From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262433AbTJTIMb (ORCPT ); Mon, 20 Oct 2003 04:12:31 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262432AbTJTIMb (ORCPT ); Mon, 20 Oct 2003 04:12:31 -0400 Received: from dyn-ctb-210-9-246-89.webone.com.au ([210.9.246.89]:46087 "EHLO chimp.local.net") by vger.kernel.org with ESMTP id S262434AbTJTIMX (ORCPT ); Mon, 20 Oct 2003 04:12:23 -0400 Message-ID: <3F9398DB.8030004@cyberone.com.au> Date: Mon, 20 Oct 2003 18:12:11 +1000 From: Nick Piggin User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4) Gecko/20030827 Debian/1.4-3 X-Accept-Language: en MIME-Version: 1.0 To: Andrew Morton CC: rwhron@earthlink.net, linux-kernel@vger.kernel.org Subject: Re: [BENCHMARK] I/O regression after 2.6.0-test5 References: <20031020003745.GA2794@rushmore> <3F933BE7.5080700@cyberone.com.au> <20031019215259.7b1c7a01.akpm@osdl.org> In-Reply-To: <20031019215259.7b1c7a01.akpm@osdl.org> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Andrew Morton wrote: >Nick Piggin wrote: > >> >> >>rwhron@earthlink.net wrote: >> >> >>>There was about a 50% regression in jobs/minute in AIM7 >>>database workload on quad P3 Xeon. The CPU time has not >>>gone up, so the extra run time is coming from something >>>else. (I/O or I/O scheduler?) >>> >>>tiobench sequential reads has a significant regression too. >>> >>>Regression appears unrelated to filesystem type. >>> >>>dbench was not affected. >>> >>>The AIM7 was run on ext2. >>> >>> >>Yeah I'd say its all due to the IO scheduler. There is a problem >>I'm thinking about how to fix - its the likely cause of this too. >> >> > >What change do you think it was due to? > > I was thinking: [PATCH] fix AS crappy performance (It still doesn't work properly) >It's rather strange that test6 is slow but test6-mm is not: generally the >IO scheduler regressions enter -mm first ;) > But if test6-mm isn't slow then maybe it is due to something else