575 BETA release feedback & discussion

Graphics / Linux Linux
43

Hi birdie, Thank you for starting the thread.

@matt-schwartz , can you share the gamescope version? I will file an internal bug. Noted this happens with Color accuracy set to ‘Prefer Efficiency’ [EDIT:- Prefer Colour Accuracy and not Prefer Efficiency]

@icalingua , I will check the linked thread for the details and file a bug. Can you please share your display model?

@dmakc , is this a driver regression on your setup? Did the 570 series drivers have the same issue?

@v19930312 , Thank you for the logs. I will file a bug and share the number for tracking. Do you mind collecting a NVIDIA bug report from the system and attaching it to your post?

1 Like
44

No updates on the VRAM exhaustion issues yet. Sorry

3 Likes
45

hi @abchauhan,

I was using gamescope version 3.16.2 at the time, but I’ve tested from latest master commit (3.16.4) and it had the same issue. it seemed like making gamescope -- vkcube fullscreen was enough to trigger the computer freeze.

the issue only occurs when using “Prefer color accuracy” on Plasma 6 Wayland. The issue is not present when using “Prefer efficiency,” which is the default choice in KDE. In case it matters, my monitor model is an LG 45GX950A-B plugged in via DisplayPort

1 Like
46

Hello. Using kernel 6.14.3, driver file NVIDIA-Linux-x86_64-575.51.02.run, MIT/GPL
branch.

When I launch steam, I get multiple errors in the terminal:
MESA: error: DRM_VIRTGPU_RESOURCE_CREATE_BLOB failed with Invalid argument
MESA: error: Failed to create virtgpu AddressSpaceStream
MESA: error: vulkan: Failed to get host connection

This repeats for a while until it settles down. In dmesg, this is coupled
with lots of these:
[ 892.133346] [drm:nv_drm_gem_map_offset_ioctl [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00004b00] Failed to lookup gem object for map: 0x00000002

And occasionally a couple of these:
[ 892.155097] [drm:nv_drm_gem_export_nvkms_memory_ioctl [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00004b00] Failed to lookup NVKMS gem object for export: 0x00000003
[ 892.157932] [drm:__nv_drm_nvkms_gem_obj_init [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00004b00] NvKmsKapiMemory 0x00000000b59b3e19 size should be in a multiple of page size to create a gem object

Bug report:
nvidia-bug-report.log.gz (665.0 KB)

There is something funky with nvidia-drm, in that I can’t install the drivers without
“–skip-module-load”. Other than these oddities, they work fine.

I’m using Linux Mint 22.1 on a System76 Thelio, card is an
NVIDIA GeForce RTX 3090 Ti (GA102-A). Any ideas?

48

Hi all,

In case you are reporting a driver issue on this thread, would you mind following the template below? This will help us keep track of issues that require internal bugs.

==== starts here ====
Reporting BUG

Synopsis:
Brief Description:
Steps to Reproduce and Frequency:
System Configuration: [System model][Operating System][Desktop Environment][GPU][Monitors - Model and Refresh Rates]
Linux Bug Report:
Additional details: [Screenshots, Screencapture, console output, logs or any details that can help with debugging]

====ends here ====

@amrits and I will review the comments, file internal bugs and list them all in this comment here.

Listing bugs filed for the issues reported on this thread:-

I.

Testing on a dual-GPU laptop (AMD Radeon 680M + NVIDIA GeForce RTX 4060 Laptop GPU). System completely locks up when it’s time to show the login screen and won’t let me do anything.

#5236637 - System lockup at login with 575-open BETA drivers on Notebook systems for v19930312’s report
STATUS : Checking if this can reproduced

II.

CTX SWITCH TIMEOUT` when running games

#5239137 - Running Black Myth Wukong benchmark results in a Xid 109, CTX SWITCH TIMEOUT with R575 beta drivers for xeeynamo’s report.
STATUS: Checking if this can be reproduced

III.

I just had a pretty serious crash where I was dumped into TTY while testing gamescope on Plasma 6 Wayland. After repeatedly fullscreening then windowing gamescope – vkcube ~10 times with Super + F, suddenly my PC locked up

#5239158 Running gamescope --vkcube repeatedly in full screen mode results in system lock up with R575 beta drivers for matt-schwartz’s report.
STATUS: Checking if this can be reproduced

IV.

Just tried new beta under Debian 12 with kernel 6.12.12+bpo-amd64 and I can’t successfully logon to KDE Plasma X11 when using proprietary NVidia kernel module with options nvidia NVreg_EnableGpuFirmware=1

#5239197 External monitor has a black screen for KDE X11 sessions on notebook systems with 575 beta drivers for dmakc’s report
STATUS: Confirmed. Under investigation.

V.

there’s a graphical bug in The Elder Scrolls IV: Oblivion Remastered (2623190) that is exclusive to the 575 driver series.

#5245355 - Graphical artifacts in The Elder Scrolls IV: Oblivion Remastered for matt-schwartz’s report.
STATUS: Confirmed. Under investigation

VI.

