Cell Library Characterization for Composite Current Source Models Based on Gaussian Process Regression and Active Learning

TL;DR

This paper introduces a Gaussian Process Regression and Active Learning framework for efficient and accurate characterization of composite current source models, significantly reducing data, time, and storage costs while outperforming existing tools.

Contribution

The paper presents a novel GPR with active learning approach for CCS characterization, improving accuracy and efficiency over traditional methods.

Findings

Achieved an average absolute error of 2.05 ps in current waveform prediction.

Reduced runtime to 27% of commercial tools.

Decreased storage requirements by up to 19.5 times.

Abstract

The composite current source (CCS) model has been adopted as an advanced timing model that represents the current behavior of cells for improved accuracy and better capability than traditional non-linear delay models (NLDM) to model complex dynamic effects and interactions under advanced process nodes. However, the high accuracy requirement, large amount of data and extensive simulation cost pose severe challenges to CCS characterization. To address these challenges, we introduce a novel Gaussian Process Regression(GPR) model with active learning(AL) to establish the characterization framework efficiently and accurately. Our approach significantly outperforms conventional commercial tools as well as learning based approaches by achieving an average absolute error of 2.05 ps and a relative error of 2.27% for current waveform of 57 cells under 9 process, voltage, temperature (PVT) corners with TSMC 22nm process. Additionally, our model drastically reduces the runtime to 27% and the storage by up to 19.5x compared with that required by commercial tools.