Newsgroups: comp.lang.lisp,comp.lang.scheme,comp.arch
Path: cantaloupe.srv.cs.cmu.edu!nntp.club.cc.cmu.edu!goldenapple.srv.cs.cmu.edu!das-news2.harvard.edu!news.dfci.harvard.edu!camelot.ccs.neu.edu!news.mathworks.com!howland.erols.net!ix.netcom.com!hbaker
From: hbaker@netcom.com (Henry Baker)
Subject: Re: Switching (was: Will Java VM kill Lisp?  How to fight it.)
Content-Type: text/plain; charset=ISO-8859-1
Message-ID: <hbaker-1004970823270001@10.0.2.1>
Sender: hbaker@netcom7.netcom.com
Content-Transfer-Encoding: 8bit
Organization: nil
X-Newsreader: Yet Another NewsWatcher 2.2.0
References: <5ii9i6$eaj@news-rocq.inria.fr>
Mime-Version: 1.0
Date: Thu, 10 Apr 1997 16:23:27 GMT
Lines: 18
Xref: glinda.oz.cs.cmu.edu comp.lang.lisp:26606 comp.lang.scheme:19478 comp.arch:76580

In article <5ii9i6$eaj@news-rocq.inria.fr>, harley@pauillac.inria.fr
(Robert Harley) wrote:

> Well the physics books on my planet say that decreasing the entropy of
> a system by one bit dissipates k.T.ln 2 joules of energy as heat.
> 
> So if you want to have some hope of avoiding heat, you would need
> perfectly reversible computations which, as I mentioned before, is not
> a realistic prospect for the near future.

As readers of PhysComp Proceedings, etc., already know, even close
approximations to reversible computations help _a lot_.  Suppose that a
'reversible' computation manages to dissipate 10% of its energy per bit
switched.  This is still _ten times_ better than the 'non-reversible' gate,
which manages to dissipate 100% of its energy per bit switched.

While this may not be good enough for quantum computation, it sure beats the
heck out of boolean and/or logic.
