Currently despite using more CPU resources multithreading still finishes rendering in roughly the same time. This was the case even if I disabled autoresetevents that force threads to wait to retain frame order. Maybe has something to do with threads flipping back and forth between it and ffmpeg?
If I put extensive maths calculations into frame calculations that caused the render time to increase significantly, then multithreading became noticeably faster than singlethreading.
Currently despite using more CPU resources multithreading still finishes rendering in roughly the same time. This was the case even if I disabled autoresetevents that force threads to wait to retain frame order. Maybe has something to do with threads flipping back and forth between it and ffmpeg? If I put extensive maths calculations into frame calculations that caused the render time to increase significantly, then multithreading became noticeably faster than singlethreading.
Branch #multithreading-test