home *** CD-ROM | disk | FTP | other *** search
- Path: nntp.hut.fi!usenet
- From: Osma.Ahvenlampi@hut.fi (Osma Ahvenlampi)
- Newsgroups: comp.sys.amiga.misc
- Subject: Re: AFS speed
- Date: 20 Jan 1996 19:21:31 +0200
- Organization: What, me, organised?
- Sender: oahvenla@hyppynaru.cs.hut.fi
- Distribution: inet
- Message-ID: <jdjwx6mrdd0.fsf@hyppynaru.cs.hut.fi>
- References: <4d7776$rru@detroit.freenet.org>
- <898.6587T1160T1958@mbox3.swipnet.se> <30FD3CB4.34AC@pop.gpnet.it>
- NNTP-Posting-Host: hyppynaru.cs.hut.fi
- Mime-Version: 1.0
- Content-Type: text/plain; charset=US-ASCII
- In-reply-to: Mark Marin's message of Wed, 17 Jan 1996 19:15:16 +0100
- X-Newsreader: Gnus v5.1
-
- In article <30FD3CB4.34AC@pop.gpnet.it> Mark Marin <megavolt@pop.gpnet.it> writes:
- >Has anybody tried comparing AFS with FFS, if you set both to a
- >similar buffer size? I have my FFS set to 100 buffers, which helped
- >greatly. I wonder if 300 would be comparable to AFS?
-
- FFS doesn't use the buffers for speeding up file access, only to cache
- directory blocks. Thus, buffers don't affect it that much. With 50
- buffers, a decent value for FFS, AFS is painfully slow. Giving as much
- actual cache memory to FFS volumes using DynamiCache gives results
- comparably to AFS with large buffers.. Difference being that
- DynamiCache can share the same cache memory between a lot of volumes
- (using it where it is needed), while AFS needs separate buffers for
- each volume.
- --
- foo
-