From owner-freebsd-stable Wed Aug 11 15:47:34 1999 Delivered-To: freebsd-stable@freebsd.org Received: from relay.ripco.com (relay.ripco.com [209.100.227.3]) by hub.freebsd.org (Postfix) with SMTP id 478CA14BF2 for ; Wed, 11 Aug 1999 15:47:26 -0700 (PDT) (envelope-from aphor@ripco.NOSPAM.com) Received: (qmail 7540 invoked from network); 11 Aug 1999 22:46:44 -0000 Received: from 228-121.ppp.ripco.net (HELO ripco.NOSPAM.com) (@209.100.228.121) by relay.ripco.com with SMTP; 11 Aug 1999 22:46:44 -0000 Message-ID: <37B1FD73.C8DDB723@ripco.NOSPAM.com> Date: Wed, 11 Aug 1999 17:47:15 -0500 From: Jeremy McMillan Reply-To: aphor@ripco.com Organization: Loose.. X-Mailer: Mozilla 4.6 [en] (X11; I; FreeBSD 3.2-STABLE i386) X-Accept-Language: en MIME-Version: 1.0 To: FreeBSD-Stable@FreeBSD.ORG Subject: Re: On freezes in 3.2-Stable References: <21100.934391386@localhost> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG In 5 years of tweaking PC hardware, I've found you get what you pay for. The thing with memory and buses is all timing. More expensive motherboards will either come burned-in with RAM, support only certain (premium priced) DIMMS, or they will sacrifice a nominal amount of performance to put extra buffer chips between memory modules to isolate the electrical interfaces of the RAM from the actual memory bus. The sad part of this is QC is usually "sampling" parts off the production line. That means they have a margin for error on the parts that get shipped. Usually, the better the replacement policy is, the less sampling they do, and the bigger the number of bad parts end up in customers' hands. On top of that, you may belong to the 5% of the population that has worse operating conditions than their QC test bench, thus even if you get a burned-in part you could still be out of luck. The more of a power-user you are, the further out on the statistical curve you get. Please remember, we are dealing with manufacturers and vendors who make their money on *commodity* hardware, which is usually never tested to (though it may have been designed for) each checkbox on the spec sheet. Wise men have no time for bad PC hardware. "Jordan K. Hubbard" wrote: > > I'm virtually certain thta this is a bogus motherboard. I know of a > number of systems running 1GB of memory that are perfectly stable > (including the quad processor Xeon at WC we use for testing) just as I > know of several situations in the past were a motherboard completely > fell over after having all of its SIMM sockets populated. As someone > else noted, there are often electrical limitations which present > themselves only when you stuff the board full of RAM. It would be an > easy test to simply swap this motherboard and move the memory back in. > > Also, how did you test the memory? With a hardware tester, I hope, > since all software memory tests are inherently bogus. Don't trust > them to tell you anything more than the fact that there's power to the > memory. :) > > - Jordan > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-stable" in the body of the message -- PLEASE NOTICE: THERE MAY BE NOSPAM IN THE HEADERS WHEN YOU HIT "REPLY"!!! Jeremy McMillan | Ask for PGP-2.6.2 or 5.0i Chicago FreeBSD Users Group http://pages.ripco.com/~aphor/ChiFUG.html To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message