|
kusano |
7d535a |
|
|
kusano |
7d535a |
/*! @file csp_blas3.c
|
|
kusano |
7d535a |
* \brief Sparse BLAS3, using some dense BLAS3 operations
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* -- SuperLU routine (version 2.0) --
|
|
kusano |
7d535a |
* Univ. of California Berkeley, Xerox Palo Alto Research Center,
|
|
kusano |
7d535a |
* and Lawrence Berkeley National Lab.
|
|
kusano |
7d535a |
* November 15, 1997
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
*/
|
|
kusano |
7d535a |
/*
|
|
kusano |
7d535a |
* File name: sp_blas3.c
|
|
kusano |
7d535a |
* Purpose: Sparse BLAS3, using some dense BLAS3 operations.
|
|
kusano |
7d535a |
*/
|
|
kusano |
7d535a |
|
|
kusano |
7d535a |
#include "slu_cdefs.h"
|
|
kusano |
7d535a |
|
|
kusano |
7d535a |
/*! \brief
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* Purpose
|
|
kusano |
7d535a |
* =======
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* sp_c performs one of the matrix-matrix operations
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* C := alpha*op( A )*op( B ) + beta*C,
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* where op( X ) is one of
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* op( X ) = X or op( X ) = X' or op( X ) = conjg( X' ),
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* alpha and beta are scalars, and A, B and C are matrices, with op( A )
|
|
kusano |
7d535a |
* an m by k matrix, op( B ) a k by n matrix and C an m by n matrix.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* Parameters
|
|
kusano |
7d535a |
* ==========
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* TRANSA - (input) char*
|
|
kusano |
7d535a |
* On entry, TRANSA specifies the form of op( A ) to be used in
|
|
kusano |
7d535a |
* the matrix multiplication as follows:
|
|
kusano |
7d535a |
* TRANSA = 'N' or 'n', op( A ) = A.
|
|
kusano |
7d535a |
* TRANSA = 'T' or 't', op( A ) = A'.
|
|
kusano |
7d535a |
* TRANSA = 'C' or 'c', op( A ) = conjg( A' ).
|
|
kusano |
7d535a |
* Unchanged on exit.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* TRANSB - (input) char*
|
|
kusano |
7d535a |
* On entry, TRANSB specifies the form of op( B ) to be used in
|
|
kusano |
7d535a |
* the matrix multiplication as follows:
|
|
kusano |
7d535a |
* TRANSB = 'N' or 'n', op( B ) = B.
|
|
kusano |
7d535a |
* TRANSB = 'T' or 't', op( B ) = B'.
|
|
kusano |
7d535a |
* TRANSB = 'C' or 'c', op( B ) = conjg( B' ).
|
|
kusano |
7d535a |
* Unchanged on exit.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* M - (input) int
|
|
kusano |
7d535a |
* On entry, M specifies the number of rows of the matrix
|
|
kusano |
7d535a |
* op( A ) and of the matrix C. M must be at least zero.
|
|
kusano |
7d535a |
* Unchanged on exit.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* N - (input) int
|
|
kusano |
7d535a |
* On entry, N specifies the number of columns of the matrix
|
|
kusano |
7d535a |
* op( B ) and the number of columns of the matrix C. N must be
|
|
kusano |
7d535a |
* at least zero.
|
|
kusano |
7d535a |
* Unchanged on exit.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* K - (input) int
|
|
kusano |
7d535a |
* On entry, K specifies the number of columns of the matrix
|
|
kusano |
7d535a |
* op( A ) and the number of rows of the matrix op( B ). K must
|
|
kusano |
7d535a |
* be at least zero.
|
|
kusano |
7d535a |
* Unchanged on exit.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* ALPHA - (input) complex
|
|
kusano |
7d535a |
* On entry, ALPHA specifies the scalar alpha.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* A - (input) SuperMatrix*
|
|
kusano |
7d535a |
* Matrix A with a sparse format, of dimension (A->nrow, A->ncol).
|
|
kusano |
7d535a |
* Currently, the type of A can be:
|
|
kusano |
7d535a |
* Stype = NC or NCP; Dtype = SLU_C; Mtype = GE.
|
|
kusano |
7d535a |
* In the future, more general A can be handled.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* B - COMPLEX PRECISION array of DIMENSION ( LDB, kb ), where kb is
|
|
kusano |
7d535a |
* n when TRANSB = 'N' or 'n', and is k otherwise.
|
|
kusano |
7d535a |
* Before entry with TRANSB = 'N' or 'n', the leading k by n
|
|
kusano |
7d535a |
* part of the array B must contain the matrix B, otherwise
|
|
kusano |
7d535a |
* the leading n by k part of the array B must contain the
|
|
kusano |
7d535a |
* matrix B.
|
|
kusano |
7d535a |
* Unchanged on exit.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* LDB - (input) int
|
|
kusano |
7d535a |
* On entry, LDB specifies the first dimension of B as declared
|
|
kusano |
7d535a |
* in the calling (sub) program. LDB must be at least max( 1, n ).
|
|
kusano |
7d535a |
* Unchanged on exit.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* BETA - (input) complex
|
|
kusano |
7d535a |
* On entry, BETA specifies the scalar beta. When BETA is
|
|
kusano |
7d535a |
* supplied as zero then C need not be set on input.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* C - COMPLEX PRECISION array of DIMENSION ( LDC, n ).
|
|
kusano |
7d535a |
* Before entry, the leading m by n part of the array C must
|
|
kusano |
7d535a |
* contain the matrix C, except when beta is zero, in which
|
|
kusano |
7d535a |
* case C need not be set on entry.
|
|
kusano |
7d535a |
* On exit, the array C is overwritten by the m by n matrix
|
|
kusano |
7d535a |
* ( alpha*op( A )*B + beta*C ).
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* LDC - (input) int
|
|
kusano |
7d535a |
* On entry, LDC specifies the first dimension of C as declared
|
|
kusano |
7d535a |
* in the calling (sub)program. LDC must be at least max(1,m).
|
|
kusano |
7d535a |
* Unchanged on exit.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
* ==== Sparse Level 3 Blas routine.
|
|
kusano |
7d535a |
*
|
|
kusano |
7d535a |
*/
|
|
kusano |
7d535a |
|
|
kusano |
7d535a |
int
|
|
kusano |
7d535a |
sp_cgemm(char *transa, char *transb, int m, int n, int k,
|
|
kusano |
7d535a |
complex alpha, SuperMatrix *A, complex *b, int ldb,
|
|
kusano |
7d535a |
complex beta, complex *c, int ldc)
|
|
kusano |
7d535a |
{
|
|
kusano |
7d535a |
int incx = 1, incy = 1;
|
|
kusano |
7d535a |
int j;
|
|
kusano |
7d535a |
|
|
kusano |
7d535a |
for (j = 0; j < n; ++j) {
|
|
kusano |
7d535a |
sp_cgemv(transa, alpha, A, &b[ldb*j], incx, beta, &c[ldc*j], incy);
|
|
kusano |
7d535a |
}
|
|
kusano |
7d535a |
return 0;
|
|
kusano |
7d535a |
}
|