A New Statistical Electromigration Analysis Methodology that Incorporates Across-Chip Temperature Variation

ted sun,  ayhan mutlu,  mahmud rahman

santa clara university

Abstract

In this paper, we present a new approach to estimate a design's EM risk based on the statistical nature of EM failures as well as an additional consideration of the across-chip temperature variation.

The results of this new methodology assure a more realistic way of predicting EM violation risk by considering the different temperature zones of a block or a chip. Using our methodology on an example block has verified that the EM failure probability is reduced from 14.85% to 0.0431% when temperature variation information is incorporated.

Especially, with the increased design size, the designers need to focus on the violations that are posing a real EM risk to reduce time to market. Our proposed method can be effectively used to realize shorter design cycles.