On Improving Hotspot Detection Through Synthetic Pattern-Based Database Enhancement

TL;DR

This paper introduces a novel database enhancement method using synthetic pattern generation to improve hotspot prediction accuracy in IC design, addressing high false alarms of existing ML and pattern matching techniques.

Contribution

It proposes a synthetic pattern-based database enhancement approach using Design of Experiments to reduce false alarms in hotspot detection.

Findings

Effective reduction in false alarm rates compared to existing methods

Validated on 45nm process with industry-standard tools and designs

Improved hotspot prediction accuracy through synthetic pattern augmentation

Abstract

Continuous technology scaling and the introduction of advanced technology nodes in Integrated Circuit (IC) fabrication is constantly exposing new manufacturability issues. One such issue, stemming from complex interaction between design and process, is the problem of design hotspots. Such hotspots are known to vary from design to design and, ideally, should be predicted early and corrected in the design stage itself, as opposed to relying on the foundry to develop process fixes for every hotspot, which would be intractable. In the past, various efforts have been made to address this issue by using a known database of hotspots as the source of information. The majority of these efforts use either Machine Learning (ML) or Pattern Matching (PM) techniques to identify and predict hotspots in new incoming designs. However, almost all of them suffer from high false-alarm rates, mainly because they are oblivious to the root causes of hotspots. In this work, we seek to address this limitation by using a novel database enhancement approach through synthetic pattern generation based on carefully crafted Design of Experiments (DOEs). Effectiveness of the proposed method against the state-of-the-art is evaluated on a 45nm process using industry-standard tools and designs.