Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!ihnp4!zehntel!hplabs!sdcrdcf!sdcsvax!dcdwest!ittvax!decvax!genrad!mit-eddie!godot!ima!pbear!peterb
From: peterb@pbear.UUCP
Newsgroups: net.arch
Subject: Re: Alternate Computer Architectures (Na
Message-ID: <31@pbear.UUCP>
Date: Tue, 29-Jan-85 04:50:13 EST
Article-I.D.: pbear.31
Posted: Tue Jan 29 04:50:13 1985
Date-Received: Sat, 2-Feb-85 14:48:14 EST
Lines: 25
Nf-ID: #R:marlin:-59600:pbear:18800001:000:1001
Nf-From: pbear!peterb    Jan 28 12:55:00 1985


	Mike,

In optical computers, There has been quite abit of work on adapting some
of the refractive properties of lenses to do matrix multiplications.

	I remember reading an article recently (I don't remember the name or
the publication) in where a company is attempting to produce an optical
matrix multiplier of 32 bit precision that operates as an attached processor.
The speed of this thing is enormous. from the time that the registers for a
100 x 100 element matrix is loaded, the result is available for unloading in
under 1 microsecond.

	It was estimated that something faster than a cray-1 would be needed
as an input/output processor in order to take advantage of the speed of this
AP. (And I thought is was bad that you needed a 370 to work as to i/o
processor for a cray-1!)

	As for the oterh types of optical computers, my memory is too fuzzy
in roder to be of any help in this department.

					Peter Barada
					ima!pbear!peterb

"...where the antelopes and teddy bears playyyy...."