On the Use of a Genetic Algorithm in High Performance Computer Benchmark Tuning

Abstract : The High-Performance Linpack (HPL) [14] package is a reference benchmark used worldwide to evaluate high-performance computing platforms. Adjustment of HPLpsilas seventeen tuning parameters to achieve maximum performance is a time-consuming task that must be performed by hand. In this paper, we show how a genetic algorithm may be exploited to automatically determine the best parameters possible to maximize the future results of the benchmark. Indeed we propose a GA based approach, even if we do not really specify a particular GA as our investigation relies on the Acovea framework [11], which managed repeated runs of the benchmark to explore the very large space of parameter combinations on the test-case cluster. This work opens the possibility of creating a fully-automatic benchmark tuning tool.
Type de document :
Communication dans un congrès
Proceedings of the IEEE International Symposium on Performance Evaluation of Computer and Telecommunication Systems (SPECTS'08), Jun 2008, Edinburgh, United Kingdom. 2008
Liste complète des métadonnées

https://hal.inria.fr/hal-00684874
Contributeur : Ist Rennes <>
Soumis le : mardi 3 avril 2012 - 13:44:24
Dernière modification le : lundi 20 juin 2016 - 14:10:32

Identifiants

  • HAL Id : hal-00684874, version 1

Collections

Citation

Dominic Dunlop, Sébastien Varette, Pascal Bouvry. On the Use of a Genetic Algorithm in High Performance Computer Benchmark Tuning. Proceedings of the IEEE International Symposium on Performance Evaluation of Computer and Telecommunication Systems (SPECTS'08), Jun 2008, Edinburgh, United Kingdom. 2008. 〈hal-00684874〉

Partager

Métriques

Consultations de la notice

32