From owner-freebsd-amd64@FreeBSD.ORG Sat Jan 28 10:12:24 2006 Return-Path: X-Original-To: freebsd-amd64@freebsd.org Delivered-To: freebsd-amd64@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 8388316A420 for ; Sat, 28 Jan 2006 10:12:24 +0000 (GMT) (envelope-from lars.tunkrans@bredband.net) Received: from mxfep04.bredband.com (mxfep04.bredband.com [195.54.107.79]) by mx1.FreeBSD.org (Postfix) with ESMTP id 6C8D343D46 for ; Sat, 28 Jan 2006 10:12:22 +0000 (GMT) (envelope-from lars.tunkrans@bredband.net) Received: from mxfep04 ([195.54.107.79] [195.54.107.79]) by mxfep04.bredband.com with SMTP id <20060128101221.ZTBV8741.mxfep04.bredband.com@mxfep04>; Sat, 28 Jan 2006 11:12:21 +0100 X-Mailer: Openwave WebEngine, version 2.8.18 (webedge20-101-1108-20050216) From: To: Daniel Rock Date: Sat, 28 Jan 2006 11:12:21 +0100 MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Message-Id: <20060128101221.ZTBV8741.mxfep04.bredband.com@mxfep04> Cc: freebsd-amd64@freebsd.org Subject: Re: Re: dual vs single core opteron 100's X-BeenThere: freebsd-amd64@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Porting FreeBSD to the AMD64 platform List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 28 Jan 2006 10:12:24 -0000 > > From: Daniel Rock > We are talking about Dual-Core Opteron 1xx. Current Opteron 1xx (beginning > with E-Stepping) use Socket 939 - just like Athlon64: > > http://www.amdcompare.com/us-en/opteron/ > http://www.amdcompare.com/us-en/desktop/ > > Athlon64 and current Opteron 1xx are the same. They even share the number of > HTT links: 1 (non coherent) - only Socket 940 Opterons provide 3 HTT links. > > Because of the integrated memory controller supporting or not supporting ECC > isn't a function of the chipset, but of the CPU itself (and the BIOS > initializing ECC right). Every Athlon64 (Socket 754, 939, 940) does support > ECC. The only difference is, that Socket 940 requires registered DIMMS while > the other ones work with unbuffered DIMMs. > > Conclusion: Apart from different CPUID the following CPUs are *exactly the same*: > > Athlon64 FX-55 Opteron 152 > Athlon64 FX-57 Opteron 154 > Athlon64 X2 4400+ Opteron 175 > Athlon64 X2 4800+ Opteron 180 O.K. Seems to defeat the purpose , If you want to build a reliable Server you want Registred ECC RAM. ( socket 940 ) If you want to build a cheap desktop machine you want un-registred non-ECC RAM. ( socket 939 ) Only other reason I can think of to have two brand names for the same product is to sell it at different price levels, but AMD generally seems to have the same price for the same clockrate on these chips. Only application I can think of for using non-reliable servers built with socket 939 is compute clusters where you have several hundred compute servers, and you are not dependent on whether an individual server runs all the time. //Lars