Energy-Optimal Configurations for Single-Node HPC Applications

TL;DR

This paper presents a methodology to identify energy-optimal configurations for single-node HPC applications by modeling power and performance, achieving significant energy savings over default schemes.

Contribution

It introduces an application-agnostic approach combining power and performance models to optimize energy use in HPC applications.

Findings

Achieved about 14X less energy compared to default Linux DVFS.

Observed 23% energy savings over the best DVFS case.

On average, 6% energy reduction across tested applications.

Abstract

Energy efficiency is a growing concern for modern computing, especially for HPC due to operational costs and the environmental impact. We propose a methodology to find energy-optimal frequency and number of active cores to run single-node HPC applications using an application-agnostic power model of the architecture and an architecture-aware performance model of the application. We characterize the application performance using Support Vector Regression. The power consumption is estimated by modeling CMOS dynamic and static power without knowledge of the application. The energy-optimal configuration is estimated by minimizing the product of the power model and the performance model's outcomes. Results for four PARSEC applications with five different inputs show that the proposed approach used about 14X less energy when compared to the worst case of the default Linux DVFS governor. For the best case of the DVFS scheme, 23% savings were observed, with an overall average of 6% less energy.