Messages in this thread | | | From | Peter Chubb <> | Date | Fri, 12 Sep 2003 12:01:05 +1000 | Subject | Re: [PATCH] schedstat-2.6.0-test5-A1 measuring process scheduling latency |
| |
>>>>> "Rick" == Rick Lindsley <ricklind@us.ibm.com> writes:
Rick> I'm not sure that it's any less or more intrusive, but it's at Rick> least another way of doing the same thing. So since you've Rick> taken some measurements, what's the length of time you find your Rick> process waits to hit the processor after getting the I/O it Rick> needs? What's the time it seems to wait when it skips (what's Rick> the cutoff at which you hear a skip versus don't hear one?)
In short, time on the run queue is negligeable. Time spend waiting for disk I/O is extensive. If you look at the graph, you'll see each disk I/O takes 0.8 seconds, which is about the same as the skip. This is on ReiserFS 2.6, laptop (4000RPM) disk, but attached to mains power.
There's a massive difference between the behaviour with hdparm -u1 and hdparm -u0 --- I don't see skips with -u1, and the time the disk light is on is much reduced. This puts my problem in the IDE layer somewhere -- which means it may not be the same problem others are seeing.
-- Dr Peter Chubb http://www.gelato.unsw.edu.au peterc AT gelato.unsw.edu.au You are lost in a maze of BitKeeper repositories, all slightly different. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |