Rapport (Rapport De Recherche) Année : 2015

Coping with silent errors in HPC applications

Résumé

This report describes a unified framework for the detection and correction of silent errors, which constitute a major threat for scientific applications at extreme-scale. We first motivate the problem and explain why checkpointing must be combined with some verification mechanism. Then we introduce a general-purpose technique based upon computational patterns that periodically repeat over time. These patterns interleave verifications and checkpoints, and we show how to determine the pattern minimizing expected execution time. Then we move to application-specific techniques and review dynamic programming algorithms for linear chains of tasks, as well as ABFT-oriented algorithms for iterative methods in sparse linear algebra.

Fichier principal
Vignette du fichier
RR-8825.pdf (912.7 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence
Loading...

Dates et versions

hal-01242369 , version 1 (11-12-2015)

Licence

Identifiants

  • HAL Id : hal-01242369 , version 1

Citer

Guillaume Aupy, Anne Benoit, Massimiliano Fasi, Yves Robert, Hongyang Sun, et al.. Coping with silent errors in HPC applications. [Research Report] RR-8825, CNRS, ENS Lyon & INRIA. 2015. ⟨hal-01242369⟩
342 Consultations
481 Téléchargements

Partager

  • More