Context-Aware DFM Rule Analysis and Scoring Using Machine Learning

TL;DR

This paper introduces a machine learning-based method for more accurate DFM rule scoring by analyzing layout context and predicting lithography impacts, improving manufacturability evaluation in physical design.

Contribution

It presents a novel machine learning approach that considers layout context for DFM scoring, addressing the limitations of rule-based methods.

Findings

Machine learning improves DFM score accuracy

Context-aware analysis enhances manufacturability predictions

Method outperforms traditional rule-based scoring

Abstract

To evaluate the quality of physical layout designs in terms of manufacturability, DFM rule scoring techniques have been widely used in physical design and physical verification phases. However, one major drawback of conventional DFM rule scoring methodologies is that resultant DFM rule scores may not accurate since the scores may not highly correspond to lithography simulation results. For instance, conventional DFM rule scoring methodologies usually use rule-based techniques to compute scores without considering neighboring geometric scenarios of targeted layout shapes. That can lead to inaccurate scoring results since computed DFM rule scores can be either too optimistic or too pessimistic. Therefore, in this paper, we propose a novel approach with the use of machine learning technology to analyze the context of targeted layouts and predict their lithography impacts on manufacturability.