From owner-freebsd-questions Fri Aug 15 15:02:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id PAA21559 for questions-outgoing; Fri, 15 Aug 1997 15:02:23 -0700 (PDT) Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id PAA21547 for ; Fri, 15 Aug 1997 15:02:13 -0700 (PDT) Received: (from root@localhost) by dyson.iquest.net (8.8.6/8.8.5) id RAA00904; Fri, 15 Aug 1997 17:01:21 -0500 (EST) From: "John S. Dyson" Message-Id: <199708152201.RAA00904@dyson.iquest.net> Subject: Re: FreeBSD is slower than Linux !? In-Reply-To: from Paul Dekkers at "Aug 14, 97 10:23:19 pm" To: psd@worldaccess.nl (Paul Dekkers) Date: Fri, 15 Aug 1997 17:01:21 -0500 (EST) Cc: shawn@luke.cpl.net, jdn@qiv.com, freebsd-questions@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL31 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-questions@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk > On Wed, 13 Aug 1997, Shawn Ramsey wrote: > > >Do be fair, I think you should mount the FreeBSd disks asyncronously. By > >default, it is set to Synchronously. Linux, at least it used to be this > >way, is mounted asynch. Disk access is HUGELY increases under FreeBSD if > >it is set to asynch. (mount -o async /dev/filesystem) > > Joking? Look at my dd-test: > Linux fBSD fBSD with async (linux with sync) > dd-test 2.61 4.95 4.78 2m06 (!!) > also with the other tests fBSD is still slower and the sync is faster, > even faster than last time without async, so FreeBSD really wrote > everything already. > and I'm sure the / was mounted with async! (checked with mount) > i'm using an i486 with 40mb's of memory so that should be enough for > FreeBSD to do something :-) > > -= Paul =- > > P.S. ext2 with sync mounted is terribly slow! Comparing that with fbsd, fbsd > is faster in writing really to disk. > With the -async option, FreeBSD actually writes long sequential writes without blocking. Those writes are done that way so that the disk-cache doesn't get backlogged with lots of dirty pages (buffers.) I have been pondering a really async option, but that would not be the default async. Our async option is really an async metadata option. John