From owner-freebsd-hardware  Tue Jul 22 14:06:44 1997
Return-Path: <owner-freebsd-hardware>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id OAA28403
          for hardware-outgoing; Tue, 22 Jul 1997 14:06:44 -0700 (PDT)
Received: from Octopussy.MI.Uni-Koeln.DE (Octopussy.MI.Uni-Koeln.DE [134.95.166.20])
          by hub.freebsd.org (8.8.5/8.8.5) with SMTP id OAA28398
          for <freebsd-hardware@freebsd.org>; Tue, 22 Jul 1997 14:06:41 -0700 (PDT)
Received: from x14.mi.uni-koeln.de (annexr2-45.slip.Uni-Koeln.DE) by Octopussy.MI.Uni-Koeln.DE with SMTP id AA18251
  (5.67b/IDA-1.5 for <freebsd-hardware@FreeBSD.ORG>); Tue, 22 Jul 1997 23:02:41 +0200
Received: (from se@localhost) by x14.mi.uni-koeln.de (8.8.6/8.6.9) id XAA15146; Tue, 22 Jul 1997 23:02:40 +0200 (CEST)
X-Face: "<d]#=8pzx);RzeqSKI86OVa7=!0/(uRa.+B.9Z9\eNUn@UG?!`y7yt2dFNn%k4'.}](uE%
 yCO)$e&Y1%3EO~ifu6Q-#YUM&JZ't,}JkPnAz,8Dj33u%@GBi%[Y#LHz$]h7a<p4)-jKI7~sKjlP-^
 EvA[G;]v&0]W!EL%shs,{7x0|oqN4YVIs5,NI#,V{9"WF):5&RkOhyj*#-IAG}Tnu;YCF,d
Message-Id: <19970722230239.56696@mi.uni-koeln.de>
Date: Tue, 22 Jul 1997 23:02:39 +0200
From: Stefan Esser <se@FreeBSD.ORG>
To: freebsd-hardware@FreeBSD.ORG
Cc: owner-crp@tssc.co.nz
Subject: Re: Best Pentium / PPro
References: <3.0.32.19970721101544.006c0e1c@mailhost> <199707211016.DAA20983@silvia.HIP.Berkeley.EDU>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.74
In-Reply-To: <199707211016.DAA20983@silvia.HIP.Berkeley.EDU>; from Satoshi Asami on Mon, Jul 21, 1997 at 03:16:22AM -0700
Sender: owner-freebsd-hardware@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

* Any comments with DIMM (10ns) vs EDO SIMM (60ns) memory, it seems like the
* 10ns DIMM is not worth the extra money due to the PCI bus speed ??

Be careful with those numbers!

The SDRAM modules are not *that* much faster than EDO,
in reality. The reason is, that while the 60ns specified
for "normal" DRAMs is the row access time (time from
the row address being stable and clocked into the RAM
chip) until data is available, while the 10ns in case
of the SDRAM is the data rate at which you can read 
further values after waiting much longer (say 60ns :)
for the first data to arrive ...

In fact, SDRAM seems to allow 5-1-1-1 clock reads at
66MHz, while EDO-RAM needs 4-2-2-2. This means that on 
average 5% of the memory reads that are not covered by
the secondary cache are going to take 8 instead of 10
clocks with SDRAM. You will see a difference, if your
application accesses large blocks of memory (much more
than fits into the cache).

In typical system benchmarks (CPU bound) SDRAM seems
to give at most a 1% to 2% improvement over EDO.

Regards, STefan