From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from 173-166-109-252-newengland.hfc.comcastbusiness.net ([173.166.109.252]:50959 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756216Ab3BUSgn (ORCPT ); Thu, 21 Feb 2013 13:36:43 -0500 Received: from brick.kernel.dk ([87.104.106.3] helo=kernel.dk) by bombadil.infradead.org with esmtpsa (Exim 4.76 #1 (Red Hat Linux)) id 1U8b0c-0006Ux-Tb for fio@vger.kernel.org; Thu, 21 Feb 2013 18:36:43 +0000 Received: from localhost (nelson.home.kernel.dk [192.168.0.33]) by kernel.dk (Postfix) with ESMTP id 553445637E for ; Thu, 21 Feb 2013 19:36:41 +0100 (CET) Date: Thu, 21 Feb 2013 19:35:09 +0100 From: Jens Axboe Subject: Fio 2.0.14 has been tagged Message-ID: <20130221183509.GY25617@kernel.dk> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Sender: fio-owner@vger.kernel.org List-Id: fio@vger.kernel.org To: fio@vger.kernel.org Hi, The 2.0.14 release is now finally out. I knew this was going to be a longer cycle, mainly due to the planned introduction of a hand rolled configure script. This makes it easier to support a variety of platforms, or even the same platform in different versions. It went a lot easier than expected, let me know if you run into build related issues due to this. But I do believe I've checked basically all platforms. In general, if it builds and works on AIX, it must be perfect. Of new notable features, the idle profiling is pretty nifty. The idea is to use any available cycles in the system, showing how much oomph would be leftover when running a particular workload. For now it wont work very well with hyperthreading (or SMT in general), though, since load on one thread will impact the other due to the shared execution units. But for non-SMT, it works really well. See setion 9 of the HOWTO. Another feature that will improve the experience, is a new verify option that greatly reduces the memory required for doing a verify. For a classic verify, fio will store a little piece info on each written block. It's a small piece, but for big (or many) devices, it adds up and has been seen to cause OOM issues in the wild. That's particularly annoying since it will happen many hours into the run. The option is not documented yet, but it's experimental_verify. For 2.0.15, it will be officially supported, but it should work fine as-is. Basically fio will rewind any needed states when the write phase has finished, then replay the same part doing the verify. Apart from that, there's been a bigger-than-usual number of great bug fixes. So I would encourage any user to upgrade to 2.0.14, and any packager to do the same. Of notable commit stats, we have 12 different committers this time, with a total of 136 commits and: 112 files changed, 3741 insertions(+), 2174 deletions(-) Thanks! -- Jens Axboe