A GPU-based hyperbolic SVD algorithm
Faculty of Mechanical Engineering and Naval Architecture, University of Zagreb, Ivana Lucica 5, 10000 Zagreb, Croatia
arXiv:1008.1371v1 [cs.NA] (7 Aug 2010)
@article{novakovic2010gpu,
title={A GPU-based hyperbolic SVD algorithm},
author={Novakovic, V. and Singer, S.},
journal={Arxiv preprint arXiv:1008.1371},
year={2010}
}
The one-sided Jacobi hyperbolic singular value decomposition (HSVD) algorithm, targeting the massively parallel graphics processing units (GPUs), is developed. The algorithm also serves as the final stage of solving the symmetric indefinite eigenvalue problem. Numerical testing demonstrates the gains in speed and accuracy over the sequential and MPI-parallelized variants of the same Jacobi-type HSVD algorithms. Finally, the possiblilites of hybrid CPU–GPU parallelism are discussed.
November 10, 2010 by hgpu
