Efficiency of general Krylov methods on GPUs - An experimental study

Conference contribution
(Conference Contribution)


Publication Details

Author(s): Anzt H, Dongarra J, Kreutzer M, Wellein G, Köhler M
Publisher: IEEE Computer Society
Publication year: 2016
Pages range: 683-691
ISBN: 9781509021406


Abstract


This paper compares different Krylov methods based on short recurrences with respect to their efficiency whenimplemented on GPUs. The comparison includes BiCGSTAB, CGS, QMR, and IDR using different shadow space dimensions. These methods are known for their good convergencecharacteristics. For a large set of test matrices taken from theUniversity of Florida Matrix Collection, we evaluate the methods'performance against different target metrics: convergence, number of sparse matrix-vector multiplications, and executiontime. We also analyze whether the methods are «orthogonal»in terms of problem suitability. We propose best practicesfor choosing methods in a «black box» scenario, where noinformation about the optimal solver is available.


FAU Authors / FAU Editors

Kreutzer, Moritz
Regionales Rechenzentrum Erlangen (RRZE)
Wellein, Gerhard Prof. Dr.
Professur für Höchstleistungsrechnen


External institutions with authors

Max-Planck-Institut für Dynamik komplexer technischer Systeme / Max Planck Institute for Dynamics of Complex Technical Systems
Oak Ridge National Laboratory


How to cite

APA:
Anzt, H., Dongarra, J., Kreutzer, M., Wellein, G., & Köhler, M. (2016). Efficiency of general Krylov methods on GPUs - An experimental study. In Proceedings of the 30th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2016 (pp. 683-691). IEEE Computer Society.

MLA:
Anzt, Hartwig, et al. "Efficiency of general Krylov methods on GPUs - An experimental study." Proceedings of the 30th IEEE International Parallel and Distributed Processing Symposium Workshops, IPDPSW 2016 IEEE Computer Society, 2016. 683-691.

BibTeX: 

Last updated on 2019-23-07 at 07:24