void CORE_dplrnt_quark(Quark *quark)
{
int m;
int n;
double *A;
int lda;
int bigM;
int m0;
int n0;
unsigned long long int seed;
quark_unpack_args_8( quark, m, n, A, lda, bigM, m0, n0, seed );
CORE_dplrnt( m, n, A, lda, bigM, m0, n0, seed );
}
/***************************************************************************//**
* Parallel tile Cholesky factorization - static scheduling
**/
void plasma_pdplrnt(plasma_context_t *plasma)
{
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_4(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_dplrnt(
tempmm, tempnn, A(m, n), ldam,
A.m, m*A.mb, n*A.nb, seed );
m = next_m;
n = next_n;
}
}
/***************************************************************************//**
* Parallel tile matrix generation - dynamic scheduling
* Same as previous function, without OpenMP pragma. Used to check solution.
**/
void plasma_pdpltmg_seq(PLASMA_desc A, unsigned long long int seed)
{
int m, n;
int ldam;
int tempmm, tempnn;
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;
double *dA = A(m, n);
CORE_dplrnt(tempmm, tempnn, dA, ldam, A.m, m*A.mb, n*A.nb, seed);
}
}
}
/***************************************************************************//**
* Parallel tile matrix generation - dynamic scheduling
**/
void plasma_pdpltmg_quark(PLASMA_desc A, unsigned long long int seed)
{
int m, n;
int ldam;
int tempmm, tempnn;
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;
double *dA = A(m, n);
hclib_pragma_marker("omp", "task depend(out:dA[0:tempnn*ldam])", "pragma38_omp_task");
CORE_dplrnt(tempmm, tempnn, dA, ldam, A.m, m*A.mb, n*A.nb, seed);
}
}
}