Machine Learning for CAD/EDA: The Road Ahead

The road ahead for machine learning (ML) and CAD/EDA is built from three elements – learning, optimization, and CAD itself. Learning is the improvement of a computer agent’s perception, knowledge or actions based on experience or data [1]. Optimization is the universal quest to do better: it is a centuries-old discipline that is at the heart of leading-edge IC design. CAD is our world: a high-stakes use domain for learning and optimization that brings staggering scale and complexity along with multiple abstractions and objectives. Learning, optimization and CAD are united in service of scaling, which is the realization of more value while consuming less resources (energy, time, area, cost). Scaling makes the impossible possible: it propels IC CAD/EDA, IC design and the broader semiconductor ecosystem forward into the future.