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