Author: I. Ercan

Abstract

In this talk, we present the FUELCOST methodology that allows us to obtain fundamental efficiency limits of complex computing structures. The heat dissipated as a result of logically irreversible operations in a circuit sets the limit to the fundamental energy requirement of information processing; therefore has a critical place amongst the challenges imposed on the realization of nanocomputing paradigms. The approach we developed provides a common ground for studying the energetic cost of logic- and associated physical-state transformations in electronic circuits on an equal footing. This allows us to analyze the correlation between these transformations and identify their unavoidable contribution to fundamental energy dissipation in information processing. Here, we present the essential features of our approach and illustrate its application to CMOS and post-CMOS paradigms to provide an alternative perspective on the fundamental limitations involved in changing trends in computation. The dynamically clocked and static logic circuit examples we studied emphasize the significance of circuit operation as well as its physical structure in determining the heat dissipation limits. We also discuss insights emerged from our analyses towards the potential use of our methodology as an assessment tool for emerging computing proposals.