You are not logged in.
Pages: 1
I have the exact issue described here:
https://www.reddit.com/r/ManjaroLinux/c … _on_144hz/
The video posted: https://imgur.com/a/5Hb1Mj0 is exactly the effect I am getting.
There is also a discussion here: https://bbs.archlinux.org/viewtopic.php?id=258638 but no solutions as it's actually a different problem.
I am using Wayfire (Wayland) but it doesn't seem to be DE dependent. I'm using a laptop with an intel and nvidia gpu. If I run the desktop environment through either gpu, the same thing happens and the person in the reddit post is using Ryzen.
I have tried two different mice (one 1000hz and one 500hz, both wireless) and get the same result.
If I move the cursor to my other screen while it's frozen I can have a frozen mouse on one display and a moving mouse on the other.
One discovery I have made is that if there is animation on the screen the issue never occurs. For example, if I run glxgears in a small window the stutter never happens. However, unless glxgears is actually displayed on the screen the issue still happens. If glxgears is minimized or running on another virtual desktop the stutter still happens
I have tried the following:
- Turning off/on adaptive sync in the monitor settings
- I thought it might be a GPU clock issue (which would explain why it doesn't happen when glxgears is running) so I tried setting the clocks to highest using `intel_gpu_frequency --max` so the clocks are always at max
- PSR or some other power saving technique might be kicking in and preventing the refresh of the display so I disabled PSR with `i915.enable_psr=0` and `i915.enable_fbc=0`
- It could be scheduling issue, I tried `i915.enable_guc=2`
- Disabled swap (I read this can happen when swapping)
None of this makes any difference. As a guess it seems like there is something in the display code somewhere that says "the frame hasn't changed, don't bother to update the display" and the small mouse cursor on the 4k display isn't enough for it to think the frame has changed, a bit like PSR is supposed to do. Perhaps it's not being disabled properly?
It can be avoided by:
- Running the display at 60hz
- Ensuring there is something animating on the screen so it refreshes correctly
- Firefox is open and visible on the monitor but not doing anything, I don't know why this is.
Any suggestions are welcome
Offline
To partially answer my own question. After switching my clocksource from tsc to hpet, the issue appears to have gone away.
However, hpet was disabled as being unreliable on my system. I'm not sure what side effects `hpet=force` and switching to it will have but it does show that the issue is with the clocksource.
It also explains why people on different hardware and different desktop environments are affected.
Is there any way to finetune tsc to potentially avoid the issue?
edit: While I had been assuming it was a GPU issue I'd ignored the CPU. I wonder if it's when the CPU clocks change that the stutter occurs and interferes with the clock...
Last edited by Tom B (2022-11-18 20:34:14)
Offline
Pages: 1