Review: A Speculative Divide-and-Conquer Optimization Method for Large Analog/Mixed-Signal Circuits

Abstract

We review the speculative divide-and-conquer (SDnC) optimization method. SDnC optimization method is a highly effective for large analog/mixed-signal (AMS) circuits. As the size of circuits grows, previous optimization methods face difficulties due to an exponential increase in design complexity. To reduce exponential design complexity, SDnC decreases the evaluation and optimization levels from the large AMS circuit to smaller module units. This significant reduction in design and optimization complexity could be validated through comparisons with other methods, such as parameter sweep and PSO. Furthermore, a key aspect of SDnC is modeling intricate interactions between modules. This results in power evaluations that differ from actual simulation results by only an average of 3.4%, demonstrating the accuracy of the SDnC approach.