Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Move FEEvaluationImplHangingNodes::run_internal() into a specialized class #12997

Merged
merged 1 commit into from
Nov 27, 2021
Merged

Move FEEvaluationImplHangingNodes::run_internal() into a specialized class #12997

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
210 changes: 120 additions & 90 deletions include/deal.II/matrix_free/evaluation_kernels_hanging_nodes.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,40 +36,32 @@ class FEEvaluationBaseData;

namespace internal
{
template <int dim, typename Number, bool is_face>
struct FEEvaluationImplHangingNodes
enum class FEEvaluationImplHangingNodesRunnerTypes
{
template <int fe_degree, int n_q_points_1d>
static bool
run(const unsigned int n_desired_components,
const FEEvaluationBaseData<dim,
typename Number::value_type,
is_face,
Number> &fe_eval,
const bool transpose,
const std::array<MatrixFreeFunctions::ConstraintKinds, Number::size()>
& c_mask,
Number *values)
{
if (transpose)
run_internal<fe_degree, true>(n_desired_components,
fe_eval,
c_mask,
values);
else
run_internal<fe_degree, false>(n_desired_components,
fe_eval,
c_mask,
values);
vectorized
};

return false;
}


template <FEEvaluationImplHangingNodesRunnerTypes,
int dim,
int fe_degree,
typename Number,
bool is_face>
class FEEvaluationImplHangingNodesRunner;



template <int dim, int fe_degree, typename Number, bool is_face>
class FEEvaluationImplHangingNodesRunner<
FEEvaluationImplHangingNodesRunnerTypes::vectorized,
dim,
fe_degree,
Number,
is_face>
{
private:
template <int structdim,
int fe_degree,
unsigned int direction,
bool transpose>
template <int structdim, unsigned int direction, bool transpose>
static void
interpolate(const unsigned int offset,
const unsigned int outer_stride,
Expand Down Expand Up @@ -121,7 +113,8 @@ namespace internal
}
}

template <int fe_degree, bool transpose>
public:
template <bool transpose>
static void
run_internal(const unsigned int n_components,
const FEEvaluationBaseData<dim,
Expand Down Expand Up @@ -185,13 +178,13 @@ namespace internal
for (unsigned int c = 0; c < n_components; ++c)
for (unsigned int face = 0; face < 2; ++face)
if (do_face[0][face])
interpolate<1, fe_degree, 0, transpose>(offsets[face],
0,
given_degree,
mask_weights[0],
mask_write[0][face],
weights,
values + c * n_dofs);
interpolate<1, 0, transpose>(offsets[face],
0,
given_degree,
mask_weights[0],
mask_write[0][face],
weights,
values + c * n_dofs);
}

// y direction
Expand All @@ -200,13 +193,13 @@ namespace internal
for (unsigned int c = 0; c < n_components; ++c)
for (unsigned int face = 0; face < 2; ++face)
if (do_face[1][face])
interpolate<1, fe_degree, 1, transpose>(offsets[face],
0,
given_degree,
mask_weights[1],
mask_write[1][face],
weights,
values + c * n_dofs);
interpolate<1, 1, transpose>(offsets[face],
0,
given_degree,
mask_weights[1],
mask_write[1][face],
weights,
values + c * n_dofs);
}
}
else if (dim == 3)
Expand Down Expand Up @@ -309,27 +302,26 @@ namespace internal
for (unsigned int c = 0; c < n_components; ++c)
for (unsigned int face = 0; face < 4; ++face)
if (process_face[0][face])
interpolate<2, fe_degree, 0, transpose>(
face_offsets[face],
outer_strides[face / 2],
given_degree,
mask_weights[0],
mask_face[0][face],
weights,
values + c * n_dofs);
interpolate<2, 0, transpose>(face_offsets[face],
outer_strides[face / 2],
given_degree,
mask_weights[0],
mask_face[0][face],
weights,
values + c * n_dofs);
}
{
const std::array<unsigned int, 4> edge_offsets = {{p0, p2, p4, p6}};
for (unsigned int c = 0; c < n_components; ++c)
for (unsigned int edge = 0; edge < 4; ++edge)
if (process_edge[0][edge])
interpolate<1, fe_degree, 0, transpose>(edge_offsets[edge],
0,
given_degree,
mask_weights[0],
mask_edge[0][edge],
weights,
values + c * n_dofs);
interpolate<1, 0, transpose>(edge_offsets[edge],
0,
given_degree,
mask_weights[0],
mask_edge[0][edge],
weights,
values + c * n_dofs);
}

