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