Performance and energy analysis of scientific workloads executing on LPSoCs

Anish Varghese*, Joshua Milthorpe, Alistair P. Rendell

*Corresponding author for this work

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

    1 Citation (Scopus)

    Abstract

    Low-power system-on-chip (LPSoC) processors provide an interesting alternative as building blocks for future HPC systems due to their high energy efficiency. However, understanding their performance-energy trade-offs and minimizing the energy-to-solution for an application running across the heterogeneous devices of an LPSoC remains a challenge. In this paper, we describe our methodology for developing an energy model which may be used to predict the energy usage of application code executing on an LPSoC system under different frequency settings. For this paper, we focus only on the CPU. Performance and energy measurements are presented for different types of workloads on the NVIDIA Tegra TK1 and Tegra TX1 systems at varying frequencies. From these results, we provide insights on how to develop a model to predict energy usage at different frequencies for general workloads.

    Original languageEnglish
    Title of host publicationParallel Processing and Applied Mathematics - 12th International Conference, PPAM 2017, Revised Selected Papers
    EditorsEwa Deelman, Roman Wyrzykowski, Konrad Karczewski, Jack Dongarra
    PublisherSpringer Verlag
    Pages113-122
    Number of pages10
    ISBN (Print)9783319780535
    DOIs
    Publication statusPublished - 2018
    Event12th International Conference on Parallel Processing and Applied Mathematics, PPAM 2017 - Czestochowa, Poland
    Duration: 10 Sept 201713 Sept 2017

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume10778 LNCS
    ISSN (Print)0302-9743
    ISSN (Electronic)1611-3349

    Conference

    Conference12th International Conference on Parallel Processing and Applied Mathematics, PPAM 2017
    Country/TerritoryPoland
    CityCzestochowa
    Period10/09/1713/09/17

    Fingerprint

    Dive into the research topics of 'Performance and energy analysis of scientific workloads executing on LPSoCs'. Together they form a unique fingerprint.

    Cite this