A dynamic switching activity generation technique for power analysis of electronic circuits

This paper describes an approach to generation of switching activities in multiclock ASIC designs. The goal is to obtain realistic worst-case power consumption of a design in order to determine die size, power rail, and power supply requirements. The software program developed is named HeatGen whose proprietary algorithms internally generate a discreet collection of switching activity scenarios for the design, and represent them in a global activity file (GAF) as the output. It optionally writes a test bench which lets designers replay those internally simulated scenarios through logic simulations. This approach does not require user-supplied external stimuli, but it creates stimulus by itself based on switching and power data in library cells. Such characterization data consists of how much energy a cell releases due to its internal signal changes, and how external loads and capacitances add to its energy usage. Cell internal power is the most significant component in generating switching activities and is being used by the proposed method in this paper.