https://bugs.kde.org/show_bug.cgi?id=525619
--- Comment #16 from Conn O'Griofa <[email protected]> --- (In reply to Zamundaaa from comment #15) > > https://invent.kde.org/plasma/kwin/-/merge_requests/9082 will probably fix > > it, but it breaks things atm. > I think I fixed it and also made it small enough for backporting. Could you > put the MR (just the first commit is enough) on top of the screencast fix > and see if it helps? I'm testing the two MRs against master and the issue seems to be mostly fixed. The specific testcase no longer reproduces and overall pacing is excellent. I'm testing normal gameplay, 60fps with vsync enabled, mangohud active on the host in-game and Moonlight's streaming stats on the client. Although the overall pacing is excellent, I did see two instance in the span of 30 minutes where the incoming framerate reduced to 40-45fps and remained there for perhaps 10-15 seconds, then the client framerate recovered. When the dip happened on the client, mangohud on the host was reporting a steady 60fps with a smooth frametime graph. There's a possibility this wasn't KWin's fault. I was testing with Sunshine using VAAPI encoding, and encoder throttling is known to occur if the EGL context used for the encoder is not created with high priority (but it was in my case). Just to be sure, I'll switch to Vulkan rendering (which doesn't have encoder throttling issues) and will test again as thoroughly as possible to try and rule out Sunshine as the culprit. -- You are receiving this mail because: You are watching all bug changes.
