From chemistry-request@ccl.net Tue May 17 14:15:20 2005 Received: from sigma957.cis.mcmaster.ca (sigma957.CIS.McMaster.CA [130.113.64.83]) by server.ccl.net (8.13.1/8.13.1) with ESMTP id j4HIFG8A003530 for ; Tue, 17 May 2005 14:15:17 -0400 Received: from coffee.psychology.mcmaster.ca (coffee.Psychology.McMaster.CA [130.113.218.59]) by sigma957.cis.mcmaster.ca (8.12.11/8.12.11) with ESMTP id j4HHCBti025888; Tue, 17 May 2005 13:12:12 -0400 (EDT) Received: by coffee.psychology.mcmaster.ca (Postfix, from userid 502) id 083328D74; Tue, 17 May 2005 13:11:41 -0400 (EDT) Received: from localhost (localhost [127.0.0.1]) by coffee.psychology.mcmaster.ca (Postfix) with ESMTP id 064C58B1D; Tue, 17 May 2005 13:11:41 -0400 (EDT) Date: Tue, 17 May 2005 13:11:40 -0400 (EDT) From: Mark Hahn X-X-Sender: hahn^at^coffee.psychology.mcmaster.ca To: Mikael Johansson Cc: Luigi Cavallo , Subject: Re: CCL:Summary Opteron vs Nocona In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-PMX-Version-Mac: 4.7.1.128075, Antispam-Engine: 2.0.3.2, Antispam-Data: 2005.5.17.18 X-PerlMx-Spam: Gauge=IIIIIII, Probability=7%, Report='__CT 0, __CT_TEXT_PLAIN 0, __HAS_MSGID 0, __MIME_VERSION 0, __SANE_MSGID 0' X-Spam-Status: No, score=0.0 required=5.0 tests=none autolearn=failed version=3.0.3 X-Spam-Checker-Version: SpamAssassin 3.0.3 (2005-04-27) on server.ccl.net > > However, there's one point about opterons. AMD just announced dual-core cpus, > > which should perform far better than the single-core shipped now. for cache-friendly applications that don't mind somewhat lower clocks. > whatever time period of choice :-) I don't think it in general is worth > waiting for the next family of CPU's. Instead, get what is available when it depends on how advantageous the change is. for instance, if you are heat or footprint-constrained in your machineroom, AMD's dual-core might well be worth waiting for, since they are compact and noticably cooler than two single-core chips. > price/performance ratio of dual cores would be, at least in the beginning, > when counting in all the costs of motherboards etc. if you have cache-friendly code, dual-core is very effective. other components are basically the same, therefore no price premium. AFAIKT, AMD hasn't announced DC pricing yet: http://www.amd.com/us-en/Weblets/0,,7832_8366_609,00.html the premium for 8xx chips is noticably higher than twice the price of the same-clock single-core. but of course you save with fewer motherboards, chassis, etc. > Also, after the dual core chips ship, there will be a lag before proper > drivers/kernel patches for Linux appear. It took a year after release no, DC is such a minor architectural change that it's already supported. regards, mark hahn. sharcnet/mcmaster