// direction 1:
Expand All @@ -342,28 +334,27 @@ namespace internal
for (unsigned int c = 0; c < n_components; ++c)
for (unsigned int face = 0; face < 4; ++face)
if (process_face[1][face])
interpolate<2, fe_degree, 1, transpose>(
face_offsets[face],
outer_strides[face / 2],
given_degree,
mask_weights[1],
mask_face[1][face],
weights,
values + c * n_dofs);
interpolate<2, 1, transpose>(face_offsets[face],
outer_strides[face / 2],
given_degree,
mask_weights[1],
mask_face[1][face],
weights,
values + c * n_dofs);
}

{
const std::array<unsigned int, 4> edge_offsets = {{p0, p1, p4, p5}};
for (unsigned int c = 0; c < n_components; ++c)
for (unsigned int edge = 0; edge < 4; ++edge)
if (process_edge[1][edge])
interpolate<1, fe_degree, 1, transpose>(edge_offsets[edge],
0,
given_degree,
mask_weights[1],
mask_edge[1][edge],
weights,
values + c * n_dofs);
interpolate<1, 1, transpose>(edge_offsets[edge],
0,
given_degree,
mask_weights[1],
mask_edge[1][edge],
weights,
values + c * n_dofs);
}

// direction 2:
Expand All @@ -375,28 +366,27 @@ namespace internal
for (unsigned int c = 0; c < n_components; ++c)
for (unsigned int face = 0; face < 4; ++face)
if (process_face[2][face])
interpolate<2, fe_degree, 2, transpose>(
face_offsets[face],
outer_strides[face / 2],
given_degree,
mask_weights[2],
mask_face[2][face],
weights,
values + c * n_dofs);
interpolate<2, 2, transpose>(face_offsets[face],
outer_strides[face / 2],
given_degree,
mask_weights[2],
mask_face[2][face],
weights,
values + c * n_dofs);
}

{
const std::array<unsigned int, 4> edge_offsets = {{p0, p1, p2, p3}};
for (unsigned int c = 0; c < n_components; ++c)
for (unsigned int edge = 0; edge < 4; ++edge)
if (process_edge[2][edge])
interpolate<1, fe_degree, 2, transpose>(edge_offsets[edge],
0,
given_degree,
mask_weights[2],
mask_edge[2][edge],
weights,
values + c * n_dofs);
interpolate<1, 2, transpose>(edge_offsets[edge],
0,
given_degree,
mask_weights[2],
mask_edge[2][edge],
weights,
values + c * n_dofs);
}
}
else
Expand All @@ -407,6 +397,46 @@ namespace internal
};



template <int dim, typename Number, bool is_face>
struct FEEvaluationImplHangingNodes
{
public:
template <int fe_degree, int n_q_points_1d>
static bool
run(const unsigned int n_desired_components,
const FEEvaluationBaseData<dim,
typename Number::value_type,
is_face,
Number> &fe_eval,
const bool transpose,
const std::array<MatrixFreeFunctions::ConstraintKinds, Number::size()>
& c_mask,
Number *values)
{
using RunnerType = FEEvaluationImplHangingNodesRunner<
FEEvaluationImplHangingNodesRunnerTypes::vectorized,
dim,
fe_degree,
Number,
is_face>;

if (transpose)
RunnerType::template run_internal<true>(n_desired_components,
fe_eval,
c_mask,
values);
else
RunnerType::template run_internal<false>(n_desired_components,
fe_eval,
c_mask,
values);

return false;
}
};


} // end of namespace internal


Expand Down