Date: Wed, 11 Jul 2001 16:43:01 -0500 (CDT) From: Mike Silbersack <silby@silby.com> To: Ian Dowse <iedowse@maths.tcd.ie> Cc: Dag-Erling Smorgrav <des@ofug.org>, <cvs-committers@FreeBSD.org>, <cvs-all@FreeBSD.org> Subject: Re: cvs commit: src/sys/i386/conf NOTES src/sys/conf files options src/sys/ufs/ufs dirhash.h ufs_dirhash.c inode.h ufs_inode.c ufs_lookup.c Message-ID: <20010711163937.R2662-100000@achilles.silby.com> In-Reply-To: <200107112116.aa67322@salmon.maths.tcd.ie>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 11 Jul 2001, Ian Dowse wrote: > In message <xzpg0c38f5q.fsf@flood.ping.uio.no>, Dag-Erling Smorgrav writes: > >Ian Dowse <iedowse@FreeBSD.org> writes: > >> Bring in dirhash, a simple hash-based lookup optimisation for large > > > >Are you going to MFC this in a while? > > Yeah, I have a releng_4 version that I've been running without any > problems, but I'll probably wait at least a week or two for any > bugs or suggestions to show up. There are almost certainly still > some access patterns where dirhash is a pessimisation over the > non-dirhash case - I think the worst of these (sequential access, > large working set) have mostly been addressed, but some filesystem > benchmarks may well come out worse with dirhash switched on. > > I'd be very interested in any such results, or any suggestions for > improvements. > > Ian I did a few runs with "postmark" from the ports tree, and found that performance was mostly unchanged. I'm not sure if it's a good benchmark, but investigating why it shows no overall performance increase would probably be worthwhile for you. In any case, I think that you should do the MFC ASAP. Since it's an optional component, you don't have to worry about breaking anything, and you'll need the larger -stable audience to get a good variety of testing done. Mike "Silby" Silbersack To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010711163937.R2662-100000>