The applications are using pthread lib, in case process A has 100 threads ready, after that process B has one thread ready to run, will scheduler 1) pick and execute thread(s) from process A maxmumlly to one time slice, then pick process B thread. Or scheduler will 2) execute 100 threads from process A first, then execute process B's thread? That means process B thread has to wait 100 x time slice to get a chance to run, will this starve process B?
If the 2nd one, it is quite different behavior from Linux CFS scheduler, right?
