On the Accuracy of Monte Carlo Yield Estimators

Abstract

Since the statistical timing analysis of VLSI circuits became a necessity, the accuracy of the proposed statistical static timing analysis (SSTA) methods is checked with the Monte Carlo (MC) estimation methods, which are called golden. However, the MC methods can have very different levels of accuracy depending on the model used beneath the method. In this paper, we build and compare three different MC yield estimators to see the effect of the models used beneath the estimator on the accuracy.