pzplgsy.c File Reference

#include "common.h"

Include dependency graph for pzplgsy.c:

Go to the source code of this file.

Macros
#define	A(m, n)   BLKADDR(A, PLASMA_Complex64_t, m, n)

Functions
void	plasma_pzplgsy (plasma_context_t *plasma)
void	plasma_pzplgsy_quark (PLASMA_Complex64_t bump, PLASMA_desc A, unsigned long long int seed, 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 normal z -> c d s

Definition in file pzplgsy.c.

---

Macro Definition Documentation

#define A	(		m,
			n
	)		BLKADDR(A, PLASMA_Complex64_t, m, n)

Definition at line 17 of file pzplgsy.c.

---

Function Documentation

void plasma_pzplgsy

(

plasma_context_t *

plasma

)

Parallel tile Cholesky factorization - static scheduling

Definition at line 21 of file pzplgsy.c.

References A, BLKLDD, CORE_zplgsy(), 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_RANK, PLASMA_SIZE, PLASMA_SUCCESS, plasma_unpack_args_5, and plasma_sequence_t::status.

{
    PLASMA_Complex64_t bump;
    PLASMA_desc A;
    unsigned long long int seed;
    PLASMA_sequence *sequence;
    PLASMA_request *request;

    int m, n;
    int next_m;
    int next_n;
    int ldam;
    int tempmm, tempnn;

    plasma_unpack_args_5(bump, A, seed, sequence, request);
    if (sequence->status != PLASMA_SUCCESS)
        return;

    n = 0;
    m = PLASMA_RANK;
    while (m >= A.mt) {
        n++;
        m = m - A.mt;
    }

    while ( n < A.nt ) {
        next_n = n;
        next_m = m;

        next_m += PLASMA_SIZE;
        while ( next_m >= A.mt && next_n < A.nt ) {
            next_n++;
            next_m = next_m - A.mt;
        }

        tempmm = m == A.mt-1 ? A.m-m*A.mb : A.mb;
        tempnn = n == A.nt-1 ? A.n-n*A.nb : A.nb;
        ldam = BLKLDD(A, m);

        CORE_zplgsy( 
            bump, tempmm, tempnn, A(m, n), ldam,
            A.m, m*A.mb, n*A.nb, seed );

        m = next_m;
        n = next_n;
    }
}

Here is the call graph for this function:

Here is the caller graph for this function:

void plasma_pzplgsy_quark	(	PLASMA_Complex64_t	bump,
		PLASMA_desc	A,
		unsigned long long int	seed,
		PLASMA_sequence *	sequence,
		PLASMA_request *	request
	)

Parallel tile Cholesky factorization - dynamic scheduling

Definition at line 72 of file pzplgsy.c.

References BLKLDD, 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_SUCCESS, plasma_context_struct::quark, QUARK_CORE_zplgsy(), 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 ldam;
    int 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);

    for (m = 0; m < A.mt; m++) {
        tempmm = m == A.mt-1 ? A.m-m*A.mb : A.mb;
        ldam = BLKLDD(A, m);
        
        for (n = 0; n < A.nt; n++) {
            tempnn = n == A.nt-1 ? A.n-n*A.nb : A.nb;
            
            QUARK_CORE_zplgsy( 
                plasma->quark, &task_flags,
                bump, tempmm, tempnn, A(m, n), ldam,
                A.m, m*A.mb, n*A.nb, seed );
        }
    }
}

Here is the call graph for this function: