Introduction to machine leaning in electronic design automation (EDA)

Abstract

The rapid growth of the intergraded circuit industry as predicted by Moores law has significantly increased the importance of efficient design processes. In addition, due to physical constraint, we will soon reach the limit of how small the size of the transistor can become, and the design processes will become more complex than ever. In order to cope with those challenges and introduce the product to the market within the time to market, the industry has been developing ways to apply machine learning to concurrent EDA tools. This paper will aim to introduce how machine learning is applied to varies processes of electronic design and how they improve the current EDA tools. We will also show the limitations and opportunities of ML based EDA tools and provide a rough idea of the potential future to those who are looking into this area.