From owner-freebsd-hackers Sun Feb 24 7: 1:29 2002 Delivered-To: freebsd-hackers@freebsd.org Received: from chiark.greenend.org.uk (chiark.greenend.org.uk [212.22.195.2]) by hub.freebsd.org (Postfix) with ESMTP id 2C0A037B405 for ; Sun, 24 Feb 2002 07:01:23 -0800 (PST) Received: from andrewm by chiark.greenend.org.uk with local (Exim 3.12 #2) id 16f09N-0002w9-00 (Debian); Sun, 24 Feb 2002 15:01:17 +0000 From: Andrew Mobbs MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <15481.61.57511.222531@chiark.greenend.org.uk> Date: Sun, 24 Feb 2002 15:01:17 +0000 (GMT) To: Matthew Dillon Cc: hackers@FreeBSD.ORG Subject: Re: Test patch for msync/object-flushing performance (for stable) In-Reply-To: <200202222201.g1MM1f431236@apollo.backplane.com> References: <15478.31998.459219.178549@chiark.greenend.org.uk> <200202222042.g1MKg4u22700@apollo.backplane.com> <200202222201.g1MM1f431236@apollo.backplane.com> X-Mailer: VM 6.72 under 21.1 (patch 10) "Capitol Reef" XEmacs Lucid Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Matthew Dillon writes: > Ok, here is a test patch. Now, there are some instructions to go along > with this patch, so continue reading. > > I have implemented two optimizations. You can turn either or both (or > neither) on with a sysctl. I would like those interested to test all > four combinations. Be sure to delete any test files and 'sync' a couple > of times between each test run so you do not skew the results. Thanks for the patch. Test results from an overnight run, following the above instructions. Unfortunatly, it looks like there was a little cron activity from time-to-time that interfered with a couple of datapoints, but the general trend is quite clear: The kernel was -STABLE, cvsupped about 18:00 GMT yesterday, and patched. Results are time in seconds for completing a standard workload. Each test was repeated five times. vm.msync_flush_flags | 0 | 1 | 2 | 3 | -------+-------+-------+-------+-------| write | 519 | 517 | 1632 | 519 | sync | 2227 | 176 | 848 | 177 | -------+-------+-------+-------+-------| write | 514 | 517 | 518 | 516 | sync | 2215 | 175 | 2219 | 176 | -------+-------+-------+-------+-------| write | 511 | 649 | 517 | 513 | sync | 2209 | 125 | 2223 | 176 | -------+-------+-------+-------+-------| write | 514 | 518 | 515 | 517 | sync | 2217 | 176 | 2209 | 176 | -------+-------+-------+-------+-------| write | 516 | 516 | 517 | 518 | sync | 2219 | 176 | 2222 | 177 | Nearly 13 times improvement in sync times, very nice. I also have vmstat logs for each test run, if you're interested (taken at a 10 second interval): (51kB) -- Andrew Mobbs - http://www.chiark.greenend.org.uk/~andrewm/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message