Pré-Publication, Document De Travail Année : 2015

Reproducible Triangular Solvers for High-Performance Computing

Résumé

On modern parallel architectures, floating-point computations may become non-deterministic and, therefore, non-reproducible mainly due to non-associativity of floating-point operations. We propose an algorithm to solve dense triangular systems by leveraging the standard parallel triangular solver and our, recently introduced, multi-level exact summation approach. Finally, we present implementations of the proposed fast repro-ducible triangular solver and results on recent NVIDIA GPUs.

Fichier principal
Vignette du fichier
trsv.pdf (216.33 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-01116588 , version 1 (13-02-2015)
hal-01116588 , version 2 (14-02-2015)

Licence

Identifiants

  • HAL Id : hal-01116588 , version 1

Citer

Roman Iakymchuk, David Defour, Caroline Collange, Stef Graillat. Reproducible Triangular Solvers for High-Performance Computing. 2015. ⟨hal-01116588v1⟩
1448 Consultations
1223 Téléchargements

Partager

  • More