Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxn!ihnp4!houxm!hjuxa!petsd!peora!jer
From: jer@peora.UUCP (J. Eric Roskos)
Newsgroups: net.arch
Subject: Re: talk about performance...
Message-ID: <2122@peora.UUCP>
Date: Tue, 29-Apr-86 09:10:20 EDT
Article-I.D.: peora.2122
Posted: Tue Apr 29 09:10:20 1986
Date-Received: Fri, 2-May-86 07:26:35 EDT
References: <3080@ncsu.UUCP> <3600001@hplabsb.UUCP>
Organization: Concurrent Computer Corporation, Orlando, Fl
Lines: 25

> A slug has about 20000 nuerons and the functions of some are well known.
> The "taste" function of a slug is handled by 6 neurons . The claim is that
> "This is at clearly a capability of at least microprocessor caliber"
> so I suppose we could assume a vlsi slug would need something on the
> order of a cray to function in true slug form!

That statement in EE Times was sort of irritating.  They seemed to be
reasoning that "we need a whole microprocessor to perform this function,
so a lot of microprocessors in parallel would be required to simulate a
slug."  Of course, slugs don't have any microprocessors, so that is a sort
of absurd claim... rather, perhaps, the functional units in the slug
dedicated to "taste", etc., are very simple but specialized to "taste".
This maybe suggests that a slug-simulator should be built out of less
generalized functional units than whole microprocessors.

Related to this, I have been thinking of an interesting thing.  It would
be relatively hard, maybe, to build a dense array of neuron-like
functional units because we usually do so by depositing layers
photographically onto a silicon wafer.  But in biological systems, the
array of neurons making up the brain grows in 3-dimensions as it develops.
It would be really useful to be able to devise IC fabrication techniques
that "grew" the ICs in 3 dimensions at once. (Not that it is possible
nowadays, of course! :-))
-- 
E. Roskos