Skip to content

Commit

Permalink
review updates:
Browse files Browse the repository at this point in the history 
- renaming
- documentation

Co-authored-by: Tobias Ribizel <ribizel@kit.edu>
  • Loading branch information
@MarcelKoch @upsj
MarcelKoch and upsj committed Nov 28, 2022
1 parent 6a2d1bd commit d70f984
Show file tree
Hide file tree
Showing 2 changed files with 138 additions and 67 deletions.
30 changes: 14 additions & 16 deletions benchmark/blas/blas_common.hpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -84,7 +84,7 @@ class BenchmarkOperation {
};

template <typename Generator, typename T>
auto as_vector(T&& p)
auto as_vector(T& p)
{
return gko::as<typename Generator::Vec>(gko::lend(p));
}
Expand All @@ -95,13 +95,13 @@ class CopyOperation : public BenchmarkOperation {
public:
CopyOperation(std::shared_ptr<const gko::Executor> exec,
const Generator& generator, gko::size_type rows,
gko::size_type cols, gko::size_type istride,
gko::size_type ostride)
gko::size_type cols, gko::size_type stride_in,
gko::size_type stride_out)
{
in_ = generator.create_multi_vector_strided(
exec, gko::dim<2>{rows, cols}, istride);
exec, gko::dim<2>{rows, cols}, stride_in);
out_ = generator.create_multi_vector_strided(
exec, gko::dim<2>{rows, cols}, ostride);
exec, gko::dim<2>{rows, cols}, stride_out);
as_vector<Generator>(in_)->fill(1);
}

Expand Down Expand Up @@ -379,19 +379,17 @@ struct dimensions {
};


gko::size_type get_optional(rapidjson::Value& obj, const char* name,
gko::size_type default_value)
{
if (obj.HasMember(name)) {
return obj[name].GetUint64();
} else {
return default_value;
}
}


dimensions parse_dims(rapidjson::Value& test_case)
{
auto get_optional = [](rapidjson::Value& obj, const char* name,
gko::size_type default_value) {
if (obj.HasMember(name)) {
return obj[name].GetUint64();
} else {
return default_value;
}
};

dimensions result;
result.n = test_case["n"].GetInt64();
result.k = get_optional(test_case, "k", result.n);
Expand Down
175 changes: 124 additions & 51 deletions benchmark/utils/stencil_matrix.hpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -65,45 +65,62 @@ bool is_in_box(const IndexType i, const IndexType bound)
* Generates matrix data for a 2D stencil matrix. If restricted is set to true,
* creates a 5-pt stencil, if it is false creates a 9-pt stencil.
*
* If
* strong_scaling is set to true, creates the same problem size independent of
* the number of ranks, if it false the problem size grows with the number of
* ranks.
* If `dim != [1 1]` then the matrix data is a subset of a larger matrix.
* The total matrix is a discretization of `[0, dims[0]] x [0, dims[1]]`, and
* each box is square. The position of the box defines the subset of the matrix.
* The degrees of freedom are ordered box-wise and the boxes themselves are
* ordered lexicographical. This means that the indices are with respect to the
* larger matrix, i.e. they might not start with 0.
*
* @param dims The number of boxes in each dimension.
* @param positions The position of this box with respect to each dimension.
* @param target_local_size The desired size of the boxes. The actual size can
* deviate from this to accommodate the square size of
* the boxes.
* @param restricted If true, a 5-pt stencil is used, else a 9-pt stencil.
*
* @return matrix data of a box using either 5-pt or 9-pt stencil.
*/
template <typename ValueType, typename IndexType>
gko::matrix_data<ValueType, IndexType> generate_2d_stencil_box(
std::array<int, 2> dims, std::array<int, 2> position,
std::array<int, 2> dims, std::array<int, 2> positions,
const gko::size_type target_local_size, bool restricted)
{
auto num_boxes =
std::accumulate(dims.begin(), dims.end(), 1, std::multiplies<>{});

