https://bugs.kde.org/show_bug.cgi?id=525619
--- Comment #5 from Conn O'Griofa <[email protected]> --- Testing against commit e29d8f841220cba2daf501e50dfa32ee335961d9 (commit-timing) with 3fad2bbcdcbd087c11a9f3b7848bc2043ccc2341 fully reverted, the 120fps issue doesn't appear, but the test case can still be reproduced. Stuttering with the 60 -> 30 -> 60 rate change (or from random 1% lows). It seems that reverting that commit doesn't reintroduce the ~3% rate loss as seen on 6.7 branch, so the incoming rate is always ~60.x instead of ~57. This means that vsync on can *sometimes* be smooth and vsync off seems about as smooth as 6.7.90. (I would still want to keep the nanosecond wait precision from that commit regardless). When the stuttering issue manifests, the incoming rate as reported by Moonlight doesn't drop, and mangohud never reports any frametime deviations from ~16.6ms. Moonlight does do frame pacing based on the encoded frame timestamps, but we don't use the pipewire pts metadata directly for this (yet) and instead just sample steady_clock::now() on arrival, because older KWin versions were providing very stale timestamps when returning from idle. -- You are receiving this mail because: You are watching all bug changes.
