Mikhail Panasyuk <otoko@xxxxxxxxx> wrote: > Experimental schedulers (based on r33922): > http://otoko.narod.ru/files/haiku/haiku_kernel_schedulers.zip Note that you don't use the current sources as a base for your work; I've recently changed the affine scheduler to work like the BeOS one after your findings, ie. it will now skip per priority level, not just 20%. This should improve the latency, but I couldn't notice any actual changes. The priority_info optimizations you made make definitely sense (only looked at your changes in the simple scheduler so far, but I guess you did the same for affine?). How about cleaning them up (removing dead code, and coding style violations), and submit a patch? Bye, Axel.