F_SGEMM/F_DGEMM/F_CGEMM/F_ZGEMM

General matrix-matrix multiply

 

 

BLAS_TRANS k-by-mmatrix AT

 

 

BLAS_CONJ_TRANS k-by-mmatrix A*

 

LDA

Leading dimension of array A.

 

 

Error conditions for lda depend on the value of transa.

 

 

Each of the following conditions generates an error flag

 

 

that is passed to the error handler:

 

 

• lda < 1

 

 

• TRANSA = BLAS_NO_TRANS and lda < m

 

 

• TRANSA = BLAS_TRANS and lda < k

 

 

• TRANSA = BLAS_CONJ_TRANS and lda < k

 

B

Array containing the matrix B, whose size is indicated

 

 

by TRANSB:

 

 

BLAS_NO_TRANS k-by-nmatrix B

 

 

BLAS_TRANS n-by-kmatrix BT

 

 

BLAS_CONJ_TRANS n-by-kmatrix B*

 

LDB

Leading dimension of array B.

 

 

Error conditions for ldb depend on the value of transb.

 

 

Each of the following conditions generates an error flag

 

 

that is passed to the error handler:

 

 

• ldb < 1

 

 

• TRANSB = BLAS_NO_TRANS and ldb < k

 

 

• TRANSB = BLAS_TRANS and ldb < n

 

 

• TRANSB = BLAS_CONJ_TRANS and ldb < n

 

BETA

The scalar BETA.

 

C

The m-by-nmatrix operand C. The representation of

 

 

the matrix entry ci,j in C is denoted by C(i, j) for all (i, j)

 

 

in the interval [0... m - 1] x [0... n - 1].

 

LDC

Leading dimension of array C. If ldc < 1 or ldc < m, an

 

 

error flag is generated and passed to the error handler.

Output

C

The updated m-by-nmatrix C replaces the input.

C ← αop( A)op( B) + βC

364HP MLIB User’s Guide