New Row-grouped CSR format for storing the sparse matrices on GPU with implementation in CUDA
Department of mathematics, Faculty of Nuclear Sciences and Physical Engineering, Czech
arXiv:1012.2270 [cs.DC] (10 Dec 2010)
@article{2010arXiv1012.2270O,
author={Oberhuber}, T. and {Suzuki}, A. and {Vacata}, J.},
title={“{New Row-grouped CSR format for storing the sparse matrices on GPU with implementation in CUDA}”},
journal={ArXiv e-prints},
archivePrefix={“arXiv”},
eprint={1012.2270},
primaryClass={“cs.DC”},
keywords={Computer Science – Distributed, Parallel, and Cluster Computing, D.1.3},
year={2010},
month={dec},
adsurl={http://adsabs.harvard.edu/abs/2010arXiv1012.2270O},
adsnote={Provided by the SAO/NASA Astrophysics Data System}
}
In this article we present a new format for storing sparse matrices. The format is designed to perform well mainly on the GPU devices. We present its implementation in CUDA. The performance has been tested on 1,600 different types of matrices and we compare our format with the Hybrid format. We give detailed comparison of both formats and show their strong and weak parts.
December 13, 2010 by hgpu
