MAVIREC: ML-Aided Vectored IR-DropEstimation and Classification

TL;DR

MAVIREC employs machine learning techniques to efficiently identify worst-case IR drop vectors and accurately predict IR drop, significantly reducing analysis time while improving coverage and precision in chip power integrity assessment.

Contribution

It introduces a novel ML-based approach using 3D convolutions for vectored IR drop analysis, outperforming traditional heuristics and industrial methods in speed and accuracy.

Findings

Profiles 100K-cycle vectors in under 30 minutes

Achieves better coverage than existing industrial flow

Provides IR drop predictions with 10x speedup and under 4mV RMSE

Abstract

Vectored IR drop analysis is a critical step in chip signoff that checks the power integrity of an on-chip power delivery network. Due to the prohibitive runtimes of dynamic IR drop analysis, the large number of test patterns must be whittled down to a small subset of worst-case IR vectors. Unlike the traditional slow heuristic method that select a few vectors with incomplete coverage, MAVIREC uses machine learning techniques -- 3D convolutions and regression-like layers -- for accurately recommending a larger subset of test patterns that exercise worst-case scenarios. In under 30 minutes, MAVIREC profiles 100K-cycle vectors and provides better coverage than a state-of-the-art industrial flow. Further, MAVIREC's IR drop predictor shows 10x speedup with under 4mV RMSE relative to an industrial flow.