InKS, a Programming Model to Decouple Performance from Algorithm in HPC Codes

Abstract : Existing programming models tend to tightly interleave algorithm and optimization in HPC simulation codes. This requires scientists to become experts in both the simulated domain and the optimization process and makes the code difficult to maintain and port to new architectures. This paper proposes the InKS programming model that decouples these two concerns with distinct languages for each. The simulation algorithm is expressed in the InKS pia language with no concern for machine-specific optimizations. Optimizations are expressed using both a family of dedicated optimizations DSLs (InKS O) and plain C++. InKS O relies on the InKS pia source to assist developers with common optimizations while C++ is used for less common ones. Our evaluation demonstrates the soundness of the approach by using it on synthetic benchmarks and the Vlasov-Poisson equation. It shows that InKS offers separation of concerns at no performance cost.
Type de document :
Communication dans un congrès
Repara 2018 - 4th International Workshop on Reengineering for Parallelism in Heterogeneous Parallel Platforms, Aug 2018, Turin, Italy. pp.1-12
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01890132
Contributeur : Ksander Ejjaaouani <>
Soumis le : lundi 8 octobre 2018 - 12:56:25
Dernière modification le : mercredi 10 octobre 2018 - 01:25:47

Fichier

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

Identifiants

  • HAL Id : hal-01890132, version 1

Citation

Ksander Ejjaaouani, Olivier Aumage, Julien Bigot, Michel Mehrenberger, Hitoshi Murai, et al.. InKS, a Programming Model to Decouple Performance from Algorithm in HPC Codes. Repara 2018 - 4th International Workshop on Reengineering for Parallelism in Heterogeneous Parallel Platforms, Aug 2018, Turin, Italy. pp.1-12. 〈hal-01890132〉

Partager

Métriques

Consultations de la notice

32

Téléchargements de fichiers

16