Like bug 1462503, I expect this to be a regression from bug 1371668. Here's a profile I took of running https://output.jsbin.com/surane/quiet on my machine: https://perfht.ml/2NBnNUx It has 28ms refresh driver ticks but isn't keeping the main thread or the renderer thread busy. I think the intention here is to fall down to 30fps and do two refresh driver ticks per each 66ms window. And that is what happening in the tail end of the profile, but during the first half of the profile, there are many instances of doing two refresh ticks per 85ms. It just seems like our refresh tick scheduling is leaving too much idle time as soon as we can't keep up with 60fps.
Bug 1487864 Comment 0 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
Like bug 1462503, I expect this to be a regression from bug 1371668. Here's a profile I took of running https://output.jsbin.com/surane/quiet on my machine: https://perfht.ml/2NBnNUx Edit 2023-09-20: Updated profile: https://share.firefox.dev/3POXUDu It has 28ms refresh driver ticks but isn't keeping the main thread or the renderer thread busy. I think the intention here is to fall down to 30fps and do two refresh driver ticks per each 66ms window. And that is what happening in the tail end of the profile, but during the first half of the profile, there are many instances of doing two refresh ticks per 85ms. It just seems like our refresh tick scheduling is leaving too much idle time as soon as we can't keep up with 60fps.