What you need to know to use these subprograms

What you need to know to use these subprograms

The following sections describe overall considerations for using matrix subprograms:

•Subroutine naming convention

•Operator arguments in the BLAS Standard

Subroutine naming convention

The Extended BLAS uses a subroutine naming convention that encodes the function of each subroutine into its name. Extended BLAS subprogram names consist of four, five, or six characters in the form TXXY, TXXYY, or TXXYYY.

The BLAS Standard uses the same naming convention as the Extended BLAS, with the addition of F_ at the beginning of each routine name. That is, BLAS Standard subprogram names take the form F_TXXY, F_TXXYY, or F_TXXYYY.

For example, the legacy BLAS single-precision, triangular-solve routine is named STRSM and its BLAS Standard counterpart is named F_STRSM. Refer to “Legacy BLAS routines” on page 211 and “BLAS Standard routines” on page 339.

The first letter, denoted by T, in the naming convention indicates one of the four Fortran data types, as shown in Table 3-1.

Table 3-1 Extended BLAS Naming Convention—Data Type

TData Type S Single Precision REAL D Double Precision REAL

C Single Precision COMPLEX Z Double Precision COMPLEX

Chapter 3 Basic Matrix Operations 207