Afficher la notice abrégée

Composing multiple StarPU applications over heterogeneous machines: A supervised approach

hal.structure.identifierEfficient runtime systems for parallel architectures [RUNTIME]
hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
dc.contributor.authorHUGO, Andra
hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
hal.structure.identifierHigh-End Parallel Algorithms for Challenging Numerical Simulations [HiePACS]
dc.contributor.authorGUERMOUCHE, Abdou
hal.structure.identifierEfficient runtime systems for parallel architectures [RUNTIME]
hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
dc.contributor.authorWACRENIER, Pierre-André
hal.structure.identifierEfficient runtime systems for parallel architectures [RUNTIME]
hal.structure.identifierLaboratoire Bordelais de Recherche en Informatique [LaBRI]
dc.contributor.authorNAMYST, Raymond
dc.date.accessioned2024-04-15T09:57:33Z
dc.date.available2024-04-15T09:57:33Z
dc.date.issued2014-02
dc.identifier.issn1094-3420
dc.identifier.urihttps://oskar-bordeaux.fr/handle/20.500.12278/198921
dc.description.abstractEnEnabling HPC applications to perform efficiently when invoking multiple parallel libraries simultaneously is a great chal-lenge. Even if a uniform runtime system is used underneath, scheduling tasks or threads coming from different libraries over the same set of hardware resources introduces many issues, such as resource oversubscription, undesirable cache flushes and memory bus contention. This paper presents an extension of StarPU, a runtime system specifically designed for heterogeneous architectures, that allows multiple parallel codes to run concurrently with minimal interference. Such parallel codes run within schedul-ing contexts that provide confined execution environments which can be used to partition computing resources. Scheduling contexts can be dynamically resized to optimize the allocation of computing resources among concurrently running libraries. We introduce a hypervisor that automatically expands or shrinks contexts using feedback from the run-time system (e.g. resource utilization). We demonstrate the relevance of our approach using benchmarks invoking multi-ple high performance linear algebra kernels simultaneously on top of heterogeneous multicore machines. We show that our mechanism can dramatically improve the overall application run time (-34%), most notably by reducing the average cache miss ratio (-50%).
dc.language.isoen
dc.publisherSAGE Publications
dc.subject.enresource allocation
dc.subject.enruntime optimisation
dc.subject.enParallel composition
dc.subject.enheterogeneous architectures
dc.subject.enscheduling
dc.title.enComposing multiple StarPU applications over heterogeneous machines: A supervised approach
dc.typeArticle de revue
dc.identifier.doi10.1177/1094342014527575
dc.subject.halInformatique [cs]/Calcul parallèle, distribué et partagé [cs.DC]
bordeaux.journalInternational Journal of High Performance Computing Applications
bordeaux.page285 - 300
bordeaux.volume28
bordeaux.hal.laboratoriesLaboratoire Bordelais de Recherche en Informatique (LaBRI) - UMR 5800*
bordeaux.institutionUniversité de Bordeaux
bordeaux.institutionBordeaux INP
bordeaux.institutionCNRS
bordeaux.peerReviewedoui
hal.identifierhal-01101045
hal.version1
hal.popularnon
hal.audienceInternationale
hal.origin.linkhttps://hal.archives-ouvertes.fr//hal-01101045v1
bordeaux.COinSctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.jtitle=International%20Journal%20of%20High%20Performance%20Computing%20Applications&rft.date=2014-02&rft.volume=28&rft.spage=285%20-%20300&rft.epage=285%20-%20300&rft.eissn=1094-3420&rft.issn=1094-3420&rft.au=HUGO,%20Andra&GUERMOUCHE,%20Abdou&WACRENIER,%20Pierre-Andr%C3%A9&NAMYST,%20Raymond&rft.genre=article


Fichier(s) constituant ce document

FichiersTailleFormatVue

Il n'y a pas de fichiers associés à ce document.

Ce document figure dans la(les) collection(s) suivante(s)

Afficher la notice abrégée