I am still having issues with crash to desktop when playing Indiana Jones TGC with FATAL ERROR: vkAcquireNextImageKHR failed with error (VK_TIMEOUT)

#5252126 - Indiana Jones TGC crashes to desktop with ‘FATAL ERROR: vkAcquireNextImageKHR failed with error (VK_TIMEOUT)’ for jimrd’s report
STATUS: Confirmed. Under investigation.

5 Likes
49

Nothing on the HDMI / DP issues?

50

Here’s a bug report collected on the closed source version of modules (as open modules won’t let me do anything other than forcibly power the system down): nvidia-bug-report.log.gz (1.5 MB).

51

Hi, what about this one? Non-existent shared VRAM on NVIDIA Linux drivers - #70 by hecate

3 Likes
52

I can confirm Display modes above 2560x1440p@120hz with HDR enabled cause flickering/corruption within gamescope-session - #6 by matt-schwartz still happens on 575 as well, with an updated nvidia-bug-report.sh output in my last comment there. Has been a bug since the first NV driver version that supported gamescope’s DRM backend.

1 Like
53

570 series driver has exactly the same issues. I can’t successfully logon with Plasma/Xorg in case of open-source kernel module. Only proprietary with options nvidia NVreg_EnableGpuFirmware=0 works for me.

54

The display model is included in the report, which is KTC H24T09P (KTC H24T09 Plus)

55

I saw the template for bug reports, will use that in the future. I see
I left out some info:
Monitor: Asustek Computer Inc PG42UQ
Display: X11: 3840x2160@120Hz
Attached on DP5, also used for sound.

56

Hey, thanks for your reply.
This is my nvidia-bug-report.log.gz (868.9 KB) with the reproducible CTX SWITCH TIMEOUT.

I also double-checked and I confirm I cannot reproduce this error when downgrading to 570.133.07.

========================

Reporting BUG

Brief Description: CTX SWITCH TIMEOUT when running games
Steps to Reproduce and Frequency: Download and install Black Myth Wukong Benchmark Tool, then run the benchmark. I can consistently reproduce it with 100% of success rate within a minute.
System Configuration: CachyOS on Kernel 6.14.2-2, KDE Plasma 6.3.4, NVIDIA RTX 5080, Monitor LG 32GR93U 4K@144 no VRR and no HDR on DP.
Linux Bug Report: nvidia-bug-report.log.gz (868.9 KB)
Additional details: I tried different configurations and patches, I tried using HDMI instead of DP and I disconnected all my non-primary monitors and I could still consistently reproduce the problem. This is not limited with Black Myth: Wukong, it happens on any game.

1 Like
57

Hi @scott-nv, Sorry, does Steam fail to launch on your system?

58

Reporting BUG

Synopsis: Open Kernel Modules fail to wake one / all monitors after suspend

Brief Description: After waking from suspend, GPU fails to wake up all screens attached (in my case 2 monitors). It can be worked around sometimes by going to tty2, unplugging monitor which isn’t working, plugging back in, and going back to graphical environment.

Steps to Reproduce and Frequency: Using the open kernel modules and using the boot parameter nvidia.NVreg_PreserveVideoMemoryAllocations=1, attempt to suspend the computer. After some time (maybe some hours), attempt to wake the computer up with more than one display connected. You will by chance notice that not all of the monitors wake correctly and remains black. In some cases, both monitors will work fine, and in some cases a hard reboot may be required to recover.

System Configuration:

  • [OS: NixOS 25.05 Warbler]
  • [Kernel: Linux 6.14.1]
  • [GUI Environment: Hyprland]
  • [GPU: NVIDIA RTX 3070]
  • [Monitors - Gigabyte M27Q (2560x1440@170Hz), Philips 273V7 (1920x1080@75Hz)]

Linux Bug Report: Many have been shared here: Suspend doesn't work when PreserveVideoMemoryAllocations is enabled · Issue #472 · NVIDIA/open-gpu-kernel-modules · GitHub

Additional details: Doesn’t happen 100% of the time weirdly. Also doesn’t happen with the closed drivers afaik.

59

Hi @faz
Thanks again for reporting here, however it would be great to share fresh nvidia bug report with latest 575 beta driver version.
Also, please confirm the last passed driver.

60

Hi UrbenLegend,
Can you please share linux bug report so that I can try to match config and reproduce issue internally.
Also do you know if there was any last passing driver.

61

Funny you ask, I was actually just able to reproduce the issue whilst being suspended for only about 10 minutes:
nvidia-bug-report.log.gz (421.3 KB)

In the case here, both screens were black, but I could still switch to tty2 which showed the secondary monitor working but the first never came to life.

62

#5240452 - Display modes above 2560x1440p@120hz with HDR enabled cause flickering/corruption within gamescope-session with driver 555.42.02

STATUS: Checking if this can be reproduced

63

Any progress on #5004630?
It’s still here in 575.51.02 :(