[Bug 217537] New: [AMDGPU] RDNA Freesync problem with CVT-Reduced display profile
bugzilla-daemon at kernel.org
bugzilla-daemon at kernel.org
Sat Jun 10 03:05:45 UTC 2023
https://bugzilla.kernel.org/show_bug.cgi?id=217537
Bug ID: 217537
Summary: [AMDGPU] RDNA Freesync problem with CVT-Reduced
display profile
Product: Drivers
Version: 2.5
Hardware: AMD
OS: Linux
Status: NEW
Severity: normal
Priority: P3
Component: Video(DRI - non Intel)
Assignee: drivers_video-dri at kernel-bugs.osdl.org
Reporter: contato-myghi63 at protonmail.com
Regression: No
Created attachment 304393
--> https://bugzilla.kernel.org/attachment.cgi?id=304393&action=edit
edid of my AOC 24G2 Monitor
The problem:
When a CVT-RB or CVT-RB2 monitor profile is enabled, using Freesync on a game
that demands 100% of GPU utilization makes the mouse pointer to perform very
slow (2~3fps) on Wayland, even if the game is running above 90fps smoothly.
Recording the screen results in a video without this problem happening (maybe
pipewire simulates the mouse?)
With x11, the game itself starts jumping frames while recording the screen
(CVT-RB only), or always have a terrible struttering (CVT-RB2).
Why do I want to use CVT-RB or CVT-RB2 monitor profiles?
I want to decrease the idle power consumption of my GPU. Using one of those
monitor profiles do the trick (15W -> 3W).
Does these profiles work on Windows?
Yes, without any issue. I'm sure the problem is only happening on Linux
distros.
Which things I've done to try fixing the issues?
Tried linux-lts, switched between both amdvlk and radv (vulkan-radeon), tested
amdgpu.dc=0 (it ended up freezing the GPU before launching display server),
made a clean install of Arch Linux, tried different display profiles while
modifying the edid and none of these options helped at all.
Hardware:
GPU: AMD RX 5500XT 8GB (GV-R55XTOC-8GD)
Monitor: AOC 24G2 1920x1080 144hz Freesync Premium
Both are connected with DisplayPort 1.2
Software:
OS: Arch Linux
DE: Plasma 5.27.5
Kernel: linux 6.3.6 / linux-lts 6.1.33
Mesa: 23.1.1-1
I'm attaching the original edid of my monitor, just in case if someone wants to
analyze it (dumped with read-edid).
--
You may reply to this email to add a comment.
You are receiving this mail because:
You are watching the assignee of the bug.
More information about the dri-devel
mailing list