merkle.pa@XEROX.COM (06/06/89)
Drexler's proposed rod logic dissipates about 10**-21 joules per "lock" operation (roughly equivalent to a gate operation). This is very small by present day standards. Interestingly, K. K. Likharev proposed a fully reversible device, the PQ (Parametric Quantron), based on Josephson Junction device technology, which dissipates 10**-24 joules per operation when operated at 4 degrees Kelvin at a speed of 10**-9 seconds (1 nanosecond) per operation. This compares with rod-logic dissipating about 10**-21 joules at 300 degrees Kelvin at a speed of 50 x 10**-12 seconds (50 picoseconds) per operation. This comparison is not, strictly speaking, apples and apples. By lowering the speed of Drexler's rod-logic (or raising the speed of Likharev's PQ) and by lowering the temperature around the rods, the difference in energy dissipation largely disappears. Still, I do not believe Likharev's proposal is widely known in the nanotechnology community, nor is there any particular reason to believe either proposal is "optimal," in the sense of dissipating minimal energy per logic operation at a given speed and temperature of operation. Can anyone propose (or has anyone heard of) a device that can be built from atoms which has lower energy dissipation per logic operation? Note that all such devices will have to be fully reversible (see: The Fundamental Physical Limits of Computation, by Charles H. Bennett and Rolf Landauer, Scientific American, Vol. 253, July 1985, pages 48-56). Reference: "Classical and Quantum Limitations on Energy Consumption in Computation" by K. K. Likharev, International Journal of Theoretical Physics, Vol. 21, Nos. 3/4, 1982, pages 311-326.