From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from merlin.infradead.org ([205.233.59.134]:44027 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752994Ab3BUIkR (ORCPT ); Thu, 21 Feb 2013 03:40:17 -0500 Date: Thu, 21 Feb 2013 09:38:43 +0100 From: Jens Axboe Subject: Re: Random vs Sequential Read in FIO Message-ID: <20130221083843.GD25617@kernel.dk> References: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Sender: fio-owner@vger.kernel.org List-Id: fio@vger.kernel.org To: David N Cc: fio@vger.kernel.org On Thu, Feb 21 2013, David N wrote: > Hi, > > I was using FIO to get response time of random reads and sequential > reads from phone's flash memory and I got like 9x greater response > time for random read. Since there is no seek time as in conventional > rotating storage, I'm wondering where this huge jump of random IO > response time comes from? How is the total IO delay actually > calculated in FIO? One option is that your sequential reads are coalesced and issued as one command, whereas random flash reads will still trigger a tread and tstream pause for each one. You need to look at device stats with the workload to see what the ratio between fio-request:device-request is. -- Jens Axboe