lew (10/12/82)
Jerry Leichter said, "In fact, thermodynamics gives you a lower bound on how much energy ANY chip you could build would require to store a bit." We can think of a bit as a bistable potential well with the 0 and 1 bit states separated by energy E. Then E must be > kT for the bit to be stable. But the entropy production per bit flip is E/T which must be >k. Right? The energy can be made small by cooling but it is entropy we are talking about. I don't think that it's our knowledge of the state of a memory that makes its entropy zero, but its stability. The bit transitions must have a high enough energy to be "frozen out" at the operation temperature. Lew Mammel, Jr. ihuxv!lew