The shift-left design paradigm of EDA: progress and challenges

Abstract

Shift-left and fusion is one of the important innovative design paradigms in the field of electronic design automation (EDA). The traditional EDA flow can be divided into multiple serial design stages, resulting in a long design cycle. Different from the traditional waterfall design paradigm used in EDA design flow, it proposes the use of fusion and parallelization in different stages. More clearly, through the integration and parallelization of adjacent design stages, the design cycle can be optimized effectively, and efficiency can be improved. Shift-left and fusion usually involve sharing data models between different design stages and adopting a new abstraction that can be applied to various stages of EDA flow, including high-level synthesis, logic synthesis, placement and routing, verification, process optimization, etc. In addition, with the help of early data analysis, possible problems in the late stage can be identified and addressed early, thereby reducing development and verification costs. In summary, left shift and fusion are of great importance for improving the efficiency and quality of the EDA flow. This paper will introduce and discuss the progress and challenges of the left-shift and fusion paradigm and look forward to the potential frontier research directions recommended to the National Natural Science Foundation of China.