Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!husc6!yale!lisper-bjorn
From: lisper-bjorn@CS.YALE.EDU (Bjorn Lisper)
Newsgroups: comp.arch
Subject: Re: Lightning Fast Photonic Optical Computers
Keywords: optical, photonic, supercomputers, switching
Message-ID: <31058@yale-celray.yale.UUCP>
Date: 8 Jun 88 20:30:29 GMT
References: <1933@uhccux.uhcc.hawaii.edu>
Sender: root@yale.UUCP
Reply-To: lisper-bjorn@CS.YALE.EDU (Bjorn Lisper)
Organization: Yale University Computer Science Dept, New Haven CT  06520-2158
Lines: 52

In article <1933@uhccux.uhcc.hawaii.edu> garvin@uhccux.UUCP (Jay Garvin)
writes:
>How many of you have heard of photonic switching? ....

I'm not an expert but I do have some impressions of this. I'll comment below
on the points where I have something to say.

>2.  Should we really be letting photons do our computing for us?

I see no moral reasons to say no. Photons should have to work as well as
electrons. :-)

>    What are the forseeable problems with this technology?

Size. I think the size of an optical switch is in the millimeter range right
now whereas electronic switches are in the micrometer and sub-micrometer range.
This is absolutely necessary to improve if optical computing is to have any
substantial impact.

>    What types of things are promising about photonic switching besides
>      the speed?

Communication seems easier to facilitate. In electronics wiring is a
problem, long wires occupy precious area and slow down the circuit. Thus you
want to connect physically close cells only. In optics the speed of light is
the limit.  I've seen a proposal how to arrange a parallel optical computer
where the processors are in a plane. Over them is a mirror and they
communicate by sending a laser beam so that it bounces off the mirror to the
right recepient.  In this communication scheme all processors are equally
close w.r.t. communication regardless of physical proximity.

Optical computers will be more resistant to electrical disturbances such as
static electricity, solar wind bursts and EMP. This is by the way the reason
fibre optics is used today for computer communication in electrically noisy
environments such as factories.

I've heard that the same optical switch actually can switch signals carried
by light of different frequencies simultaneously. If this is true it will
open new possibilities for parallelity, where computations can take place
simultaneously not only in different processors but actually in *the same*
one.

>3.  How long before one of these things gets off the drawing board and
>      into production?

Some years ago a Californian company by the name of, I think, Guiltech, was
developing a systolic array type parallel processor with optical computing
elements. Apparently they dropped the project since nothing has been heard
since. I don't think they dropped it for technical reasons, however, but
rather since the price/performance quotient wasn't good enough.

Bjorn Lisper