Studying the Impact of Power Capping on MapReduce-based, Data-intensive Mini-applications on Intel KNL and KNM Architectures

TL;DR

This study quantifies how data movement affects performance and energy consumption in MapReduce applications on Intel KNL and KNM architectures, highlighting the benefits of combiner optimizations.

Contribution

It provides a detailed analysis of data movement impacts on performance and energy in MapReduce applications on HPC architectures, using power capping measurements.

Findings

Data movement significantly increases energy and runtime costs.

Combiner optimization reduces data movement costs.

Performance varies with dataset characteristics and system architecture.

Abstract

In this poster, we quantitatively measure the impacts of data movement on performance in MapReduce-based applications when executed on HPC systems. We leverage the PAPI 'powercap' component to identify ideal conditions for execution of our applications in terms of (1) dataset characteristics (i.e., unique words); (2) HPC system (i.e., KNL and KNM); and (3) implementation of the MapReduce programming model (i.e., with or without combiner optimizations). Results confirm the high energy and runtime costs of data movement, and the benefits of the combiner optimization on these costs.