Characterization of power usage and performance in data-intensive applications sing MapReduce over MPI

Joshua Davis, Tao Gao, Sunita Chandrasekaran, Heike Jagode, Anthony Danalis, Jack Dongarra, Pavan Balaji, Michela Taufer

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

This paper presents a quantitative evaluation of the power usage over time in data-intensive applications that use MapReduce over MPI. We leverage the PAPI powercap tool to identify ideal conditions for execution of our mini-applications in terms of (1) dataset characteristics (e.g., unique words in datasets); (2) system characteristics (e.g., KNL and KNM); and (3) implementation of the MapReduce programming model (e.g., impact of various optimizations). Results illustrate the high power utilization and runtime costs of data management on HPC architectures.

Original languageEnglish
Title of host publicationParallel Computing
Subtitle of host publicationTechnology Trends
EditorsIan Foster, Gerhard R. Joubert, Ludek Kucera, Wolfgang E. Nagel, Frans Peters
PublisherIOS Press BV
Pages287-298
Number of pages12
ISBN (Electronic)9781643680705
DOIs
StatePublished - 2020
Externally publishedYes

Publication series

NameAdvances in Parallel Computing
Volume36
ISSN (Print)0927-5452
ISSN (Electronic)1879-808X

Funding

This work was supported by NSF CCF 1841758. This work was supported by NSF CCF 1841758

FundersFunder number
NSF CCF1841758
National Science FoundationCCF 1841758

    Keywords

    • Combiner optimizations
    • Data management
    • KNL
    • KNM
    • PAPI

    Fingerprint

    Dive into the research topics of 'Characterization of power usage and performance in data-intensive applications sing MapReduce over MPI'. Together they form a unique fingerprint.

    Cite this