Path: utzoo!utgpu!attcan!uunet!lll-winken!lll-tis!ames!nrl-cmf!mailrus!cornell!uw-beaver!rice!sun-spots-request
From: dvorak@iam.unibe.ch (Jiri Dvorak)
Newsgroups: comp.sys.sun
Subject: Re: SMD-4 vs. 7053
Message-ID: <183:dvorak@iam.unibe.ch>
Date: 15 Dec 88 21:50:05 GMT
Sender: usenet@rice.edu
Organization: Rice University, Houston, Texas
Lines: 19
Approved: Sun-Spots@rice.edu
Original-Date: 6  Dec 88 10:55 +0100
X-Sun-Spots-Digest: Volume 7, Issue 58, message 4 of 14

Many thanks to all who sent a response to my question on the differences
between the new Sun controller SMD-4 and the xy7053. The most detailed
response I got from 

>  Brian Wong - TSE Sun Washington DC <bwong@sun.com>

>  The SMD-4 is similar to the Xylogics 753; however it has radically different
>  proms which cause significantly different caching behaviour.  The (silicon)
>  logic is reasonably close.  The difference is that the 753 just takes huge
>  blocks and pre-reads, while the SMD-4 actually utilizes a demand-replacement
>  algorithm.  Performance is notably different for some kinds of applications
>  (like servers for diskless clients).

The hardware and software requirements for a SMD-4 were already discussed
in earlier issues of sun-spots.

regards,
Jiri Dvorak                  dvorak@iam.unibe.ch   or
University of Berne          dvorak%iam.unibe.ch@relay.cs.net
Switzerland