From owner-freebsd-questions Wed Sep 23 10:11:41 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id KAA03065 for freebsd-questions-outgoing; Wed, 23 Sep 1998 10:11:41 -0700 (PDT) (envelope-from owner-freebsd-questions@FreeBSD.ORG) Received: from post.mail.demon.net (post-12.mail.demon.net [194.217.242.41]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id KAA03041 for ; Wed, 23 Sep 1998 10:11:34 -0700 (PDT) (envelope-from searle@longacre.demon.co.uk) Received: from [158.152.156.24] (helo=longacre.demon.co.uk) by post.mail.demon.net with smtp (Exim 2.03 #1) id 0zLsRq-00012M-00 for questions@freebsd.org; Wed, 23 Sep 1998 17:11:27 +0000 Date: Wed, 23 Sep 1998 15:36:31 +0100 (BST) From: Michael Searle Subject: Re: mixing RAM for FreeBSD ? To: rick hamell In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII X-Organization: Who, me? X-Mailer: ANT RISCOS Marcel [ver 1.46] Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Tue 22 Sep, rick hamell wrote: > > >inaccurate, just that the thing being measured at 12ns is *not* > > >the same thing that is measured at 60ns in the SIMMS. When you > > >start comparing apples/apples, the numbers are more like 55ns vs. > > >60ns. > > > > > >Or maybe this is a false memory. > > > > You could be right and I'm more than willing to be educated. Anybody know? > > I've heard something like this too, though I don't remeber what > it was. I think it had to do with the BUS speed, hence we're seeing > "100mhz compatible memory," instead of 8ns or 10ns or whatever. Pretty > much to answer the first posters questions, it don't matter how fast > your memory is, your computer can handle it. Now... weather it handles > it with no parity problems, etc, is a measure of the quality of the > memory and your motherboard. I.e. Packard Bells are famous for having > 70ns memory, but it tended to run at 80 or 90ns pretty easily. :) > The difference is that normal SIMMs are asynchronous to the host bus, the 60ns is the minimum access time for the first word of a 4 word cache line. (the next 3 words are read at a faster rate) EDO RAM may have timings of 5-2-2-2 : 5 clocks for the first word, 2 clocks for each next word. (non- EDO RAM will be around 6-3-3-3.) Some chipsets can read teo lines at once. (5-2-2-2/2-2-2-2). SDRAMs are synchronous, so the access time measured is the clock rate (ie 10ns = 100MHz). The timing is something like 7-1-1-1 for a 66MHz SDRAM, so it is faster but not as fast as it appears to be - multiplying by 5 is about right. (10ns SDRAM ~= 50ns EDO) Faster SDRAMs are generally not much faster for reading the first word of the line, but can still read a word every cycle for the rest of the burst at a higher clock speed, which is why SDRAM speed is usually given by a frequency. 100MHz SDRAM would have timing around 10-1-1-1. (Also, as you said a lot of memory is low quality and can't really run at the specified speed: speeds given in ns are where this happens most.) For more information, see Tom's Hardware Guide (http://tomshardware.com/) -- csubl@csv.warwick.ac.uk To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message