Date: Wed, 22 Dec 1999 20:52:42 -0600 From: David Kelly <dkelly@HiWAAY.net> To: freebsd-questions@FreeBSD.ORG Subject: Re: Process Scheduling (was: Re: Sys Admin article on Linux emulation) Message-ID: <199912230252.UAA57478@nospam.hiwaay.net> In-Reply-To: Message from Lowell Gilbert <lowell@world.std.com> of "22 Dec 1999 08:17:20 EST." <rd61z8funbz.fsf@world.std.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Moved to -questions "because I want to know". Please cc: me (dkelly@hiwaay.net) as I'm not subscribed to -questions and its probably not wise to with the holidays forthcoming with less time to pay attention to massive volumes of email. Lowell Gilbert writes: > David Kelly <dkelly@hiwaay.net> writes: > > > Would have hoped/expected a process at maximum niceness would not run at > > all if a normal-nice process wanted to run. > > No, "nice"d processes should not get completely frozen out. > What you're describing is idprio(1). Reason I bring it up is with a P-II 400 and 3.2-RELEASE I had one of the last rc5des (before they got named dnetc) crunchers running. With "systat -v" was watching "cvs checkout" run. The CVS archive was on a mostly empty 7G partition on one SCSI HD, writing to another mostly empty large partition on another. Both were on the same AIC7890 family U2LVD controller on an Asus P2B-S MB. Drives are UW, not LVD. Anyway, cvs had less than 5% of the CPU. Drives were doing about 200k bytes/sec. Rc5des had the other 95%. Killed rc5des and drives moved up to about 500k/sec. Did not have softupdates enabled. Also I didn't take good notes, but cvs ran about 3 times faster once rc5des was stopped. Am pretty sure it was a situation where cvs had very little to do once it got a time slice then yeilded it when it blocked for I/O. Rc5des got the remainder of the slice. Started the next slice. I/O completed. And I suspect cvs had to wait until rc5des used all its slice before getting a chance at the incoming data. That machine was scheduled for a wipe and update last month. Maybe I'll get around to it next week. :-) My PPro (oc'ed to 210) with only one 9G HD for both target and destination seems to run cvs faster than the P-II 400. Have a newer FreeBSD on the PPro. Also have softupdates enabled. Other differences is the PPro has X running all the time, the P-II doesn't have X installed at all. One would think that would be in favor of the P-II. Anyway, one of the things about rtprio and family that I don't understand is why a plain user can't use such to decrease the priority of his own processes? Or that a process can't decrease itself. The way nice can be used? -- David Kelly N4HHE, dkelly@hiwaay.net ===================================================================== The human mind ordinarily operates at only ten percent of its capacity -- the rest is overhead for the operating system. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199912230252.UAA57478>