A high performance algorithm using pre-processing for the sparse matrix-vector multiplication

In this paper we propose a feature extraction based algorithm (FEBA) for the sparse matrix-vector multiplication. The key idea of FEB A is to exploit any regular structure present in the sparse matrix by extracting it and processing it separately. The order in which these structures are extracted is determined by the relative efficiency with which they can be processed. We have tested FEB A on IBM 3090 VF for matrices from Harwell Boeing and OSL collection.