ErikFabrizzi/NuSiF-Solver
1
0
Fork 0
forked from moebiusband/NuSiF-Solver
Code Pull Requests Activity

checkpoint: bad overhead

Browse Source
This commit is contained in:
indiano 2025-05-05 01:21:53 +02:00
parent 946fefae5f
commit 5dd7f83dc5
1 changed files with 44 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

65
PoissonSolver/2D-omp/src/solver.c
View File

@ -62,30 +62,53 @@ void initSolver(Solver* solver, Parameter* params, int problem)
solver->p = allocate(64, bytesize);
solver->rhs = allocate(64, bytesize);
double dx = solver->dx;
double dy = solver->dy;
double* p = solver->p;
double* rhs = solver->rhs;
#pragma omp parallel for collapse(2)
for (int i = 0; i < imax + 2; i++) {
for (int j = 0; j < jmax + 2; j++) {
// P(i, j) = sin(2.0 * PI * i * dx * 2.0) + sin(2.0 * PI * j * dy * 2.0);
P(i, j) = 0;
}
double dx = solver->dx;
double dy = solver->dy;
double* p = solver->p;
double* rhs = solver->rhs;
int dim[2] = { 0 };
int num_threads = 1;
#pragma omp parallel
{
#pragma omp critical
num_threads = omp_get_num_threads();
}
omp_create_dim(num_threads, dim);
printf("%d: { %d, %d}\n", num_threads, dim[0], dim[1]);
#pragma omp parallel
{
int jsw, isw;
double local_res = 0.0;
int li_start = get_dim_start(get_x_choord(omp_get_thread_num(), dim),
dim[0],
solver->imax);
int lj_start = get_dim_start(get_y_choord(omp_get_thread_num(), dim),
dim[1],
solver->jmax);
int limax = li_start + distribute_dim(get_x_choord(omp_get_thread_num(), dim),
dim[0],
solver->imax);
int ljmax = lj_start + distribute_dim(get_y_choord(omp_get_thread_num(), dim),
dim[1],
solver->jmax);
if (problem == 2) {
#pragma omp parallel for collapse(2)
for (int i = 0; i < imax + 2; i++) {
for (int j = 0; j < jmax + 2; j++) {
RHS(i, j) = sin(2.0 * PI * i * dx);
for (int i = li_start; i < limax + 2; i++) {
for (int j = lj_start; j < ljmax + 2; j++) {
P(i, j) = sin(2.0 * PI * i * dx * 2.0) + sin(2.0 * PI * j * dy * 2.0);
}
}
} else {
#pragma omp parallel for collapse(2)
for (int i = 0; i < imax + 2; i++) {
for (int j = 0; j < jmax + 2; j++) {
RHS(i, j) = 0.0;
if (problem == 2) {
for (int i = li_start; i < limax + 2; i++) {
for (int j = lj_start; j < ljmax + 2; j++) {
RHS(i, j) = sin(2.0 * PI * i * dx);
}
}
} else {
for (int i = li_start; i < limax + 2; i++) {
for (int j = lj_start; j < ljmax + 2; j++) {
RHS(i, j) = 0.0;
}
}
}
}
@ -117,9 +140,9 @@ void solveRB(Solver* solver)
omp_create_dim(num_threads, dim);
printf("%d: { %d, %d}\n", num_threads, dim[0], dim[1]);
while ((res >= epssq) && (it < itermax)) {
res = 0.0;
#pragma omp parallel
{
res = 0.0;
int jsw, isw;
double local_res = 0.0;
int li_start = get_dim_start(get_x_choord(omp_get_thread_num(), dim),