mdspan: Multi-dimensional Non-owning View#

#include <raft/core/mdspan.hpp>

template<typename ElementType, typename Extents, typename LayoutPolicy = layout_c_contiguous, typename AccessorPolicy = std::experimental::default_accessor<ElementType>> using raft::mdspan = std::experimental::mdspan<ElementType, Extents, LayoutPolicy, AccessorPolicy>#

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous, bool is_host_accessible = false, bool is_device_accessible = true, size_t... Extents> constexpr auto raft::make_mdspan(ElementType *ptr, extents<IndexType, Extents...> exts)#

Create a raft::mdspan.

Template Parameters:

ElementType – the data type of the matrix elements
IndexType – the index type of the extents
LayoutPolicy – policy for strides and layout ordering
is_host_accessible – whether the data is accessible on host
is_device_accessible – whether the data is accessible on device

Parameters:

ptr – Pointer to the data
exts – dimensionality of the array (series of integers)

Returns:

raft::mdspan

template<typename IndexType, typename ...Extents, typename = ensure_integral_extents<Extents...>> constexpr auto raft::make_extents(Extents... exts)#

Create raft::extents to specify dimensionality.

Template Parameters:

IndexType – The type of each dimension of the extents
Extents – Dimensions (a series of integers)

Parameters:

exts – The desired dimensions

Returns:

raft::extents

template<typename Extents, typename Strides> auto raft::make_strided_layout(Extents extents, Strides strides)#

Create a layout_stride mapping from extents and strides.

Parameters:

extents – [in] the dimensionality of the layout
strides – [in] the strides between elements in the layout

Returns:

raft::layout_stride::mapping<Extents>

template<typename Idx, typename IndexType, typename LayoutPolicy, size_t... Exts> RAFT_INLINE_FUNCTION auto unravel_index(Idx idx, extents<IndexType, Exts...> shape, LayoutPolicy const &layout)#

Turns linear index into coordinate. Similar to numpy unravel_index.

auto m = make_host_matrix<float>(7, 6);
auto m_v = m.view();
auto coord = unravel_index(2, m.extents(), typename decltype(m)::layout_type{});
std::apply(m_v, coord) = 2;

Parameters:

idx – The linear index.
shape – The shape of the array to use.
layout – Must be layout_c_contiguous (row-major) in current implementation.

Returns:

A std::tuple that represents the coordinate.

template<typename Extents, typename Strides> auto is_c_contiguous(Extents const &extents, Strides const &strides) -> bool#: Whether the strides imply a c-contiguous layout.

template<typename Extents, typename Strides> auto is_f_contiguous(Extents const &extents, Strides const &strides) -> bool#: Whether the strides imply a f-contiguous layout.

template<class ElementType, class Extents, class Layout, class Accessor> auto make_const_mdspan(mdspan<ElementType, Extents, Layout, Accessor> mds)#

Create a copy of the given mdspan with const element type.

Template Parameters:

ElementType – the const-qualified data type of the mdspan elements
Extents – raft::extents for dimensions
Layout – policy for strides and layout ordering
Accessor – Accessor policy for the input and output

Parameters:

mds – raft::mdspan object

Returns:

raft::mdspan

Device Vocabulary#

#include <raft/core/device_mdspan.hpp>

template<typename ElementType, typename Extents, typename LayoutPolicy = layout_c_contiguous, typename AccessorPolicy = std::experimental::default_accessor<ElementType>> using raft::device_mdspan = mdspan<ElementType, Extents, LayoutPolicy, device_accessor<AccessorPolicy>>#: std::experimental::mdspan with device tag to avoid accessing incorrect memory location.

template<typename T, bool B> struct is_device_mdspan : public std::false_type#

template<typename T> using raft::is_device_mdspan_t = is_device_mdspan<T, is_mdspan_v<T>>#: \brief Boolean to determine if template type T is either raft::device_mdspan or a derived type

template<typename T> using raft::is_input_device_mdspan_t = is_device_mdspan<T, is_input_mdspan_v<T>>#

template<typename T> using raft::is_output_device_mdspan_t = is_device_mdspan<T, is_output_mdspan_v<T>>#

template<typename ...Tn> using raft::enable_if_device_mdspan = std::enable_if_t<is_device_mdspan_v<Tn...>>#

template<typename ...Tn> using raft::enable_if_input_device_mdspan = std::enable_if_t<is_input_device_mdspan_v<Tn...>>#

template<typename ...Tn> using raft::enable_if_output_device_mdspan = std::enable_if_t<is_output_device_mdspan_v<Tn...>>#

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous> using raft::device_matrix_view = device_mdspan<ElementType, matrix_extent<IndexType>, LayoutPolicy>#

Shorthand for c-contiguous device matrix view.

