From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Moyer Subject: Re: New driver mtipx2xx submission Date: Wed, 01 Jun 2011 15:51:41 -0400 Message-ID: References: <22A973199D2C2F46933448F6E7990A300204F2BC@ntxboimbx31.micron.com> <20110428230605.78c55c70@lxorguk.ukuu.org.uk> <22A973199D2C2F46933448F6E7990A300204F728@ntxboimbx31.micron.com> <20110502184206.25907c5e@lxorguk.ukuu.org.uk> <22A973199D2C2F46933448F6E7990A300214B0D6@ntxboimbx31.micron.com> <20110511202013.075b07cc@lxorguk.ukuu.org.uk> <4DD722DB.8030303@micron.com> <22A973199D2C2F46933448F6E7990A300239EA77@ntxboimbx31.micron.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from mx1.redhat.com ([209.132.183.28]:61193 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1759089Ab1FATvz (ORCPT ); Wed, 1 Jun 2011 15:51:55 -0400 In-Reply-To: <22A973199D2C2F46933448F6E7990A300239EA77@ntxboimbx31.micron.com> (Asai Thambi Samymuthu Pattrayasamy's message of "Wed, 1 Jun 2011 13:40:29 -0600") Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: "Asai Thambi Samymuthu Pattrayasamy (asamymuthupa) [CONTRACTOR]" Cc: Alan Cox , linux-ide@vger.kernel.org "Asai Thambi Samymuthu Pattrayasamy (asamymuthupa) [CONTRACTOR]" writes: > On 5/25/2011 8:36 AM, Jeff Moyer wrote: >> Asai Thambi S P writes: >> >>> On 5/11/2011 1:20 PM, Alan Cox wrote: >>>> So a bigger queue helped (at least in 2006). The AHCI driver can be >>>> taught your bigger queue easily enough. The question is where with a >>>> *current* kernel are any remaining bottlenecks if you do that and > how do >>>> we fix them. >>> >>> Attached image/table shows the performance numbers on current kernel. >>> >>> The main objectives of our new mtipx2xx driver are >>> 1.) highest performance (see attached image/table), >>> 2.) lowest CPU utilization, and >> >> Can you collect perf data to show why the ahci driver is taking up so >> much more CPU for the random I/O case? >> > > Collected the perf data for ahci driver. As the call graph is getting > distorted in the email, attaching the perf data call graph report. Thanks, Asai! I don't think cfq is the ideal I/O scheduler to be testing. Could you run again with deadline and/or noop and see how that changes your throughput and perf report? Also, just for completeness, could you tell us which kernel you ran this against? Thanks! Jeff