TNL::Algorithms::Reduction3D< Devices::Cuda > Struct Reference

Static Public Member Functions
template<typename Result, typename Fetch, typename Reduction, typename Index, typename Output>
static void	reduce (Result identity, Fetch fetch, Reduction reduction, Index size, int m, int n, Output hostResult)
	Performs reduction on a 3D dataset into a 2D output.

Member Function Documentation

reduce()

template<typename Result, typename Fetch, typename Reduction, typename Index, typename Output>

void TNL::Algorithms::Reduction3D< Devices::Cuda >::reduce ( Result identity, Fetch fetch, Reduction reduction, Index size, int m, int n, Output hostResult )

static

Performs reduction on a 3D dataset into a 2D output.

This function applies a reduction operation across all elements along the first dimension of a 3D dataset, combining values for each position in the second and third dimensions. The reduction starts with an identity element and aggregates all values along the first dimension.

Template Parameters

Result	Type representing the identity element and result values.
Fetch	Callable type used to fetch values from the 3D dataset.
Reduction	Callable type representing the reduction operation.
Index	Integral type used for the first dimension's size.
Output	Callable type used to store the resulting values.

Parameters

identity	[in] The identity element for the reduction operation. This value does not affect the result of the reduction.
fetch	[in] Callable object such that fetch(i, j, k) yields a value from the 3D dataset given by i ranging from 0 to size-1, j from 0 to m-1, and k from 0 to n-1.
reduction	[in] Callable object representing the reduction operation. Examples include instances of std::plus, std::logical_and, std::logical_or, etc.
size	[in] The size of the first dimension (reduction axis) of the 3D dataset.
m	[in] Number of elements in the second dimension of the 3D dataset.
n	[in] Number of elements in the third dimension of the 3D dataset.
hostResult	[out] Callable object returning a modifiable reference to the output array. Used as result(j, k) = value for j=0, ..., m-1 and k=0, ..., n-1. For example, a 2D NDArrayView of size m × n. Note that the output array must be allocated on Host.

---

The documentation for this struct was generated from the following files:

• src/TNL/Algorithms/Reduction3D.h
• src/TNL/Algorithms/Reduction3D.hpp