On synthesizing workloads emulating MPI applications

Javier Navaridas, Jose Miguel-Alonso, Francisco Javier Ridruejo

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

    Abstract

    Evaluation of high performance parallel systems is a delicate issue, due to the difficulty of generating workloads that represent, with fidelity, those that will run on actual systems. In this paper we make an overview of the most usual methodologies used to generate workloads for performance evaluation purposes, focusing on the network: random traffic, patterns based on permutations, traces, execution-driven, etc. In order to fill the gap between purely synthetic and applicationdriven workloads, we present a set of pseudo-synthetic workloads that mimic applications behavior, emulating some widely-used implementations of MPI collectives and some communication patterns commonly used in scientific applications. This mimicry is done in terms of spatial distribution of messages as well as in the causal relationship among them. As an example of the proposed methodology, we use a subset of these workloads to confront tori and fat-trees. ©2008 IEEE.
    Original languageEnglish
    Title of host publicationIPDPS Miami 2008 - Proceedings of the 22nd IEEE International Parallel and Distributed Processing Symposium, Program and CD-ROM|IPDPS Miami - Proc. IEEE Int. Parallel Distrib. Process. Symp., Program CD-ROM
    DOIs
    Publication statusPublished - 2008
    EventIPDPS 2008 - 22nd IEEE International Parallel and Distributed Processing Symposium - Miami, FL
    Duration: 1 Jul 2008 → …
    http://dblp.uni-trier.de/db/conf/ipps/ipdps2008.html#KotselidisAJLKW08http://dblp.uni-trier.de/rec/bibtex/conf/ipps/KotselidisAJLKW08.xmlhttp://dblp.uni-trier.de/rec/bibtex/conf/ipps/KotselidisAJLKW08

    Conference

    ConferenceIPDPS 2008 - 22nd IEEE International Parallel and Distributed Processing Symposium
    CityMiami, FL
    Period1/07/08 → …
    Internet address

    Fingerprint

    Dive into the research topics of 'On synthesizing workloads emulating MPI applications'. Together they form a unique fingerprint.

    Cite this