Iterative Krylov Methods for Gravity Problems on Graphics Processing Unit

Author

Chiek Ahamed, Abal-Kassim ; Magoules, Frederic

Author_Institution

Ecole Centrale Paris, Paris, France

fYear

2013

Firstpage

16

Lastpage

20

Abstract

This paper presents the performance of linear algebra operations together with their uses within iterative Krylov methods for solving the gravity equations on Graphics Processing Unit (GPU). Numerical experiments performed on a set of real gravity matrices arising from the Chicxulub crater are exposed, showing the performance, robustness andefficiency of our algorithms, with a speed-up of up to thirty in double precision arithmetics.

Keywords

floating point arithmetic; graphics processing units; iterative methods; mathematics computing; matrix algebra; Chicxulub crater; FLOPS; GPU; double precision arithmetics; floating point operations-per-second; graphics processing unit; gravity equations; gravity problems; iterative Krylov methods; linear algebra operations; real gravity matrices; Graphics processing units; Gravity; Iterative methods; Mathematical model; Sparse matrices; Vectors; CUDA; GPU; Gravity equations; Iterative Krylov methods; Parallel computing; SpMV;

fLanguage

English

Publisher

ieee

Conference_Titel

Distributed Computing and Applications to Business, Engineering & Science (DCABES), 2013 12th International Symposium on

Conference_Location

Kingston upon Thames, Surrey, UK

Type

conf

DOI

10.1109/DCABES.2013.10

Filename

6636411