Template Parameters:

ElementType – the data type of the matrix elements
IndexType – the index type of the extents
LayoutPolicy – policy for strides and layout ordering

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous> using raft::device_vector_view = device_mdspan<ElementType, vector_extent<IndexType>, LayoutPolicy>#

Shorthand for 1-dim device mdspan.

Template Parameters:

ElementType – the data type of the vector elements
IndexType – the index type of the extents
LayoutPolicy – policy for strides and layout ordering

template<typename ElementType, typename IndexType = std::uint32_t> using raft::device_scalar_view = device_mdspan<ElementType, scalar_extent<IndexType>>#

Shorthand for 0-dim host mdspan (scalar).

Template Parameters:

ElementType – the data type of the scalar element
IndexType – the index type of the extents

Device Factories#

#include <raft/core/device_mdspan.hpp>

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous> auto raft::make_device_matrix_view(ElementType *ptr, IndexType n_rows, IndexType n_cols)#

Create a 2-dim c-contiguous mdspan instance for device pointer. It’s expected that the given layout policy match the layout of the underlying pointer.

Template Parameters:

ElementType – the data type of the matrix elements
LayoutPolicy – policy for strides and layout ordering
IndexType – the index type of the extents

Parameters:

ptr – [in] on device to wrap
n_rows – [in] number of rows in pointer
n_cols – [in] number of columns in pointer

template<typename ElementType, typename IndexType, typename LayoutPolicy = layout_c_contiguous> auto raft::make_device_vector_view(ElementType *ptr, IndexType n)#

Create a 1-dim mdspan instance for device pointer.

Template Parameters:

ElementType – the data type of the vector elements
IndexType – the index type of the extents
LayoutPolicy – policy for strides and layout ordering

Parameters:

ptr – [in] on device to wrap
n – [in] number of elements in pointer

Returns:

raft::device_vector_view

template<typename ElementType, typename IndexType = std::uint32_t> auto raft::make_device_scalar_view(ElementType *ptr)#

Create a 0-dim (scalar) mdspan instance for device value.

Template Parameters:

ElementType – the data type of the matrix elements
IndexType – the index type of the extents

Parameters:

ptr – [in] on device to wrap

Managed Vocabulary#

#include <raft/core/device_mdspan.hpp>

..doxygentypedef:: raft::managed_mdspan

project:: RAFT

template<typename T, bool B> struct is_managed_mdspan : public std::false_type#

template<typename T> using raft::is_managed_mdspan_t = is_managed_mdspan<T, is_mdspan_v<T>>#: \brief Boolean to determine if template type T is either raft::managed_mdspan or a derived type

template<typename T> using raft::is_input_managed_mdspan_t = is_managed_mdspan<T, is_input_mdspan_v<T>>#

template<typename T> using raft::is_output_managed_mdspan_t = is_managed_mdspan<T, is_output_mdspan_v<T>>#

template<typename ...Tn> using raft::enable_if_managed_mdspan = std::enable_if_t<is_managed_mdspan_v<Tn...>>#

template<typename ...Tn> using raft::enable_if_input_managed_mdspan = std::enable_if_t<is_input_managed_mdspan_v<Tn...>>#

template<typename ...Tn> using raft::enable_if_output_managed_mdspan = std::enable_if_t<is_output_managed_mdspan_v<Tn...>>#

Managed Factories#

#include <raft/core/device_mdspan.hpp>

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous, size_t... Extents> auto raft::make_managed_mdspan(ElementType *ptr, extents<IndexType, Extents...> exts)#

Create a raft::managed_mdspan.

