pslaswp.c File Reference

#include "common.h"

Include dependency graph for pslaswp.c:

Go to the source code of this file.

Macros
#define	B(m, n)   BLKADDR(B, float, m, n)
#define	IPIV(k)   &(IPIV[(int64_t)B.mb*(int64_t)(k)])

Functions
void	plasma_pslaswp_quark (PLASMA_desc B, int *IPIV, int inc, PLASMA_sequence *sequence, PLASMA_request *request)

---

Detailed Description

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

Version:

2.4.5

Author:

Mathieu Faverge

Date:

2010-11-15 s Tue Nov 22 14:35:43 2011

Definition in file pslaswp.c.

---

Macro Definition Documentation

#define B	(		m,
			n
	)		BLKADDR(B, float, m, n)

Definition at line 17 of file pslaswp.c.

#define IPIV

(

k

)

&(IPIV[(int64_t)B.mb*(int64_t)(k)])

Definition at line 18 of file pslaswp.c.

---

Function Documentation

void plasma_pslaswp_quark	(	PLASMA_desc	B,
		int *	IPIV,
		int	inc,
		PLASMA_sequence *	sequence,
		PLASMA_request *	request
	)

Parallel tile row interchanges - dynamic scheduling

Definition at line 23 of file pslaswp.c.

References B, IPIV, plasma_desc_t::m, plasma_desc_t::mb, plasma_desc_t::mt, plasma_desc_t::n, plasma_desc_t::nb, plasma_desc_t::nt, plasma_context_self(), plasma_desc_submatrix(), PLASMA_SUCCESS, plasma_context_struct::quark, QUARK_CORE_slaswp_ontile(), plasma_sequence_t::quark_sequence, QUARK_Task_Flag_Set(), Quark_Task_Flags_Initializer, plasma_sequence_t::status, and TASK_SEQUENCE.

{
    plasma_context_t *plasma;
    Quark_Task_Flags task_flags = Quark_Task_Flags_Initializer;

    int m, n;
    int tempi, tempm, tempmm, tempnn;

    plasma = plasma_context_self();
    if (sequence->status != PLASMA_SUCCESS)
        return;
    QUARK_Task_Flag_Set(&task_flags, TASK_SEQUENCE, (intptr_t)sequence->quark_sequence);

    if ( inc > 0 ) 
    {
        for (m = 0; m < B.mt; m++) {
            tempi = m * B.mb;
            tempm = B.m - tempi;
            tempmm = m == B.mt-1 ? tempm : B.mb;

            for (n = 0; n < B.nt; n++) {
                tempnn = n == B.nt-1 ? B.n - n * B.nb : B.nb;
                
                QUARK_CORE_slaswp_ontile(
                    plasma->quark, &task_flags,
                    plasma_desc_submatrix(B, tempi, n*B.nb, tempm, tempnn),
                    B(m, n), 1, tempmm, IPIV(m), inc, B(B.mt-1, n) );
            }
        }
    } 
    else 
    {
        for (m = B.mt-1; m > -1; m--) {
            tempi = m * B.mb;
            tempm = B.m - tempi;
            tempmm = m == B.mt-1 ? tempm : B.mb;

            for (n = 0; n < B.nt; n++) {
                tempnn = n == B.nt-1 ? B.n - n * B.nb : B.nb;
                
                QUARK_CORE_slaswp_ontile(
                    plasma->quark, &task_flags,
                    plasma_desc_submatrix(B, tempi, n*B.nb, tempm, tempnn),
                    B(m, n), 1, tempmm, IPIV(m), inc, B(0, n) );
            }
        }
    } 
}

Here is the call graph for this function: