Alt-BEAM Archive
Message #00555
To: beam@corp.sgi.com
From: Jean auBois aubois@trail.com
Date: Wed, 17 Feb 1999 17:26:35 -0700
Subject: [alt-beam] Bicore reverse photovores, chips extra
>And I think it's possible for the Nu's to directly control the reversing
>process. So It's just a matter of preference.
Sounds fairly close to Tilden's BEAMAnts from a year or so ago. Uses a
'240: four gates for two bicores, 1 gate each to make 2 Nv drivers, 1 gate
each to make 2 Nu drivers. He uses, as you write, the Nu's to perform a
reversing function (you have to understand his notion of an "XOR motor
driver" to get this part.) I'll bet he was doing something clever with the
enable lines as well, but I honestly don't remember.
Now, how he puts all of that stuff together is his secret -- and that
secret really IS the cleverest part, particularly given the large parameter
space involved. Please don't ask me 'cuz I only remember some vague
descriptions and I don't have a clue how they were individually implemented.
However, -anyone- could look at the combinatorial arrangements of Nv's and
Nu's and bicores and stuff given the number of gates that a '240 (or some
other useful chip) has and come up with all sorts of interesting building
block sets -- LEGO, anyone? As I've noted, the hard part would be figuring
out what you could do with the combinations (and one of Still/Tilden's
papers/abstracts helps here, btw.) It makes me shudder (pleasantly) to
think of what Rigter's N/2 neurons (well, I guess you'd have to have two
chips) might add to the stew.
Zoz
------------------------------------------------------------------------
eGroup home: http://www.eGroups.com/list/alt-beam
Free Web-based e-mail groups by eGroups.com
Home