[PERF] Lots of display errors in peformance graph/stuttering
I'm seeing the same thing with a very similar configuration:
Distribution: Arch
SteamVR version: 2.8.8
Steam client version: 1738026274
Opted into Steam client beta?: No
Graphics driver version: nvidia-open 565.77-12
SteamVR system report: https://gist.github.com/chr-1x/ab31edfca50a060ca2acad7d9842aa99
Performance data:
CPU: Ryzen 7 7800X3D
GPU: RTX 4070S
gist for perf top: https://gist.github.com/chr-1x/dc6ddfbd23a146ba13c9e418ad21fb7c
link to gpuvis trace: "gpu_profiler_capture" debug setting doesn't seem to be doing anything for me
Very similar issue. Lots of pink bars in the perf graph even though steamVR is well below the target frame time.
Hello @machinedgod, @chr-1x, check if this is the NVIDIA driver regression from GSP firmware being enabled https://download.nvidia.com/XFree86/Linux-x86_64/565.77/README/gsp.html https://forums.developer.nvidia.com/t/enabling-gsp-makes-steamvr-unplayable-on-rtx-3090ti/298694/12
Hey @kisak-valve ,
thank you for the reply.
My query with nvidia-smi shows it is; I'll try following instructions in disabling it and post results in few hours.
Update:
that worked! Thank you @kisak-valve, you're a star ⭐
@chr-1x
Just in case its needed (sorry if it isn't), to save you tiny bit of time - disabling the GSP isn't as straightforward as documentation states.
- if you modify the kernel parameter, aside making sure not to modify
grub.cfgbut/etc/defaults/grub(and then rungrub-mkconfigas Arch docs specify) - you have to prefix the parameter with module name, so itsnvidia.NVreg_EnableGpuFirmware=0. I have not tried this however! - if you go the
/etc/modprobe.droute (which is what I've done) - don't be like me and forget to alsosudo mkinitcpio -P😂
I'll close this as resolved.
P.S. For any future Arch wanderer who might stumble upon this and needs more handholding with module parameters - these are the contents of /etc/modprobe.d/40-nvidia.conf:
options nvidia NVreg_EnableGpuFirmware=0
and also please read this before tampering with module or kernel parameters:
Arch Wiki - Kernel parameters
Arch Wiki - Kernel modules settings
Yep, that did it. Exact steps:
sudo pacman -S nvidia(answer 'y' when it asks to replacenvidia-open)sudo vim /etc/modprobe.d/40-nvidia.confand put the following in the file:options nvidia NVreg_EnableGpuFirmware=0sudo mkinitcpio -P- reboot
I verified that a GSP firmware version was not listed in the output of cat /proc/driver/nvidia/gpus/0000:01:00.0/information. Upon starting steamVR, there were far fewer pink lines and the view in SteamVR home was stable 👍
Edit:
scratch the below (leaving for achives).
I just tested previous and stable branch - and getting exactly same behaviour. Must be something else. I'll investigate and post here if I find anything. If there's any ideas, shoot them my way please.
Edit 2: Solution, for anyone who experienced the same issue. ZERO display errors at Ultra+ settings in Elite Dangerous!
Add new nvidia module option and re-enable the firmware if you disabled it (/etc/modprobe.d/40-nvidia.conf).
...
#### VR
# This used to fix the lines, but ever since the update on 2026/06/24
# they're back - so I'm trying to re-enable the firmware
#options nvidia NVreg_EnableGpuFirmware=0
options nvidia NVreg_EnableGpuFirmware=1
# 2026/06/24 - enbling this because it says on the forum its a special
# feature to help?
#
# https://forums.developer.nvidia.com/t/enabling-gsp-makes-steamvr-unplayable-on-rtx-3090ti/298694/12
# Implemented a feature that allows low latency display interrupts to
# be serviced even when the system is under heavy contention. This
# is especially useful for reducing stutter when using virtual reality.
#
# This feature is experimental and disabled by default.
#
# This feature can be enabled by loading nvidia.ko with the
# `NVreg_RegistryDwords=RMIntrLockingMode=1` kernel module parameter.
options NVreg_RegistryDwords=RMIntrLockingMode=1
...
Also, could be irrelevant - but I switched SteamVR to Stable for now. Likely it doesn't matter if you remain on beta, but, I wanna play now and don't feel like investigating further 😅
Original report
Hi, hope this is a good place to re-report possible regression in beta? If not - my apologies, please close the issue (or I can, whatever works).The latest beta (Jun 24, 2026) has either regressed this bug, or has introduced display errors elsewhere.
However, after latest update, Elite Dangerous is borderline unplayable - the graph looks almost the same as the original one I posted, display errors all over (and stable 15-18ms for main render loop of the game, 90Hz, Advanced Filtering is Off).
I went back to check if nvidia screwed something up too, as I couldn't remember updating, and I tried tinkering with new options they added (NVreg_RegistryDwords=RMIntrLockingMode=1 supposed to help with low latency display ints) and tried re-enabling the GpuFirmware, but it had no (perceptible) effect either way.
nvidia-smi relevant output:
NVIDIA-SMI 610.43.02 KMD Version: 610.43.02 CUDA UMD Version: 13.3
Will gladly provide any other info that's needed. Hardware didn't change.
Nothing extracted yet.
Describe the bug
Moving the headset causes a lot of stuttering. Observing perf graph shows those as pink lines/display errors.
To Reproduce
Simply run SteamVR, wait for few seconds until all loading stabilizes and use the headset as normal.
Expected behavior
There are no display errors - frame time is well below limits (6-7ms out of 11.1ms for 90Hz)
System Information (please complete the following information):
Performance Data (optional, but very helpful):
Screenshots
Additional context
I've attempted:
Nothing resolved the issue but some configurations seem to create less display errors than the others. Could just be random, though.