Energy constrained limits to operation and assembly of information processing systems: Lessons for directions of nanoscale systems

Amongst the most popular questions related to information systems of today are the ones related to the search of materials, devices, circuits and architectural approaches that allow us to access and utilize the nanoscale. The broad reach of these questions leads to excessively speculative and exaggerated claims because of the nature of information processing as a complex system problem. The goal of this paper is to bring out one key theme: the centrality of energy dissipation as the primary constraint independent of property of state variable utilized for digitization of the information. When using charge transport and change of electromagnetic fields in devices and systems, non-linearity, collective effects, and a hierarchy of design across length and time scales is central to efficient information processing through manipulation and transmission of bits. We show that in the limits of nanometer scale, the dominant practical constraints arise from power dissipation in ever smaller volumes and of efficient signal interconnectivity commensurate with the large density of devices. These limitations are tied to the physical basis in charge transport and changes of fields, and apply equally to all materials -hard, soft or molecular -and are encapsulated in a very simple equation: tau=QAeff/alphaU where tau is the time constant of any energy consuming operation, U the energy consumption per operation, alpha the activity factor, and Ae(f the cross-sectional area of heat removal through which heat is removed at the rate of Q . This equation points to the profound implications of size reduction. At the largest scale, the limitations arise from partitioning and hierarchical apportionment for system performance, ease of design and manufacturing. We take a hierarchical view of the underlying fundamental and practical challenges of the conventional and unconventional approaches using the analytic framework appropriate to the length scale to distinguish between fact and fantasy, a- nd to point to practical emerging directions with a system-scale perspective. I argue that Adaptive Electronics focusing on power dissipation at nanoscale is the need of the hour. Examples of possible approaches are shown.