core_dgessm.c File Reference

#include <cblas.h>
#include <lapacke.h>
#include "common.h"

Include dependency graph for core_dgessm.c:

Go to the source code of this file.

Functions
int	CORE_dgessm (int M, int N, int K, int IB, int *IPIV, double *L, int LDL, double *A, int LDA)
void	QUARK_CORE_dgessm (Quark *quark, Quark_Task_Flags *task_flags, int m, int n, int k, int ib, int nb, int *IPIV, double *L, int ldl, double *A, int lda)
void	CORE_dgessm_quark (Quark *quark)

---

Detailed Description

PLASMA core_blas kernel PLASMA is a software package provided by Univ. of Tennessee, Univ. of California Berkeley and Univ. of Colorado Denver

Version:

2.4.5

Author:

Hatem Ltaief

Mathieu Faverge

Jakub Kurzak

Date:

2010-11-15 d Tue Nov 22 14:35:18 2011

Definition in file core_dgessm.c.

---

Function Documentation

int CORE_dgessm	(	int	M,
		int	N,
		int	K,
		int	IB,
		int *	IPIV,
		double *	L,
		int	LDL,
		double *	A,
		int	LDA
	)

CORE_dgessm applies the factor L computed by CORE_dgetrf_incpiv to a complex M-by-N tile A.

Parameters:

[in]	M	The number of rows of the tile A. M >= 0.
[in]	N	The number of columns of the tile A. N >= 0.
[in]	K
[in]	IB	The inner-blocking size. IB >= 0.
[in]	IPIV	as returned by CORE_dgetrf_incpiv.
[in]	L	The NB-by-NB lower triangular tile.
[in]	LDL	The leading dimension of the array L. LDL >= max(1,NB).
[in,out]	A	On entry, the M-by-N tile A. On exit, updated by the application of L.
[in]	LDA	The leading dimension of the array A. LDA >= max(1,M).

Returns:

Return values:

PLASMA_SUCCESS	successful exit
<0	if INFO = -k, the k-th argument had an illegal value

Definition at line 68 of file core_dgessm.c.

References cblas_dgemm(), cblas_dtrsm(), CblasColMajor, CblasLeft, CblasLower, CblasNoTrans, CblasUnit, coreblas_error, max, min, and PLASMA_SUCCESS.

{
    static double zone  =  1.0;
    static double mzone = -1.0;
    static int                ione  =  1;

    int i, sb;
    int tmp, tmp2;

    /* Check input arguments */
    if (M < 0) {
        coreblas_error(1, "Illegal value of M");
        return -1;
    }
    if (N < 0) {
        coreblas_error(2, "Illegal value of N");
        return -2;
    }
    if (K < 0) {
        coreblas_error(3, "Illegal value of K");
        return -3;
    }
    if (IB < 0) {
        coreblas_error(4, "Illegal value of IB");
        return -4;
    }
    if ((LDL < max(1,M)) && (M > 0)) {
        coreblas_error(7, "Illegal value of LDL");
        return -7;
    }
    if ((LDA < max(1,M)) && (M > 0)) {
        coreblas_error(9, "Illegal value of LDA");
        return -9;
    }

    /* Quick return */
    if ((M == 0) || (N == 0) || (K == 0) || (IB == 0))
        return PLASMA_SUCCESS;

    for(i = 0; i < K; i += IB) {
        sb = min(IB, K-i);
        /*
         * Apply interchanges to columns I*IB+1:IB*( I+1 )+1.
         */
        tmp  = i+1;
        tmp2 = i+sb;
        LAPACKE_dlaswp_work(LAPACK_COL_MAJOR, N, A, LDA, tmp, tmp2, IPIV, ione);
        /*
         * Compute block row of U.
         */
        cblas_dtrsm(
            CblasColMajor, CblasLeft, CblasLower, CblasNoTrans, CblasUnit,
            sb, N, (zone),
            &L[LDL*i+i], LDL,
            &A[i], LDA );

        if (i+sb < M) {
        /*
        * Update trailing submatrix.
        */
        cblas_dgemm(
            CblasColMajor, CblasNoTrans, CblasNoTrans,
            M-(i+sb), N, sb,
            (mzone), &L[LDL*i+(i+sb)], LDL,
            &A[i], LDA,
            (zone), &A[i+sb], LDA );
        }
    }
    return PLASMA_SUCCESS;
}

Here is the call graph for this function:

Here is the caller graph for this function:

void CORE_dgessm_quark

(

Quark *

quark

)

Definition at line 172 of file core_dgessm.c.

References A, CORE_dgessm(), IPIV, L, and quark_unpack_args_9.

{
    int m;
    int n;
    int k;
    int ib;
    int *IPIV;
    double *L;
    int ldl;
    double *A;
    int lda;

    quark_unpack_args_9(quark, m, n, k, ib, IPIV, L, ldl, A, lda);
    CORE_dgessm(m, n, k, ib, IPIV, L, ldl, A, lda);
}

Here is the call graph for this function:

Here is the caller graph for this function:

void QUARK_CORE_dgessm	(	Quark *	quark,
		Quark_Task_Flags *	task_flags,
		int	m,
		int	n,
		int	k,
		int	ib,
		int	nb,
		int *	IPIV,
		double *	L,
		int	ldl,
		double *	A,
		int	lda
	)

Definition at line 145 of file core_dgessm.c.

References CORE_dgessm_quark(), DAG_CORE_GESSM, INOUT, INPUT, QUARK_Insert_Task(), QUARK_REGION_L, and VALUE.

{
    DAG_CORE_GESSM;
    QUARK_Insert_Task(quark, CORE_dgessm_quark, task_flags,
        sizeof(int),                        &m,     VALUE,
        sizeof(int),                        &n,     VALUE,
        sizeof(int),                        &k,     VALUE,
        sizeof(int),                        &ib,    VALUE,
        sizeof(int)*nb,                      IPIV,          INPUT,
        sizeof(double)*nb*nb,    L,             INPUT | QUARK_REGION_L,
        sizeof(int),                        &ldl,   VALUE,
        sizeof(double)*nb*nb,    A,             INOUT,
        sizeof(int),                        &lda,   VALUE,
        0);
}

Here is the call graph for this function:

Here is the caller graph for this function: