Messages in this thread | | | From | "Stephen C. Tweedie" <> | Date | Mon, 14 Feb 2000 20:32:37 +0000 (GMT) | Subject | Re: elevator-starvation-4 (2.2.14 && 2.3.42) |
| |
Hi,
On Fri, 11 Feb 2000 06:26:32 -0800, Bruce Thompson <bruce@otherother.com> said:
> It isn't terribly surprising to me that the problem seems to > develop more with IDE than SCSI: Time spent generating requests for > an IDE device is time that the driver cannot service those requests > since the CPU must actively work to service the requests.
It looks as if that doesn't have anything to do with the situation here --- the IDE drivers use the default elevator queuing, whereas the SCSI drivers take the requests off the elevator queue immediately and then do their own queuing within the SCSI layers. The difference in the two queuing policies appears to be what is dominating the performance differences between the two, as fixing IDE's queuing does give us a very much fairer request response curve.
--Stephen
- 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/
| |