From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from merlin.infradead.org ([205.233.59.134]:44324 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752524Ab3H1REc (ORCPT ); Wed, 28 Aug 2013 13:04:32 -0400 Message-ID: <521E2D91.80600@kernel.dk> Date: Wed, 28 Aug 2013 11:04:17 -0600 From: Jens Axboe MIME-Version: 1.0 Subject: Re: filling buffers with data from block device References: In-Reply-To: Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: fio-owner@vger.kernel.org List-Id: fio@vger.kernel.org To: Alket Memushaj Cc: fio@vger.kernel.org On 08/27/2013 05:15 PM, Alket Memushaj wrote: > is it possible to have fio fill the I/O buffers with data from a block > device? when testing systems with both deduplication and compression > enabled, it would be easier to fill the buffers with existing data. It would not be impossible. But if you look at how fio allocates and uses it's IO buffers, then you would need to continually fill them before use since it reuses the same amount of them (defined by iodepth * jobs). It might be more feasible to analyze the data contents, and then improve fio to be able to express that more easily. -- Jens Axboe