From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S263654AbTDXNTa (ORCPT ); Thu, 24 Apr 2003 09:19:30 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S263655AbTDXNTa (ORCPT ); Thu, 24 Apr 2003 09:19:30 -0400 Received: from ns.suse.de ([213.95.15.193]:13836 "EHLO Cantor.suse.de") by vger.kernel.org with ESMTP id S263654AbTDXNT3 (ORCPT ); Thu, 24 Apr 2003 09:19:29 -0400 To: rwhron@earthlink.net Cc: linux-kernel@vger.kernel.org, lord@sgi.com Subject: Re: [benchmarks] various filesystems on 2.5.68 References: <20030424124728.GA1477@rushmore.suse.lists.linux.kernel> From: Andi Kleen Date: 24 Apr 2003 15:31:22 +0200 In-Reply-To: <20030424124728.GA1477@rushmore.suse.lists.linux.kernel> Message-ID: User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.2 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org rwhron@earthlink.net writes: > mount options > mount -t ext2 -o defaults,noatime /dev/sdc1 /fs1 > mount -t ext3 -o defaults,noatime,data=writeback /dev/sdc1 /fs1 > mount -t reiserfs -o defaults,noatime /dev/sdc1 /fs1 > mount -t jfs -o defaults,noatime /dev/sdc1 /fs1 > mount -t xfs -o logbufs=8,logbsize=32768,noatime /dev/sdc1 /fs1 > > mkfs command > mke2fs -q /dev/sdc1 > mke2fs -q -j -J size=400 /dev/sdc1 > yes "y" | mkreiserfs /dev/sdc1 >/tmp/mkr.out 2>&1 > jfs_mkfs -q /dev/sdc1 > mkfs.xfs -l size=32768b -f /dev/sdc1 > > Very recent version of xfsprogs/jfsutils/reiserfsprogs/e2fsprogs. > > XFS mount/mkfs options came from the XFS FAQ and are very desireable > for these tests based on an earlier run without them. If anyone It's my experience too that XFS often performs badly with the default mkfs/mount options, especially for metadata intensive workloads. The drawback is that it eats memory. I've been thinking about making it set the bigger mount option by default based on the available physical memory; this would probably help a lot of users. iirc one drawback is that it eats from the vmalloc area which is only 64MB by default on 32bit. But still should not be that big an issue. On 64bit it's not an issue at all of course. Of course the log size cannot be changed this way, but perhaps the default in mkfs is just too small for today's disk sizes? -Andi