Date: Fri, 22 May 1998 15:39:56 +0930 From: Greg Lehey <grog@lemis.com> To: Guido Kollerie <gkoller@cs.vu.nl>, freebsd-scsi@FreeBSD.ORG Subject: Re: CAM performance on -stable Message-ID: <19980522153956.K27201@freebie.lemis.com> In-Reply-To: <Pine.SOL.3.96.980521111937.10781A-100000@keg.cs.vu.nl>; from Guido Kollerie on Thu, May 21, 1998 at 11:34:26AM %2B0200 References: <Pine.SOL.3.96.980521111937.10781A-100000@keg.cs.vu.nl>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 21 May 1998 at 11:34:26 +0200, Guido Kollerie wrote: > To get an indication of the perfomance of CAM I have performed three make > -j4 buildworld's. Once with the non-CAM kernel and twice with the CAM > kernel. /usr was mounted async,noatime and I used -O -pipe > -DCAM_OLD_TIMEOUT for CFLAGS, NOPROFILE was true > > To my surprise all the three buildworld times were nearly identical (3hr > 9min +/- 1min). I had expected the CAM kernel to perform much better as it > allows my BT-958 to use Command Tagged Queueing were as the non-CAM kernel > did not. > > Has anyone else found similar results on a -stable system? No, but I'd expect it. Make world doesn't really stress the SCSI system, since it doesn't handle more than one process at a time (well, not in a UP system, anyway). You'll see a lot more improvement if you use soft updates or async. Greg -- See complete headers for address and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19980522153956.K27201>