Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site rti-sel.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxt!houxm!mtuxo!mtunh!mtung!mtunf!ariel!vax135!timeinc!phri!pesnta!amd!amdcad!decwrl!decvax!mcnc!rti-sel!trt
From: trt@rti-sel.UUCP (Tom Truscott)
Newsgroups: net.physics
Subject: Re: reversible computing
Message-ID: <274@rti-sel.UUCP>
Date: Sun, 30-Jun-85 15:49:37 EDT
Article-I.D.: rti-sel.274
Posted: Sun Jun 30 15:49:37 1985
Date-Received: Wed, 3-Jul-85 08:30:41 EDT
References: <329@sri-arpa.ARPA>
Organization: Research Triangle Institute, NC
Lines: 22

I was surprised that that Sci Amer article "The Fundamental Physical
Limits of Computation" spent so many pages on "is a minimum amount
of energy required" and almost none on "some other questions."

The question that interests me most is: how fast can computers go?
The article only suggests that "extremely fast events can take place
without any loss of energy" but hints that there might be problems
in determining the exact time at which the event took place.
But in a sequential machine the transition times are crucial!!
Years ago a letter to the CACM (I think) suggested an upper bound
on sequential transitions/second.  It used the uncertainty
principle to determine ergs/transition/second, and then E=mc^2
to determine the point at which the computer became a black hole (!).
Now, this particular line of argument might be flawed,
but surely 'MIPS' is as interesting as kilowatts/operation.

Another interesting question is: how small can a computer memory be made?
If there is some minimum volume per bit then no memory is O(1)
and binary search is not O(log n).
I wish the authors had spent more time on such things.
Posing interesting questions can be just as illuminating as answering them.
	Tom Truscott