Block-based Static Timing Analysis with Uncertainty

Abstract

Static timing analysis is a critical step in design of any digital integrated circuit. Technology and design trends have led to significant increase in environmental and process variations which need to be incorporated in static timing analysis. This paper presents a new, efficient and accurate block-based static timing analysis technique considering uncertainty. This new method is more efficient as its models arrival times as cumulative density functions (CDFs) and delays as probability functions (PDFs). Computationally simple expression are presented for basic static timing operations. The techniques are valid for any form of the probability distribution, though the use piecewise linear modeling of CDFs is highlighted in this paper. Reconvergent fanouts are handled using a new technique that avoids path tracing. Variable accuracy timing analysis can be performed by varying the modeling accuracy of the piecewise linear model. Regular and statistical timing on different parts of the circuit can be incorporated into a single timing analysis run. Accuracy and efficiency of the proposed method is demonstrated for various ISCAS benchmark circuits.