const auto dp =
const auto discretization_points =
static_cast<IndexType>(closest_nth_root(target_local_size, 2));
const auto local_size = static_cast<gko::size_type>(dp * dp);
const auto local_size = static_cast<gko::size_type>(discretization_points *
discretization_points);
const auto global_size = local_size * num_boxes;
auto A_data = gko::matrix_data<ValueType, IndexType>(
gko::dim<2>{static_cast<gko::size_type>(global_size),
static_cast<gko::size_type>(global_size)});

auto global_offset = [&](const int px, const int py) {
return static_cast<int>(local_size) * px +
static_cast<int>(local_size) * dims[0] * py;
auto global_offset = [&](const int position_x, const int position_y) {
return static_cast<int>(local_size) * position_x +
static_cast<int>(local_size) * dims[0] * position_y;
};

auto target_coords = [&](const IndexType i, const int coord) {
return is_in_box(i, dp) ? coord : i < 0 ? coord - 1 : coord + 1;
auto target_position = [&](const IndexType i, const int position) {
return is_in_box(i, discretization_points)
? position
: (i < 0 ? position - 1 : position + 1);
};

auto target_local_idx = [&](const IndexType i) {
return is_in_box(i, dp) ? i : i < 0 ? dp + i : dp - i;
return is_in_box(i, discretization_points)
? i
: (i < 0 ? discretization_points + i
: discretization_points - i);
};

auto flat_idx = [&](const IndexType ix, IndexType iy) {
auto tcx = target_coords(ix, position[0]);
auto tcy = target_coords(iy, position[1]);
if (is_in_box(tcx, dims[0]) && is_in_box(tcy, dims[1])) {
return global_offset(tcx, tcy) + target_local_idx(ix) +
auto tpx = target_position(ix, positions[0]);
auto tpy = target_position(iy, positions[1]);
if (is_in_box(tpx, dims[0]) && is_in_box(tpy, dims[1])) {
return global_offset(tpx, tpy) + target_local_idx(ix) +
target_local_idx(iy);
} else {
return static_cast<IndexType>(-1);
Expand All @@ -127,8 +144,8 @@ gko::matrix_data<ValueType, IndexType> generate_2d_stencil_box(
};
const auto diag_value = static_cast<ValueType>(nnz_in_row() - 1);

for (IndexType i = 0; i < dp; ++i) {
for (IndexType j = 0; j < dp; ++j) {
for (IndexType i = 0; i < discretization_points; ++i) {
for (IndexType j = 0; j < discretization_points; ++j) {
auto row = flat_idx(j, i);
for (IndexType d_i : {-1, 0, 1}) {
for (IndexType d_j : {-1, 0, 1}) {
Expand Down Expand Up @@ -156,50 +173,72 @@ gko::matrix_data<ValueType, IndexType> generate_2d_stencil_box(

/**
* Generates matrix data for a 3D stencil matrix. If restricted is set to true,
* creates a 7-pt stencil, if it is false creates a 27-pt stencil. If
* strong_scaling is set to true, creates the same problem size independent of
* the number of ranks, if it false the problem size grows with the number of
* ranks.
* creates a 7-pt stencil, if it is false creates a 27-pt stencil.
*
* If `dim != [1 1 1]` then the matrix data is a subset of a larger matrix.
* The total matrix is a discretization of `[0, dims[0]] x [0, dims[1]] x [0,
* dims[2]]`, and each box is a cube. The position of the box defines the subset
* of the matrix. The degrees of freedom are ordered box-wise and the boxes
* themselves are ordered lexicographical. This means that the indices are with
* respect to the larger matrix, i.e. they might not start with 0.
*
* @param dims The number of boxes in each dimension.
* @param positions The position of this box with respect to each dimension.
* @param target_local_size The desired size of the boxes. The actual size can
* deviate from this to accommodate the uniform size
* of the boxes.
* @param restricted If true, a 7-pt stencil is used, else a 27-pt stencil.
*
* @return matrix data of a box using either 7-pt or 27-pt stencil.
*/
template <typename ValueType, typename IndexType>
gko::matrix_data<ValueType, IndexType> generate_3d_stencil_box(
std::array<int, 3> dims, std::array<int, 3> position,
std::array<int, 3> dims, std::array<int, 3> positions,
const gko::size_type target_local_size, bool restricted)
{
auto num_boxes =
std::accumulate(dims.begin(), dims.end(), 1, std::multiplies<>{});

