Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!aramis.rutgers.edu!athos.rutgers.edu!nanotech From: peb@tma1.sun.com (Paul Baclaski) Newsgroups: sci.nanotech Subject: Re: Reversible logic -- is it practical? Message-ID: Date: 13 Jul 89 03:18:28 GMT Sender: nanotech@athos.rutgers.edu Lines: 23 Approved: nanotech@aramis.rutgers.edu In article , merkle.pa@xerox.com writes: > This issue is discussed in some detail in "Conservative Logic" by Edward > Fredkin and Tommaso Toffoli, Internation Journal of Theoretical Physics, > Vol. 21, Nos. 3/4, 1982. > > Basically, you run the computation forwards until you get the desired > answer, then (because the logic gates are all reversible) you run the > computation backwards (eating up the "waste" outputs as you go) until you > reach the initial condition. Result? You get exactly the output that you > want, and no "waste output" is left. They also discuss a more ingenious Is there a cost for reversal (I would guess that the machine would have a register for "direction of time".. )? Obviously, this would slow down throughput by a factor of 2. There must be a cost for reprogramming the machine or if it does any i/o. Thus, the cost may be very low, but it is not zero. Paul E. Baclaski Sun Microsystems peb@sun.com