From owner-freebsd-stable@FreeBSD.ORG Thu Apr 6 19:30:08 2006 Return-Path: X-Original-To: freebsd-stable@freebsd.org Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 1EA9216A485 for ; Thu, 6 Apr 2006 19:30:08 +0000 (UTC) (envelope-from peterjeremy@optushome.com.au) Received: from mail02.syd.optusnet.com.au (mail02.syd.optusnet.com.au [211.29.132.183]) by mx1.FreeBSD.org (Postfix) with ESMTP id E226343D66 for ; Thu, 6 Apr 2006 19:29:54 +0000 (GMT) (envelope-from peterjeremy@optushome.com.au) Received: from turion.vk2pj.dyndns.org (c220-239-19-236.belrs4.nsw.optusnet.com.au [220.239.19.236]) by mail02.syd.optusnet.com.au (8.12.11/8.12.11) with ESMTP id k36JTpEZ011256 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Fri, 7 Apr 2006 05:29:52 +1000 Received: from turion.vk2pj.dyndns.org (localhost.vk2pj.dyndns.org [127.0.0.1]) by turion.vk2pj.dyndns.org (8.13.4/8.13.4) with ESMTP id k36JTpEC002870; Fri, 7 Apr 2006 05:29:51 +1000 (EST) (envelope-from peter@turion.vk2pj.dyndns.org) Received: (from peter@localhost) by turion.vk2pj.dyndns.org (8.13.4/8.13.4/Submit) id k36JToSR002869; Fri, 7 Apr 2006 05:29:50 +1000 (EST) (envelope-from peter) Date: Fri, 7 Apr 2006 05:29:50 +1000 From: Peter Jeremy To: Pete French Message-ID: <20060406192950.GE700@turion.vk2pj.dyndns.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: X-PGP-Key: http://members.optusnet.com.au/peterjeremy/pubkey.asc User-Agent: Mutt/1.5.11 Cc: freebsd-stable@freebsd.org Subject: Re: Pros and Cons of amd64 (versus i386). X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 06 Apr 2006 19:30:08 -0000 On Thu, 2006-Apr-06 15:38:20 +0100, Pete French wrote: >I was thinking of moving this to amd64, but was kind of put off by results >from a test system I setup using an Athlon 64 3700+ to talk to this >machine. The opteron box is currently running 6.1-PRE/i386, and the 3700 is >runiing either Windows XP or 61-PRE/amd64. Under Windows I can completely >saturate the ether comming in, and get 70% bandwidth going out (it's gig >ether). Under amd64 on the client end I can only get about 55% utilisation >in both directions. This surprised me a lot as when I was running i386 >on that box it was always faster. Of course a number of variables have >changed since then (primarily moving from a broadcom gigabit card to using >the onboard realtek card), but I was concened that the difference was due to >the 64 bit operating system, as opposed to superior windws drivers, which >seemed unlikely! If you want to do a valid benchmark, you really need to use the same hardware for the testing. amd64 is not necessarily faster than i386 on the same hardware: On the downside: - amd64 has more, larger registers so context switching is slower - 64-bit long/pointer and lower code density means larger working set size and lower cache effectiveness On the upside: - more registers means less register spills (better code) - I think there's more efficient support for PIC code - 64-bit arithmetic means faster multi-precision math (think RSA/DSA/SSL) - access to >2GB virtual address space >I know that what I should do is install i386 on the client and test again, but >doing that will lose my only 64 bit environment so I am loathe to do so. Any >comments ? Backup your amd64 environment and install i386. You can re-install the amd64 once the testing is finished. The best benchmark is always your own application. -- Peter Jeremy