Overview

4 Sparse BLAS Operations

Overview

This chapter explains how to use the HP MLIB Sparse BLAS implementation that supports matrix multiply and triangular solver.

This implementation provides the NIST Fortran Sparse BLAS V.0.5 standard functionality, but with three additional data types (“Sparse BLAS Naming Convention—Data Type” on page 423) and four additional matrix forms (See MSC, MSR, BMC, and BMR in Table 4-2). The Sparse BLAS library routines are written in Fortran 77 and are callable from Fortran 90 and C routines.

Chapter objectives

After reading this chapter you will:

•Be familiar with the Sparse BLAS subroutine naming convention

•Understand argument conventions

•Know what operations the Sparse BLAS performs

•Know how to use the described subprograms

•Be familiar with the Sparse BLAS subroutines

Chapter 4 Sparse BLAS Operations 421