Fast and accurate estimation of SRAM read and hold failure probability using critical point sampling

Abstract

Owing to increase in parametric variations with technology scaling, accurate estimation of bit-cell failure probability in nano-scale static random access memory (SRAM) has become an extremely challenging task. In this study, the authors propose a method to detect the SRAM bit-cell failure, named 'critical point sampling'. Using this technique, read and hold failure probability of an SRAM bit-cell can be efficiently estimated in a simulation-based way. Simulation results show that our estimation method provides high accuracy, while being similar to 50x faster in computational speed compared to transient Monte-Carlo simulation. The method can be applied to optimise SRAM design for better yield and contributes significantly in reducing the overall design time.