https://bugs.kde.org/show_bug.cgi?id=509669
--- Comment #46 from fililip <[email protected]> --- Yeah, this is why I'm of the opinion that the cursor plane should be special cased in that it should be reusing images rendered only once for each display scale and only actually repaint them when the display configuration changes, and all of this should be controlled by another thread (which has actually been done already, I think). Redrawing the cursor on the GPU every time is risky if a stable frame time can't be achieved or the GPU is stuck/hung. It's precisely why stuff like FurMark will stall cursor repaints, and this was a problem even back on 6.4 and prior versions. Maybe it's possible to just force a CPU image blit instead here and that would be a better workaround? Not sure how bad that would be for laptop or power constrained use long-term though :/ (By the way, I have always been able to notice unstable vblank intervals on Nvidia display hardware; it was simply never as stable as AMD or Intel iGPUs/dGPUs. I even reported it here thinking KWin could do something about it but it just seems to be an Nvidia issue: https://bugs.kde.org/show_bug.cgi?id=481716) -- You are receiving this mail because: You are watching all bug changes.
