SteamVR and Monado on Fedora 42 and CatchyOS, either X11 or Wayland. I have tested with RTX 2070 and RTX 3080 GPUs, but I am still encountering jitter, though only when the framerate is low, which is when reprojection matters the most.
This can be easily verified through the abundance of the purple (Display Error) lines of the SteamVR Performance Graph.
