Cluster-based Localization of IR-drop in Test Application considering Parasitic Elements

Highly compact test patterns are vulnerable to IR-drop during testing which might lead to failures or breakdowns. An accurate analysis of all test patterns is infeasible due to the excessive analysis run time. Previous switching activity based IR-drop prediction methods are highly approximate since less data is used to analyze the test set. In this paper, we propose a dynamic IR-drop prediction methodology, which considers resistive and capacitive parasitic elements of the circuit together with the switching activity. The proposed method uses machine-learning based clustering and is more accurate than the general switching based method. More importantly, the methodology is fast enough that the complete test set can be processed to identify vulnerable patterns prone to IR-drop failure. The experiments show the effectiveness of the proposed approach for the approximate analysis of the complete test set.