You are not logged in.

#26 2025-04-24 15:17:18

seth
Member
Registered: 2012-09-03
Posts: 64,355

Re: [Solved] Microstutter with Multimonitor

xrandr -q
xrandr --output VGHDP-123 --gamma 1.5
xrandr --output VGHDP-321 --gamma 1.5
…

Does altering the gamma on a specific output (or maybe even attempting it on a disconnected one) cause such stall (and maybe an error message)?

Offline

#27 2025-04-24 21:40:05

Aedahng9
Member
Registered: 2020-08-14
Posts: 24

Re: [Solved] Microstutter with Multimonitor

Indeed it does. Every time I change the gamma, the stall occurs. But it doesn't output any error.

Last edited by Aedahng9 (2025-04-24 21:42:33)

Offline

#28 2025-04-25 07:19:39

seth
Member
Registered: 2012-09-03
Posts: 64,355

Re: [Solved] Microstutter with Multimonitor

Regardles of the output you try to change but only if there're mulitple?
And "xrandr -q" does *not* stall the displays?

This is regardless of whether xf86-video-amdgpu is used and even affects xwayland (ie. running xrandr on a wayland session)?

Offline

#29 2025-04-25 08:33:32

Aedahng9
Member
Registered: 2020-08-14
Posts: 24

Re: [Solved] Microstutter with Multimonitor

seth wrote:

Regardles of the output you try to change but only if there're mulitple?

Yes, doesn't matter which output I change. As soon as there is more than 1 monitor active, it stalls.

seth wrote:

And "xrandr -q" does *not* stall the displays?

No, this does not stall. Only changing the gamma.

seth wrote:

This is regardless of whether xf86-video-amdgpu is used and even affects xwayland (ie. running xrandr on a wayland session)?

xf86-video-amdgpu is currently not installed, and it's stalling. It affects xwayland aswell, yes.

Edit: Regardless of whether xf86-video-amdgpu is installed or not, stalling occurs on changing the gamma.

Last edited by Aedahng9 (2025-04-25 08:53:41)

Offline

#30 2025-04-25 12:38:38

seth
Member
Registered: 2012-09-03
Posts: 64,355

Re: [Solved] Microstutter with Multimonitor

https://gitlab.freedesktop.org/drm/amd/-/issues/623
=> https://gitlab.freedesktop.org/drm/amd/-/issues/3708 (symptom description is a perfect fit)
AD blames https://en.wikipedia.org/wiki/Memory-ma … mapped_I/O but that does somehow not explain the multiscreen factor.

Interestingly the OP faces this only w/ high refresh rates but the secondary comment suggests to get it w/ all rates - but two outputs.
I don't think that the pclk causing some threshold at which you run int 100ms stalls can be explained by the expense of MMIO - there's probably some "bug" where some threshold runs you into a way too generous timeout or so.

Offline

Board footer

Powered by FluxBB