... wrote: > I noticed that my network performance has gone down from ... MBytes 95.0 Mbits/sec > to [ 3] 0.0-10.0 ... tracked it back to the commit where CFS > was enabled ("sched ... the resulting debug file. Ingo - To unsubscribe from this list: send ...

linux-kernel - Ingo Molnar - Sep 26 2007 - 05:34

... The ideal > solution is of course to almost never use yield. Fortunately 99 ... to boast a little: "( another detail: due to nanosec accounting and timeline sorting ... have tested so far. )" [Documentation/sched-design-CFS.txt] Cheers, Jarek P. - ...

linux-kernel - Jarek Poplawski - Oct 3 2007 - 04:02

I noticed that my network performance has gone down from 2.6.22 from [ ... 10.0 sec 113 MBytes 95.0 Mbits/sec to [ 3] 0.0-10.0 sec 75.7 ... . I did a git bisect today and tracked it back to the commit where CFS was enabled ("sched: cfs core code ...

linux-kernel - Martin Michlmayr - Sep 26 2007 - 04:52

... wrote: > I noticed that my network performance has gone down from 2 ... and tracked it back to the commit where CFS > was enabled ("sched: ... might be a worth while to try it without it. -Mike ... the body of a message to majordomo@vger.kernel.org More ...

linux-kernel - Mike Galbraith - Sep 26 2007 - 06:23

... is somewhat confusing without additional context. I'll improve the wording. (sched-design-CFS.txt needs an update anyway) Ingo - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the ...

linux-kernel - Ingo Molnar - Oct 7 2007 - 03:18

... fix would be for iperf to use blocking network IO > > though, or maybe ... rise. Ingo: Can you clarify what CFS' current default sched_yield implementation is and ... this thread is scheduled again)? DS - To unsubscribe from this list: send the ...

linux-kernel - David Schwartz - Sep 26 2007 - 08:00

... so, it would be very difficult to find. > [...] google.com/codesearch is your friend. ... I'm not familiar enough with CFS' internals to help much on the > ... Linux apps follow that ideal solution ;-) ] Ingo - To unsubscribe from this list: send the ...

linux-kernel - Ingo Molnar - Oct 2 2007 - 02:06

... timeslice length' - timeslices in CFS are of variable length. > > > * (to see the precise ... of timeslice length" is difficult to understand? The timeslices of tasks ( ... see new sched_slice() function which seems > to return... time.) wrong again. That ...

linux-kernel - Ingo Molnar - Oct 3 2007 - 05:10

... timeslice length' - timeslices in CFS are of variable length. > > > > * (to see the ... of timeslice length" is difficult to understand? The timeslices > of tasks ( ... new sched_slice() function which seems > > to return... time.) > > wrong again. That ...

linux-kernel - Jarek Poplawski - Oct 3 2007 - 05:50

... > http://people.redhat.com/mingo/cfs-scheduler/tools/cfs-debug-info.sh > > ... the two cases can be compared) Ingo - To unsubscribe from this list: send the line ... kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info ...

linux-kernel - Ingo Molnar - Sep 26 2007 - 05:47

... time to saturate 1 gigabit of network traffic to another box. With the patch ... or other bug. In that sense CFS does the right thing by exposing such bugs =B-) Ingo -------------------------> Subject: iperf: fix locking From: Ingo Molnar < ...

linux-kernel - Ingo Molnar - Sep 26 2007 - 09:31

... saturate 1 gigabit of > network traffic to another box. With the ... removed altogether > > - i was able to measure much higher bandwidth over ... other > bug. In that sense CFS does the right thing by ... the body of a message to majordomo@vger.kernel.org More ...

linux-kernel - Stephen Hemminger - Sep 26 2007 - 11:40

... current sched_yield behavior *is* for CFS and > > what the tunable does ... the right end of the CFS rbtree. This means that for ... not familiar enough with CFS' internals to help much on the ... the time, if possible?) DS - To unsubscribe from this list: send ...

linux-kernel - David Schwartz - Oct 1 2007 - 14:23

... notion of "timeslices" in CFS. (in CFS tasks > > "earn" a right ... length' - timeslices in CFS are of variable length. > * (to see the ... there is no pending timeslice to be "cleared" as it has ... the body of a message to majordomo@vger.kernel.org More ...

linux-kernel - Ingo Molnar - Oct 3 2007 - 04:16

... would be for iperf to use blocking network IO >>>> though, or ... other > bug. In that sense CFS does the right thing by ... different kind? Regards, Jarek P. - To unsubscribe from this list: send ... the body of a message to majordomo@vger.kernel.org More ...

linux-kernel - Jarek Poplawski - Sep 27 2007 - 05:30

... but SCHED_FIFO/RR tasks (for which tasks CFS does keep the behavior). Please read ... lkml for a more detailed background: CFS: some bad numbers with Java/database ... the twiddling of that sysctl). Ingo - To unsubscribe from this list: send the ...

linux-kernel - Ingo Molnar - Sep 27 2007 - 05:46

... lkml for a more detailed > background: > > CFS: some bad numbers with Java/database ... some 'compatibility' test could be prepared to compare a few different ideas on ... by next kernels? Thanks, Jarek P. - To unsubscribe from this list: send the ...

linux-kernel - Jarek Poplawski - Sep 27 2007 - 08:27

... length' - timeslices in CFS are of variable length. > > * (to see the ... new sched_slice() function which seems to return... time.) > > In the 2 ... you count eg. in check_preempt_curr_fair to find if time is over ... and you didn't care to explain why it's so ...

linux-kernel - Jarek Poplawski - Oct 3 2007 - 04:56

... code i might be able to point out the breakage in ... well when the > > discussion shifts to yield behavior, because that shows ... time requeued to the right end of the CFS rbtree. This ... up breaking more apps.) Ingo - To unsubscribe from this list: send ...

linux-kernel - Ingo Molnar - Oct 1 2007 - 13:31

... lkml for a more detailed > > background: > > > > CFS: some bad numbers with Java/database ... 'compatibility' test could be prepared to compare > a few different ideas on ... above - hence the switch.) Ingo - To unsubscribe from this list: send the ...

linux-kernel - Ingo Molnar - Sep 27 2007 - 09:31