Modelling DVFS and UFS for Region-Based Energy Aware Tuning of HPC Applications

TL;DR

This paper introduces a neural network-based autotuning plugin for HPC applications that optimizes energy efficiency by dynamically adjusting DVFS and UFS at the region level, achieving significant energy savings.

Contribution

It presents a novel region-based autotuning approach integrating neural network predictions with hardware frequency scaling for energy-aware HPC application tuning.

Findings

Average energy savings of 16.1% with the proposed method

Compared to 7.8% savings with static tuning

Effective energy reduction demonstrated on five benchmarks

Abstract

Energy efficiency and energy conservation are one of the most crucial constraints for meeting the 20MW power envelope desired for exascale systems. Towards this, most of the research in this area has been focused on the utilization of user-controllable hardware switches such as per-core dynamic voltage frequency scaling (DVFS) and software controlled clock modulation at the application level. In this paper, we present a tuning plugin for the Periscope Tuning Framework which integrates fine-grained autotuning at the region level with DVFS and uncore frequency scaling (UFS). The tuning is based on a feed-forward neural network which is formulated using Performance Monitoring Counters (PMC) supported by x86 systems and trained using standardized benchmarks. Experiments on five standardized hybrid benchmarks show an energy improvement of 16.1% on average when the applications are tuned according to our methodology as compared to 7.8% for static tuning.