Messages in this thread | | | Date | Tue, 14 Mar 2000 16:08:12 +0100 (CET) | From | Ingo Molnar <> | Subject | Re: 2.3.51: 'timeslice transfer' |
| |
On Tue, 14 Mar 2000, Borislav Deianov wrote:
> Or you can use Fairsched. Briefly: processes are assigned to nodes, > each node has a weight. Nodes are guaranteed their "fair share" of the > CPU. Each node uses the standard scheduler internally. Therefore, no > matter how antisocial your processes are, you can only hurt your own > node.
yep - the Linux priority model is just the most obvious one and it does not try to guarantee any static percentage how CPU time is split up between processes (in the SCHED_OTHER scheduling class that it). (and does not know about CPU time assigned to a group of processes at all) Priorities are rarely used currently. Once massively parallel shared memory multiuser systems become common (with a CPU count of say more than 100), more finegrained control of CPU resources will likely get desirable. (and there it will probably be a must).
Ingo
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |