A methodology for efficient estimation of switching activity in sequential logic circuits

Abstract

W e describe a computationally ecient s c heme to approximate average switching activity in sequential circuits which requires the solution of a non-linear system of equations of size N, where the variables correspond to state line probabilities. W e show that the approximation method is within 3% of the exact Chapman-Kolmogorov method, but is orders of magnitude faster for large circuits. Previous sequential switching activity estimation methods can have signicantly greater inaccuracies.