On improving the performance of sparse matrix-vector multiplication

Author

White, James B., III ; Sadayappan, P.

Author_Institution

Ohio Supercomput. Center, Columbus, OH, USA

fYear

1997

fDate

18-21 Dec 1997

Firstpage

66

Lastpage

71

Abstract

We analyze single node performance of sparse matrix vector multiplication by investigating issues of data locality and fine grained parallelism. We examine the data locality characteristics of the compressed sparse row representation and consider improvements in locality through matrix permutation. Motivated by potential improvements in fine grained parallelism, we evaluate modified sparse matrix representations. The results lead to general conclusions about improving single node performance of sparse matrix vector multiplication in parallel libraries of sparse iterative solvers

Keywords

mathematics; mathematics computing; matrix multiplication; parallel algorithms; parallel programming; sparse matrices; compressed sparse row representation; data locality; fine grained parallelism; matrix permutation; modified sparse matrix representations; parallel libraries; single node performance; sparse iterative solvers; sparse matrix vector multiplication; Containers; Data analysis; Libraries; Load management; Operating systems; Performance analysis; Scalability; Sparse matrices; Supercomputers;

fLanguage

English

Publisher

ieee

Conference_Titel

High-Performance Computing, 1997. Proceedings. Fourth International Conference on

Conference_Location

Bangalore

Print_ISBN

0-8186-8067-9

Type

conf

DOI

10.1109/HIPC.1997.634472

Filename

634472