Statistical Modeling of SRAMs

Characterizing static random access memories (SRAMs) is difficult but necessary to understandits properties. Choosing an optimal memory requires critical characteristics such as delay, power, and area. Characterization can be done accurately using SPICE but is slow. Analytical models aim to provide quick results to allow for rapid design iterations with the memory. These models do not require perfect accuracy but must maintain fidelity. This thesis presents the implementation of two Elmore-based models and statistical models for SRAMs. In addition, this thesis assesses the models' accuracy, fidelity, speed, and additional costs