Genetic Algorithm–Driven IBIS-AMI Optimization for Robust 200G SerDes Design

As AI-driven computing increases bandwidth requirements, robust pre-silicon methodologies are becoming increasingly important for 200G SerDes design. This paper presents an enhanced IBIS-AMI modeling approach that improves equalizer adaptation accuracy and efficiency through the use of genetic algorithm–based optimization. Additional enhancements to adaptation cost functions further improve correlation with silicon behavior across a broad range of channels. The proposed techniques accelerate equalization optimization while narrowing the gap between simulation models and hardware performance.

This paper was presented at DesignCon 2025.