Hi all, I have a set of N processes and would like them to be scheduled such that each process gets only a fixed time slice (that can be set) and nothing more than it. I read through the documentation of Completely Fair Scheduler/Round Robin scheduling and resource limitation using cgroups, but all of them seem to be giving a "min guarantee" to running processes. i.e., if I have 4 processes, out of which 2 are idle, the remaining 2 get exactly 50% of the CPU. But I do not want this.. even if two processes are idle, I would like the remaining 2 to get only 25% of the CPU time and idle-wait the remaining time (a max-limit, rather than a min-guarantee).
I came across cpulimit that does precisely what I need. I am wondering if there's a linux scheduler that does the same, to avoid overheads. I also came across a webpage (I am not able to find it now) that talked something on the lines of setting the scheduling epoch interval and timeslice quanta to a set of processes. I believe this abstraction would also be useful (i.e., processes 1--500 each get 1ms time in an epoch lasting 1s; the remaining 500ms is used for the kernel and other tasks). Thanks, -- Vimal -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to [email protected] Please read the FAQ at http://kernelnewbies.org/FAQ
