Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!iuvax!pur-ee!uiucdcs!uxc.cso.uiuc.edu!ccvaxa!aglew
From: aglew@ccvaxa.UUCP
Newsgroups: comp.arch
Subject: Re: OS co-processor ??
Message-ID: <28200059@ccvaxa>
Date: Tue, 27-Oct-87 21:32:00 EST
Article-I.D.: ccvaxa.28200059
Posted: Tue Oct 27 21:32:00 1987
Date-Received: Sat, 31-Oct-87 01:44:15 EST
References: <686@its63b.ed.ac.uk>
Lines: 41
Nf-ID: #R:its63b.ed.ac.uk:686:ccvaxa:28200059:000:2240
Nf-From: ccvaxa.UUCP!aglew    Oct 27 20:32:00 1987


..> adam@its63b.ed.ac.uk writes about ICL's CAFS, a disk controller
..> which filters data as it comes off the disk, before passing it
..> to the main processor.

Pardon me for a bit of advertisement, but Gould also sells such a device,
called a HYPERSEARCH board, which filters data as the disk revolves.
It can be used under UNIX, and has great potential for grepping on
documentation :-).
    As a matter of fact, it may have been done by the same Scottish
company that did ICL's board.

These are not, however, content addressible memories in the sense of
cache controllers and the AMD file store. They are, rather, filters on
the rapid flow of data off the disk, that reduce CPU involvement in
processing. Although they probably contain some associative component,
similar devices (for PCs, for example), do not, and still have advantages.

There seems to be a transatlantic difference in the meaning ascribed to
"associative memory" or "content addressible memory" - or maybe the
difference is between subfields of computer engineering. I usually think
of an associative memory as one that you present part of a data item to,
and it returns the rest of the data item - with parallel match circuitry,
not with a microsequencer going and doing a sequential or hashed search
on the internal representation. Papers from Europe or in databases tend
to talk about associative memories as being the black box - without
caring about how the internal search is done.
	Neither definition is right or wrong. It's just useful to know
what somebody is thinking of when he says "content addressible memory".
I know that I nearly flipped when I read a Japanese article talking about
a 64M-element associative memory...


Andy "Krazy" Glew. Gould CSD-Urbana.    USEnet:  ihnp4!uiucdcs!ccvaxa!aglew
1101 E. University, Urbana, IL 61801    ARPAnet: aglew@gswd-vms.arpa

I always felt that disclaimers were silly and affected, but there are people
who let themselves be affected by silly things, so: my opinions are my own,
and not the opinions of my employer, or any other organisation with which I am
affiliated. I indicate my employer only so that other people may account for
any possible bias I may have towards my employer's products or systems.