From owner-freebsd-stable Fri Aug 13 17:46: 1 1999 Delivered-To: freebsd-stable@freebsd.org Received: from gndrsh.aac.dev.com (GndRsh.aac.dev.com [198.145.92.4]) by hub.freebsd.org (Postfix) with ESMTP id 1A38814EA1 for ; Fri, 13 Aug 1999 17:45:57 -0700 (PDT) (envelope-from rgrimes@gndrsh.aac.dev.com) Received: (from rgrimes@localhost) by gndrsh.aac.dev.com (8.9.3/8.9.3) id RAA02542; Fri, 13 Aug 1999 17:46:02 -0700 (PDT) (envelope-from rgrimes) From: "Rodney W. Grimes" Message-Id: <199908140046.RAA02542@gndrsh.aac.dev.com> Subject: Re: On freezes in 3.2-Stable (fwd) In-Reply-To: <199908132342.QAA01905@dingo.cdrom.com> from Mike Smith at "Aug 13, 1999 04:42:04 pm" To: mike@smith.net.au (Mike Smith) Date: Fri, 13 Aug 1999 17:46:01 -0700 (PDT) Cc: freebsd-stable@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL54 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > > I didn't realize this person asked this off list, so I am forwarding > > my reply onto the list, it has information in it that someone should > > use to write a FAQ/FSA, this issue has come up some many times over > > the years that I am getting sick of reading it every time memory > > generations change. > > > > Basic rule, Don't ever expect to be able to drive more than 72 DRAM > > chips with anything in the PC world, period. You have to do bus > > buffering using balanced drivers and real clean layout work on the PC Board, > > and there ain't a PC maker out there who is going to bother with > > this unless they are supporting 8 sockets or more in high dollar ^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^ > > server machines. ^^^^^^^^^^^^^^^^ > > Er. The Intel AD450NX has 32 DIMM sockets. Unless someone starts > making 2-chip DIMMs, I don't see how you would run 8GB in this box (and > I have seen it being done under other operating systems). The Intel AD450NX in in the class ``supporting 8 sockets or more''. The Intel AD450NX is in the class ``high dollar server machines''. > > You might want to qualify the issue a little further; specifically with > regard to "typical memory controllers", logic families and fanout. I think I qualified it plenty enough... As far as logic families and fanout, well they really don't matter much with regards to this issue, memory drivers are special, not made with the same configuration that the logic family is, and thus it's fanout is quite different. The constant has been 72 for a very long time, since I started duing mos memory system designs with 54xx/74xx TTL in the 80's until today with the latest in CMOS and BiCMOS. Yea.. I could design a supper honking driver that can drive 288 chips in the memory controller, but then all the RAM chips would have to have larger data drivers to handle that side of the bus. It ain't changed in 20 years, and it ain't likely to change, infact if anything it is going to decrease as f goes up. RAMBUS is already starting to push some very tough real world physical limits. -- Rod Grimes - KD7CAX - (RWG25) rgrimes@gndrsh.dnsmgr.net To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message