Points Device¶

template<std::size_t Ndim, std::floating_point TData = float, concepts::device TDev = clue::Device> class PointsDevice : public internal::points_interface<PointsDevice<Ndim, float, clue::Device>>¶

The PointsDevice class is a data structure that manages points on a device. It provides methods to allocate, access, and manipulate points in device memory.

Template Parameters:¶

std::size_t Ndim¶: The number of dimensions of the points to manage
std::floating_point TData = float¶: The data type for the point coordinates and weights
concepts::device TDev = clue::Device¶: The device type to use for the allocation. Defaults to clue::Device.

Public Types¶

using element_type = TData ¶

using value_type = std::remove_cv_t<TData>¶

Public Functions¶

template<concepts::queue TQueue> PointsDevice(TQueue &queue, std::int32_t n_points)¶

Construct a PointsDevice object.

Parameters:¶

TQueue &queue¶: The queue to use for the device operations
std::int32_t n_points¶: The number of points to allocate

template<concepts::queue TQueue> PointsDevice(TQueue &queue, std::int32_t n_points, std::span<std::byte> buffer)¶

Construct a PointsDevice object with a pre-allocated buffer.

Parameters:¶

TQueue &queue¶: The queue to use for the device operations
std::int32_t n_points¶: The number of points to allocate
std::span<std::byte> buffer¶: The buffer to use for the points

template<concepts::queue TQueue> PointsDevice(TQueue &queue, std::int32_t n_points, std::span<element_type> input, std::span<int> output)¶

Constructs a container for the points allocated on the device using interleaved data.

Note

The input buffer must contain the coordinates and weights in an SoA format

Parameters:¶

TQueue &queue¶: The queue to use for memory allocation
std::int32_t n_points¶: The number of points
input_buffer: The pre-allocated buffer containing interleaved coordinates and weights
output_buffer: The pre-allocated buffer to store the cluster indexes

template<concepts::queue TQueue> PointsDevice(TQueue &queue, std::int32_t n_points, std::span<element_type> coordinates, std::span<element_type> weights, std::span<int> output)¶

Constructs a container for the points allocated on the device using separate coordinate and weight buffers.

Note

The coordinates buffer must have a size of n_points * Ndim

Parameters:¶

TQueue &queue¶: The queue to use for memory allocation
std::int32_t n_points¶: The number of points
std::span<element_type> coordinates¶: The pre-allocated buffer containing the coordinates
std::span<element_type> weights¶: The pre-allocated buffer containing the weights
std::span<int> output¶: The pre-allocated buffer to store the cluster indexes

template<concepts::queue TQueue> PointsDevice(TQueue &queue, std::int32_t n_points, element_type *input, int *output)¶

Constructs a container for the points allocated on the device using interleaved data.

Note

The input buffer must contain the coordinates and weights in an SoA format

Parameters:¶

TQueue &queue¶: The queue to use for memory allocation
std::int32_t n_points¶: The number of points
input_buffer: The pre-allocated buffer containing interleaved coordinates and weights
output_buffer: The pre-allocated buffer to store the cluster indexes

template<concepts::queue TQueue> PointsDevice(TQueue &queue, std::int32_t n_points, element_type *coordinates, element_type *weights, int *output)¶

Constructs a container for the points allocated on the device using separate coordinate and weight buffers.

Note

The coordinates buffer must have a size of n_points * Ndim

Parameters:¶

TQueue &queue¶: The queue to use for memory allocation
std::int32_t n_points¶: The number of points
element_type *coordinates¶: The pre-allocated buffer containing the coordinates
element_type *weights¶: The pre-allocated buffer containing the weights
int *output¶: The pre-allocated buffer to store the cluster indexes

template<concepts::queue TQueue, concepts::pointer... TBuffers> PointsDevice(TQueue &queue, std::int32_t n_points, TBuffers... buffers)¶

Construct a PointsDevice object with a pre-allocated buffer.

Parameters:¶

TQueue &queue¶: The queue to use for the device operations
std::int32_t n_points¶: The number of points to allocate
TBuffers... buffers¶: The buffers to use for the points

PointsDevice(const PointsDevice&) = delete¶

PointsDevice &operator=(const PointsDevice&) = delete¶

PointsDevice(PointsDevice&&) = default¶

PointsDevice &operator=(PointsDevice&&) = default¶

~PointsDevice() = default¶

ALPAKA_FN_HOST auto size() const¶

Returns the number of points.

Returns:¶: The number of points

ALPAKA_FN_HOST auto coords(size_t dim) const¶

Returns the coordinates of the points for a specific dimension as a const span.

Parameters:¶

size_t dim¶: The dimension for which to get the coordinates

Returns:¶

A const span of the coordinates for the specified dimension

ALPAKA_FN_HOST auto coords(size_t dim)¶

Returns the coordinates of the points for a specific dimension as a span.

Parameters:¶

size_t dim¶: The dimension for which to get the coordinates

Returns:¶

A span of the coordinates for the specified dimension

ALPAKA_FN_HOST auto weights() const¶

Returns the weights of the points as a const span.

Returns:¶: A const span of the weights of the points

ALPAKA_FN_HOST auto weights()¶

Returns the weights of the points as a span.

Returns:¶: A span of the weights of the points

ALPAKA_FN_HOST auto clusterIndexes() const¶

Returns the cluster indexes of the points as a const span.

Returns:¶: A const span of the cluster indexes of the points

ALPAKA_FN_HOST auto clusterIndexes()¶

Returns the cluster indexes of the points as a span.

Returns:¶: A span of the cluster indexes of the points

ALPAKA_FN_HOST auto clustered() const¶

Indicates whether the points have been clustered.

Returns:¶: True if the points have been clustered, false otherwise

ALPAKA_FN_HOST const auto &view() const¶

Returns the view of the points.

Returns:¶: A const reference to the PointsView structure containing the points data

ALPAKA_FN_HOST auto &view()¶

Returns the view of the points.

Returns:¶: A reference to the PointsView structure containing the points data

ALPAKA_FN_HOST const auto &n_clusters()¶

Teturns the cluster properties of the points.

Note

This value is lazily evaluated and cached upon the first call

Returns:¶: The number of clusters reconstructed