const auto dp =
const auto discretization_points =
static_cast<IndexType>(closest_nth_root(target_local_size, 3));
const auto local_size = static_cast<gko::size_type>(dp * dp * dp);
const auto local_size = static_cast<gko::size_type>(
discretization_points * discretization_points * discretization_points);
const auto global_size = local_size * num_boxes;
auto A_data = gko::matrix_data<ValueType, IndexType>(
gko::dim<2>{static_cast<gko::size_type>(global_size),
static_cast<gko::size_type>(global_size)});

auto global_offset = [&](const int cx, const int cy, const int cz) {
return cx * static_cast<int>(local_size) +
cy * static_cast<int>(local_size) * dims[0] +
cz * static_cast<int>(local_size) * dims[0] * dims[1];
auto global_offset = [&](const int position_x, const int position_y,
const int position_z) {
return position_x * static_cast<int>(local_size) +
position_y * static_cast<int>(local_size) * dims[0] +
position_z * static_cast<int>(local_size) * dims[0] * dims[1];
};

auto target_coords = [&](const IndexType i, const int coord) {
return is_in_box(i, dp) ? coord : i < 0 ? coord - 1 : coord + 1;
auto target_position = [&](const IndexType i, const int position) {
return is_in_box(i, discretization_points)
? position
: (i < 0 ? position - 1 : position + 1);
};

auto target_local_idx = [&](const IndexType i) {
return is_in_box(i, dp) ? i : i < 0 ? dp + i : dp - i;
return is_in_box(i, discretization_points)
? i
: (i < 0 ? discretization_points + i
: discretization_points - i);
};

auto flat_idx = [&](const IndexType ix, const IndexType iy,
const IndexType iz) {
auto tcx = target_coords(ix, position[0]);
auto tcy = target_coords(iy, position[1]);
auto tcz = target_coords(iz, position[2]);
if (is_in_box(tcx, dims[0]) && is_in_box(tcy, dims[1]) &&
is_in_box(tcz, dims[2])) {
return global_offset(tcx, tcy, tcz) + target_local_idx(ix) +
target_local_idx(iy) * dp + target_local_idx(iz) * dp * dp;
auto tpx = target_position(ix, positions[0]);
auto tpy = target_position(iy, positions[1]);
auto tpz = target_position(iz, positions[2]);
if (is_in_box(tpx, dims[0]) && is_in_box(tpy, dims[1]) &&
is_in_box(tpz, dims[2])) {
return global_offset(tpx, tpy, tpz) + target_local_idx(ix) +
target_local_idx(iy) * discretization_points +
target_local_idx(iz) * discretization_points *
discretization_points;
} else {
return static_cast<IndexType>(-1);
}
Expand Down Expand Up @@ -227,9 +266,9 @@ gko::matrix_data<ValueType, IndexType> generate_3d_stencil_box(
};
const auto diag_value = static_cast<ValueType>(nnz_in_row() - 1);

for (IndexType i = 0; i < dp; ++i) {
for (IndexType j = 0; j < dp; ++j) {
for (IndexType k = 0; k < dp; ++k) {
for (IndexType i = 0; i < discretization_points; ++i) {
for (IndexType j = 0; j < discretization_points; ++j) {
for (IndexType k = 0; k < discretization_points; ++k) {
auto row = flat_idx(k, j, i);
for (IndexType d_i : {-1, 0, 1}) {
for (IndexType d_j : {-1, 0, 1}) {
Expand Down Expand Up @@ -258,6 +297,17 @@ gko::matrix_data<ValueType, IndexType> generate_3d_stencil_box(
}


