Efficient Process Replication for MPI Applications: Sharing Work Between Replicas

Thomas Ropars 1, * Arnaud Lefray 2 Dohyun Kim 3 André Schiper 1
* Auteur correspondant
1 EPFL - Ecole Polytechnique Fédérale de Lausanne
2 LIP - Laboratoire de l'Informatique du Parallélisme
3 School of CSE - School of Computer Science and Engineering [Seoul]
Abstract : With the increased failure rate expected in future extreme scale supercomputers, process replication might become a viable alternative to checkpointing. By default, the workload efficiency of replication is limited to 50% because of the additional resources that have to be used to execute the replicas of the application's processes. In this paper, we introduce intra-parallelization, a solution that avoids replicating all computation by introducing work-sharing between replicas. We show on a representative set of benchmarks that intra-parallelization allows achieving more than 50% efficiency without compromising fault tolerance.
Keywords : High performance computing fault tolerance replication
Type de document :
Communication dans un congrès
29th IEEE International Parallel & Distributed Processing Symposium (IPDPS2015), 2015, Hyderabad, India
Liste complète des métadonnées

Littérature citée [24 références]  Voir  Masquer  Télécharger
https://hal.inria.fr/hal-01121959
Contributeur : Thomas Ropars <>
Soumis le : lundi 2 mars 2015 - 22:57:19
Dernière modification le : mardi 24 avril 2018 - 13:52:43
Document(s) archivé(s) le : mardi 2 juin 2015 - 10:00:49

Fichier

ipdps2015.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01121959, version 1

Collections

ENS-LYON | INRIA | LIP | GRID5000 | UNIV-LYON1

Citation

Thomas Ropars, Arnaud Lefray, Dohyun Kim, André Schiper. Efficient Process Replication for MPI Applications: Sharing Work Between Replicas. 29th IEEE International Parallel & Distributed Processing Symposium (IPDPS2015), 2015, Hyderabad, India. 〈hal-01121959〉

Exporter

BibTeX TEI DC DCterms EndNote

Partager

Métriques

Consultations de la notice

237

Téléchargements de fichiers

236

Contact


archive-ouverte@inria.fr