A Randomized LU-based Solver Using GPU and Intel Xeon Phi Accelerators

Conference paper

DOI: 10.1007/978-3-319-27308-2_15

Part of the Lecture Notes in Computer Science book series (LNCS, volume 9523)
Cite this paper as:
Baboulin M., Khabou A., Rémy A. (2015) A Randomized LU-based Solver Using GPU and Intel Xeon Phi Accelerators. In: Hunold S. et al. (eds) Euro-Par 2015: Parallel Processing Workshops. Euro-Par 2015. Lecture Notes in Computer Science, vol 9523. Springer, Cham

Abstract

We present a fast hybrid solver for dense linear systems based on LU factorization. To achieve good performance, we avoid pivoting by using random butterfly transformations for which we developed efficient implementations on heterogeneous architectures. We used both Graphics Processing Units and Intel Xeon Phi as accelerators. The performance results show that the pre-processing due to randomization is negligible and that the solver outperforms the corresponding routines based on partial pivoting.

Keywords

Random Butterfly Transformations (RBT) LU factorization Graphics Processing Units (GPU) Intel Xeon Phi 

Copyright information

© Springer International Publishing Switzerland 2015

Authors and Affiliations

  1. 1.Université Paris-SudOrsayFrance

Personalised recommendations