Template Parameters:

ElementType – the data type of the matrix elements
IndexType – the index type of the extents
LayoutPolicy – policy for strides and layout ordering

Parameters:

ptr – Pointer to the data
exts – dimensionality of the array (series of integers)

Returns:

raft::managed_mdspan

Host Vocabulary#

#include <raft/core/host_mdspan.hpp>

template<typename ElementType, typename Extents, typename LayoutPolicy = layout_c_contiguous, typename AccessorPolicy = std::experimental::default_accessor<ElementType>> using raft::host_mdspan = mdspan<ElementType, Extents, LayoutPolicy, host_accessor<AccessorPolicy>>#: std::experimental::mdspan with host tag to avoid accessing incorrect memory location.

template<typename T, bool B> struct is_host_mdspan : public std::false_type#

template<typename T> using raft::is_host_mdspan_t = is_host_mdspan<T, is_mdspan_v<T>>#: \brief Boolean to determine if template type T is either raft::host_mdspan or a derived type

template<typename T> using raft::is_input_host_mdspan_t = is_host_mdspan<T, is_input_mdspan_v<T>>#

template<typename T> using raft::is_output_host_mdspan_t = is_host_mdspan<T, is_output_mdspan_v<T>>#

template<typename ...Tn> using raft::enable_if_host_mdspan = std::enable_if_t<is_input_mdspan_v<Tn...>>#

template<typename ...Tn> using raft::enable_if_input_host_mdspan = std::enable_if_t<is_input_host_mdspan_v<Tn...>>#

template<typename ...Tn> using raft::enable_if_output_host_mdspan = std::enable_if_t<is_output_host_mdspan_v<Tn...>>#

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous> using raft::host_matrix_view = host_mdspan<ElementType, matrix_extent<IndexType>, LayoutPolicy>#

Shorthand for c-contiguous host matrix view.

Template Parameters:

ElementType – the data type of the matrix elements
IndexType – the index type of the extents
LayoutPolicy – policy for strides and layout ordering

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous> using raft::host_vector_view = host_mdspan<ElementType, vector_extent<IndexType>, LayoutPolicy>#

Shorthand for 1-dim host mdspan.

Template Parameters:

ElementType – the data type of the vector elements
IndexType – the index type of the extents

template<typename ElementType, typename IndexType = std::uint32_t> using raft::host_scalar_view = host_mdspan<ElementType, scalar_extent<IndexType>>#

Shorthand for 0-dim host mdspan (scalar).

Template Parameters:

ElementType – the data type of the scalar element
IndexType – the index type of the extents

Host Factories#

#include <raft/core/host_mdspan.hpp>

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous> auto raft::make_host_matrix_view(ElementType *ptr, IndexType n_rows, IndexType n_cols)#

Create a 2-dim c-contiguous mdspan instance for host pointer. It’s expected that the given layout policy match the layout of the underlying pointer.

Template Parameters:

ElementType – the data type of the matrix elements
IndexType – the index type of the extents
LayoutPolicy – policy for strides and layout ordering

Parameters:

ptr – [in] on host to wrap
n_rows – [in] number of rows in pointer
n_cols – [in] number of columns in pointer

template<typename ElementType, typename IndexType = std::uint32_t, typename LayoutPolicy = layout_c_contiguous> auto raft::make_host_vector_view(ElementType *ptr, IndexType n)#

Create a 1-dim mdspan instance for host pointer.

Template Parameters:

ElementType – the data type of the vector elements
IndexType – the index type of the extents

Parameters:

ptr – [in] on host to wrap
n – [in] number of elements in pointer

Returns:

raft::host_vector_view

template<typename ElementType, typename IndexType = std::uint32_t> auto raft::make_device_scalar_view(ElementType *ptr)

Create a 0-dim (scalar) mdspan instance for device value.

Template Parameters:

ElementType – the data type of the matrix elements
IndexType – the index type of the extents

Parameters:

ptr – [in] on device to wrap

Validation Routines#

#include <raft/core/mdspan.hpp>

template<typename T, typename = void> struct is_mdspan : public std::false_type#

template<typename T> using raft::is_mdspan_t = is_mdspan<std::remove_const_t<T>>#

template<typename T, typename = void> struct is_input_mdspan : public std::false_type#

template<typename T> using raft::is_input_mdspan_t = is_input_mdspan<T>#

template<typename T, typename = void> struct is_output_mdspan : public std::false_type#

template<typename T> using raft::is_output_mdspan_t = is_output_mdspan<T>#

template<typename ...Tn> using raft::enable_if_mdspan = std::enable_if_t<is_mdspan_v<Tn...>>#

template<typename ...Tn> using raft::enable_if_input_mdspan = std::enable_if_t<is_input_mdspan_v<Tn...>>#

template<typename ...Tn> using raft::enable_if_output_mdspan = std::enable_if_t<is_output_mdspan_v<Tn...>>#