A Comparative Study on Optimization Algorithms in PowerSynth 2

Multi-Chip Power Modules (MCPM) are a critical component in power conversion applications. Power modules and their layout optimization have been considered a crucial step to achieving maximum performance. PowerSynth 2 (PS2) is an electronic design automation (EDA) tool for the generation and optimization of power module layouts. Currently, v2.0 uses NSGA-II and Randomization for layout optimization. However, existing NSGA-II implementation is not aware of the layout hierarchy, resulting in a less desirable solution space than Randomization. To address this limitation, this research presents a hierarchical optimization framework for the layout synthesis process in PowerSynth 2. Experimental results show that proposed hierarchical algorithms improve over the existing optimization algorithms. Moreover, MOPSO is faster than NSGA-II to converge to Pareto Front with similar solution space coverage.