From owner-freebsd-fs@FreeBSD.ORG Fri Nov 5 07:57:10 2010 Return-Path: Delivered-To: freebsd-fs@FreeBSD.ORG Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 117B1106566C for ; Fri, 5 Nov 2010 07:57:10 +0000 (UTC) (envelope-from olli@lurza.secnetix.de) Received: from lurza.secnetix.de (lurza.secnetix.de [IPv6:2a01:170:102f::2]) by mx1.freebsd.org (Postfix) with ESMTP id 843538FC15 for ; Fri, 5 Nov 2010 07:57:09 +0000 (UTC) Received: from lurza.secnetix.de (localhost [127.0.0.1]) by lurza.secnetix.de (8.14.3/8.14.3) with ESMTP id oA57uqBl030091; Fri, 5 Nov 2010 08:57:08 +0100 (CET) (envelope-from oliver.fromme@secnetix.de) Received: (from olli@localhost) by lurza.secnetix.de (8.14.3/8.14.3/Submit) id oA57uqdW030090; Fri, 5 Nov 2010 08:56:52 +0100 (CET) (envelope-from olli) Date: Fri, 5 Nov 2010 08:56:52 +0100 (CET) Message-Id: <201011050756.oA57uqdW030090@lurza.secnetix.de> From: Oliver Fromme To: freebsd-fs@FreeBSD.ORG, freebsd@jdc.parodius.com In-Reply-To: <201011041823.oA4INTIL097844@lurza.secnetix.de> X-Newsgroups: list.freebsd-fs User-Agent: tin/1.8.3-20070201 ("Scotasay") (UNIX) (FreeBSD/6.4-PRERELEASE-20080904 (i386)) MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.3.5 (lurza.secnetix.de [127.0.0.1]); Fri, 05 Nov 2010 08:57:08 +0100 (CET) Cc: Subject: Re: Using an SSD "disk" for / X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list Reply-To: freebsd-fs@FreeBSD.ORG, freebsd@jdc.parodius.com List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 05 Nov 2010 07:57:10 -0000 Oliver Fromme wrote: > With /usr/src and /usr/obj both on the same SATA-300 HDD, > a buildworld takes 10 minutes and 13 seconds. > > With /usr/src and /usr/obj on separate memory disks, > a buildworld takes 10 minutes and 4 seconds. > > This is stable/8 with empty /etc/src.conf, i.e. a complete > buildworld. The machine is an AMD Phenom II X6 (hexacore). I think I should give some more information about the system that I used for the test. The HDD is accessed via ahci(4) and ada(4) with NCQ. This might make a small difference, though I haven't benchmarked it. src and obj are mounted with noatime. In the memory disk case, src was even mounted read-only, and obj was mounted async. I made sure that there was no swapping / page-out activity during the memory disk test, so all the data really stayed in RAM (which should be even faster than an SSD, I assume). The system was otherwise idle, gstat(8) basically showed nothing on ada0 during the memory disk test. The following sysctl tunings are in effect: vfs.hirunningspace=8388608 vfs.lorunningspace=1048576 vfs.read_max=32 vfs.ufs.dirhash_maxmem=33554432 kern.sched.preempt_thresh=220 (I've got more things in /etc/sysctl.conf, but the rest shouldn't affect file system performance or buildworld, I think.) The kernel is running with HZ=300 (unfortunately HZ isn't dynamic in FreeBSD, so I'm using 300 as a trade-off). AUDIT and MAC are disabled. Best regards Oliver -- Oliver Fromme, secnetix GmbH & Co. KG, Marktplatz 29, 85567 Grafing b. M. Handelsregister: Registergericht Muenchen, HRA 74606, Geschäftsfuehrung: secnetix Verwaltungsgesellsch. mbH, Handelsregister: Registergericht Mün- chen, HRB 125758, Geschäftsführer: Maik Bachmann, Olaf Erb, Ralf Gebhart FreeBSD-Dienstleistungen, -Produkte und mehr: http://www.secnetix.de/bsd "When your hammer is C++, everything begins to look like a thumb." -- Steve Haflich, in comp.lang.c++