From owner-freebsd-fs Sat Oct 3 18:26:53 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id SAA03839 for freebsd-fs-outgoing; Sat, 3 Oct 1998 18:26:53 -0700 (PDT) (envelope-from owner-freebsd-fs@FreeBSD.ORG) Received: from smtp02.primenet.com (smtp02.primenet.com [206.165.6.132]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id SAA03811; Sat, 3 Oct 1998 18:26:40 -0700 (PDT) (envelope-from tlambert@usr06.primenet.com) Received: (from daemon@localhost) by smtp02.primenet.com (8.8.8/8.8.8) id SAA06542; Sat, 3 Oct 1998 18:26:19 -0700 (MST) Received: from usr06.primenet.com(206.165.6.206) via SMTP by smtp02.primenet.com, id smtpd006510; Sat Oct 3 18:26:10 1998 Received: (from tlambert@localhost) by usr06.primenet.com (8.8.5/8.8.5) id SAA16639; Sat, 3 Oct 1998 18:26:03 -0700 (MST) From: Terry Lambert Message-Id: <199810040126.SAA16639@usr06.primenet.com> Subject: Re: filesystem safety and SCSI disk write caching To: bde@zeta.org.au (Bruce Evans) Date: Sun, 4 Oct 1998 01:26:03 +0000 (GMT) Cc: Don.Lewis@tsc.tdk.com, gibbs@plutotech.com, freebsd-fs@FreeBSD.ORG, freebsd-scsi@FreeBSD.ORG In-Reply-To: <199810030124.LAA18635@godzilla.zeta.org.au> from "Bruce Evans" at Oct 3, 98 11:24:50 am X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-fs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org > >>BTW, with softupdates, and tagged command queuing enabled in CAM, there > >>is not much of a performance hit from turning off write caching. I > >>saw "make buildworld" increase from about 2 hours to 2 hours 5 minutes, > >>and "make -j6 buildworld" increase from about 1 hour 30 minutes to > >>1 hour 35 minutes. > > > >In this particular benchmark, perhaps not, but make buildworld is not > >indicative of most I/O loads. > > I think it can be interpreted as showing that the performance hit is > very large. `make world' is mostly cpu-bound, and most of it's i/o's > are reads (60% here). I guess it spends less than 5 minutes of its time > writing (27000 block output operations here). An increase of 5 minutes > is very large. This is without "noatime". Every inode read, is written, and every directory inode is written multiple times, and all object files and executables, as well as some generated sources, are written. If you read the Ganger/Patt paper, you will see that soft updates is within 5% of memory speed for most uses. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-fs" in the body of the message