/**
* Generates matrix data for the requested stencil.
*
* @see generate_2d_stencil_box, generate_3d_stencil_box
*
* @param stencil_name The name of the stencil.
* @param target_local_size The desired size of the matrix. The actual size can
* deviate from this to accommodate the uniform size
* of the discretization.
* @return matrix data using the requested stencil.
*/
template <typename ValueType, typename IndexType>
gko::matrix_data<ValueType, IndexType> generate_stencil(
std::string stencil_name, const gko::size_type target_local_size)
Expand All @@ -284,6 +334,12 @@ gko::matrix_data<ValueType, IndexType> generate_stencil(
#if GINKGO_BUILD_MPI


/**
* Generates matrix data for a given 2D stencil, where the position of this
* block is given by it's MPI rank.
*
* @see generate_2d_stencil_box
*/
template <typename ValueType, typename IndexType>
gko::matrix_data<ValueType, IndexType> generate_2d_stencil(
gko::experimental::mpi::communicator comm,
Expand All @@ -301,6 +357,12 @@ gko::matrix_data<ValueType, IndexType> generate_2d_stencil(
}


/**
* Generates matrix data for a given 23 stencil, where the position of this
* block is given by it's MPI rank.
*
* @see generate_3d_stencil_box
*/
template <typename ValueType, typename IndexType>
gko::matrix_data<ValueType, IndexType> generate_3d_stencil(
gko::experimental::mpi::communicator comm,
Expand All @@ -321,10 +383,12 @@ gko::matrix_data<ValueType, IndexType> generate_3d_stencil(

/**
* Generates matrix data for a 2D stencil matrix. If restricted is set to true,
* creates a 5-pt stencil, if it is false creates a 9-pt stencil. If
* strong_scaling is set to true, creates the same problemsize independent of
* the number of ranks, if it false the problem size grows with the number of
* ranks.
* creates a 5-pt stencil, if it is false creates a 9-pt stencil.
*
* The degrees of freedom are ordered such that each MPI rank has at most two
* neighbors.
*
* @see generate_2d_stencil_box
*/
template <typename ValueType, typename IndexType>
gko::matrix_data<ValueType, IndexType> generate_2d_stencil_with_optimal_comm(
Expand Down Expand Up @@ -385,10 +449,12 @@ gko::matrix_data<ValueType, IndexType> generate_2d_stencil_with_optimal_comm(

/**
* Generates matrix data for a 3D stencil matrix. If restricted is set to true,
* creates a 7-pt stencil, if it is false creates a 27-pt stencil. If
* strong_scaling is set to true, creates the same problemsize independent of
* the number of ranks, if it false the problem size grows with the number of
* ranks.
* creates a 7-pt stencil, if it is false creates a 27-pt stencil.
*
* The degrees of freedom are ordered such that each MPI rank has at most two
* neighbors.
*
* @see generate_3d_stencil_box
*/
template <typename ValueType, typename IndexType>
gko::matrix_data<ValueType, IndexType> generate_3d_stencil_with_optimal_comm(
Expand Down Expand Up @@ -457,6 +523,13 @@ gko::matrix_data<ValueType, IndexType> generate_3d_stencil_with_optimal_comm(
}


/**
* Generates matrix data for the requested stencil.
*
* @copydoc generate_stencil(const gko::size_type, bool)
*
* @param comm The MPI communicator to determine the rank.
*/
template <typename ValueType, typename IndexType>
gko::matrix_data<ValueType, IndexType> generate_stencil(
std::string stencil_name, gko::experimental::mpi::communicator comm,
Expand Down

0 comments on commit d70f984

Please sign in to comment.