Use of Carnot’s Engine and Bernoulli’s Pump to identify efficiency of information processing for computing beyond Moore's Law

In this paper, we illustrate a theretical framework that we have developed for extending Landauer-type formalism for elementary information processing to architectures and systems. We evaluate the efficiency of an ideal computing architecture at the limits of scaling. Our methodology is based on a bottom-up approach using free energy-based open system model to build an ideal computing system from a binary switch using statistical physics, thermodynamics, and quantum analysis. We devised simple computing engines such as Bernoulli’s pump and Carnot's engine to information processing operations to illustrate the premise of ideal computing subject to the physical/thermodynamic laws.

Our analysis will further demonstrate that energy can be recovered in an information processing cycle, when the computed state is brought back to the initial state. Since these gates form a complete basis for binary logic, it is possible to use arbitrarily low energy for information processing. We will also explore possibilities of reversible computing from this perspective. Similar to the design of heat engines, we hope that this framework will help in evaluating efficiences of different architectures such as analogue/network/bio-inspired systems by comparisons with a limiting ideal computing engine.