I know a lot of thread launching API calls allow specifying a priority on launch, but this doesn’t account for dynamic levels of load. I’m also aware programs that target a certain amount of loops per second can work out the duration that they took to finish an instruction, and sleep the thread for the rest of the duration