alpaka::api::util Namespace Reference

Namespaces
namespace	detail

Functions
template<std::integral auto T_limit, std::integral auto T_index, std::integral auto T_increment>
consteval auto	adjustToLimit (concepts::CVector auto const input)
	adjust the input vector to a given limit by halving all components until the product of these is below or equal to the limit
auto	adjustToLimit (concepts::Vector auto input, std::integral auto const limit)
	adjust the input vector to a given limit by halving the largest dimension until the product of all components is below or equal to the limit
template<typename T_ValueType, alpaka::concepts::Vector T_Extents>
auto	emulatedAlignedMemDescription (uint32_t alignmentInByte, T_Extents extents)
	provides a memory description to create multidimensional linewise aligned memory within a one dimensional byte area
consteval uint32_t	highestPowerOfTwo (uint32_t value)
template<typename T_ValueType>
constexpr auto	simdOptimizedAlignment (auto api, alpaka::concepts::DeviceKind auto deviceKind)
	Calculate the best alignment for SIMD optimized memory allocation.

Function Documentation

adjustToLimit() [1/2]

template<std::integral auto T_limit, std::integral auto T_index, std::integral auto T_increment>

auto alpaka::api::util::adjustToLimit ( concepts::CVector auto const input )

consteval

adjust the input vector to a given limit by halving all components until the product of these is below or equal to the limit

Definition at line 64 of file util.hpp.

adjustToLimit() [2/2]

auto alpaka::api::util::adjustToLimit ( concepts::Vector auto input, std::integral auto const limit )

inline

adjust the input vector to a given limit by halving the largest dimension until the product of all components is below or equal to the limit

Definition at line 71 of file util.hpp.

emulatedAlignedMemDescription()

template<typename T_ValueType, alpaka::concepts::Vector T_Extents>

auto alpaka::api::util::emulatedAlignedMemDescription ( uint32_t alignmentInByte, T_Extents extents )

inline

provides a memory description to create multidimensional linewise aligned memory within a one dimensional byte area

Parameters

alignment

data alignment in bytes

Returns

tuple with the linearized data blob size in bytes and multi-dimensional pitches, std::tuple(numBytes,pitchMD)

Definition at line 101 of file util.hpp.

highestPowerOfTwo()

uint32_t alpaka::api::util::highestPowerOfTwo ( uint32_t value )

consteval

Definition at line 124 of file util.hpp.

simdOptimizedAlignment()

template<typename T_ValueType>

auto alpaka::api::util::simdOptimizedAlignment ( auto api, alpaka::concepts::DeviceKind auto deviceKind )

inlineconstexpr

Calculate the best alignment for SIMD optimized memory allocation.

Parameters

api	the API to use
deviceKind	the device kind to use

Returns

the best alignment in bytes, will be a power of two value

Definition at line 141 of file util.hpp.