Skip to content

Commit

Permalink
Fuse some mfiter loops (#1100)
Browse files Browse the repository at this point in the history
  • Loading branch information
@marchdf
marchdf authored Jun 14, 2024
1 parent a037e52 commit b46eb08
Show file tree
Hide file tree
Showing 3 changed files with 38 additions and 54 deletions.
13 changes: 6 additions & 7 deletions unit_tests/wind_energy/actuator/test_act_utils.H
View file
Original file line number Diff line number Diff line change
Expand Up @@ -38,21 +38,20 @@ inline void init_field(amr_wind::Field& fld)
const auto& dx = mesh.Geom(lev).CellSizeArray();
const auto& problo = mesh.Geom(lev).ProbLoArray();

for (amrex::MFIter mfi(fld(lev)); mfi.isValid(); ++mfi) {
auto bx = mfi.growntilebox();
const auto& farr = fld(lev).array(mfi);

amrex::ParallelFor(bx, [=] AMREX_GPU_DEVICE(int i, int j, int k) {
const auto& farrs = fld(lev).arrays();
amrex::ParallelFor(
fld(lev), fld.num_grow(),
[=] AMREX_GPU_DEVICE(int nbx, int i, int j, int k) noexcept {
const amrex::Real x = problo[0] + (i + offsetx) * dx[0];
const amrex::Real y = problo[1] + (j + offsety) * dx[1];
const amrex::Real z = problo[2] + (k + offsetz) * dx[2];

for (int d = 0; d < ncomp; d++) {
farr(i, j, k, d) = x + y + z;
farrs[nbx](i, j, k, d) = x + y + z;
}
});
}
}
amrex::Gpu::synchronize();
}

} // namespace amr_wind_tests
Expand Down
10 changes: 1 addition & 9 deletions unit_tests/wind_energy/test_abl_bc.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -55,15 +55,7 @@ void init_velocity(amr_wind::Field& fld, amrex::Real vval, int dir)
fld.setVal(0.0);

for (int lev = 0; lev < nlevels; ++lev) {

for (amrex::MFIter mfi(fld(lev)); mfi.isValid(); ++mfi) {
auto bx = mfi.growntilebox();
const auto& farr = fld(lev).array(mfi);
amrex::ParallelFor(bx, [=] AMREX_GPU_DEVICE(int i, int j, int k) {
// Initialize specified dir to specified value
farr(i, j, k, dir) = vval;
});
}
fld(lev).setVal(vval, dir, 1);
}
}
} // namespace
Expand Down
69 changes: 31 additions & 38 deletions unit_tests/wind_energy/test_abl_stats.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,18 +13,15 @@ void init_field1(amr_wind::Field& fld)
const int nlevels = fld.repo().num_active_levels();

for (int lev = 0; lev < nlevels; ++lev) {

for (amrex::MFIter mfi(fld(lev)); mfi.isValid(); ++mfi) {
auto bx = mfi.growntilebox();
const auto& farr = fld(lev).array(mfi);

// Give TKE gradient for nonzero diff term
amrex::ParallelFor(bx, [=] AMREX_GPU_DEVICE(int i, int j, int k) {
farr(i, j, k, 0) =
const auto& farrs = fld(lev).arrays();
amrex::ParallelFor(
fld(lev), amrex::IntVect(0),
[=] AMREX_GPU_DEVICE(int nbx, int i, int j, int k) noexcept {
farrs[nbx](i, j, k) =
std::sin(0.01 * i) + std::pow(k, 0.2) + std::cos(0.01 * j);
});
}
}
amrex::Gpu::synchronize();
}

amrex::Real test_new_tke(
Expand All @@ -42,24 +39,23 @@ amrex::Real test_new_tke(
for (int lev = 0; lev < tke.repo().num_active_levels(); ++lev) {

// Form tke estimate by adding to the old tke field
for (amrex::MFIter mfi(tkeold(lev)); mfi.isValid(); ++mfi) {
const auto& bx = mfi.tilebox();
const auto& tke_old_arr = tkeold(lev).array(mfi);
const auto& buoy_prod_arr = buoy_prod(lev).const_array(mfi);
const auto& shear_prod_arr = shear_prod(lev).const_array(mfi);
const auto& dissipation_arr = dissipation(lev).const_array(mfi);
const auto& diffusion_arr = diffusion(lev).const_array(mfi);
const auto& conv_arr = conv_term(lev).const_array(mfi);

amrex::ParallelFor(
bx, [=] AMREX_GPU_DEVICE(int i, int j, int k) noexcept {
tke_old_arr(i, j, k) +=
dt *
(conv_arr(i, j, k) + shear_prod_arr(i, j, k) +
buoy_prod_arr(i, j, k) - dissipation_arr(i, j, k) +
diffusion_arr(i, j, k));
});
}
const auto& tke_old_arrs = tkeold(lev).arrays();
const auto& buoy_prod_arrs = buoy_prod(lev).const_arrays();
const auto& shear_prod_arrs = shear_prod(lev).const_arrays();
const auto& dissipation_arrs = dissipation(lev).const_arrays();
const auto& diffusion_arrs = diffusion(lev).const_arrays();
const auto& conv_arrs = conv_term(lev).const_arrays();
amrex::ParallelFor(
tkeold(lev), amrex::IntVect(0),
[=] AMREX_GPU_DEVICE(int nbx, int i, int j, int k) noexcept {
tke_old_arrs[nbx](i, j, k) +=
dt *
(conv_arrs[nbx](i, j, k) + shear_prod_arrs[nbx](i, j, k) +
buoy_prod_arrs[nbx](i, j, k) -
dissipation_arrs[nbx](i, j, k) +
diffusion_arrs[nbx](i, j, k));
});
amrex::Gpu::synchronize();

// Difference between tke estimate and tke calculated by the code
error_total += amrex::ReduceSum(
Expand All @@ -84,19 +80,16 @@ amrex::Real test_new_tke(
void remove_nans(amr_wind::Field& field)
{
for (int lev = 0; lev < field.repo().num_active_levels(); ++lev) {

// Form tke estimate by adding to the old tke field
for (amrex::MFIter mfi(field(lev)); mfi.isValid(); ++mfi) {
const auto& bx = mfi.tilebox();
const auto& field_arr = field(lev).array(mfi);
amrex::ParallelFor(
bx, [=] AMREX_GPU_DEVICE(int i, int j, int k) noexcept {
field_arr(i, j, k) = std::isnan(field_arr(i, j, k))
const auto& farrs = field(lev).arrays();
amrex::ParallelFor(
field(lev), amrex::IntVect(0), field(lev).nComp(),
[=] AMREX_GPU_DEVICE(int nbx, int i, int j, int k, int n) noexcept {
farrs[nbx](i, j, k, n) = std::isnan(farrs[nbx](i, j, k, n))
? 0.0
: field_arr(i, j, k);
});
}
: farrs[nbx](i, j, k, n);
});
}
amrex::Gpu::synchronize();
}
} // namespace

Expand Down

0 comments on commit b46eb08

Please sign in to comment.