Reproducible Triangular Solvers for High-Performance Computing

Author

Iakymchuk, Roman ; Defour, David ; Collange, Sylvain ; Graillat, Stef

Author_Institution

LIP6, UPMC Univ. Paris 06, Paris, France

fYear

2015

fDate

13-15 April 2015

Firstpage

353

Lastpage

358

Abstract

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 reproducible triangular solver and results on recent NVIDIA GPUs.

Keywords

floating point arithmetic; parallel architectures; NVIDIA GPU; floating-point computations; floating-point operations; high-performance computing; parallel architectures; reproducible triangular solvers; standard parallel triangular solver; triangular systems; Accuracy; Computer architecture; Graphics processing units; Instruction sets; Linear systems; Matrices; Standards; GPU accelerators; Triangular linear system; accuracy; error-free transformations; reproducibility; substitution algorithm; superaccumulator;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Technology - New Generations (ITNG), 2015 12th International Conference on

Conference_Location

Las Vegas, NV

Print_ISBN

978-1-4799-8827-3

Type

conf

DOI

10.1109/ITNG.2015.63

Filename

7113498