detail Namespace Reference

detail Namespace Reference#

8 min read time

Applies to Linux

rocCV: roccv::detail Namespace Reference

Namespaces
	math

Classes
struct	HasTypeTraits_t

struct	MakeType_t

class	StridedDataWrap

struct	SwizzleIndexMap

struct	SwizzleIndexMap< eSwizzle::XYZW, N >

struct	SwizzleIndexMap< eSwizzle::ZYXW, 1 >

struct	SwizzleIndexMap< eSwizzle::ZYXW, 3 >

struct	SwizzleIndexMap< eSwizzle::ZYXW, 4 >

struct	TypeTraits
	Returns the type traits associated with a datatype. More...

Typedefs
template<typename T >
using	BaseType = typename TypeTraits< T >::base_type
	Returns the base type of a given HIP vectorized type. More...

template<typename T , int C>
using	MakeType = MakeType_t< T, C >::type

Functions
	DEFINE_TYPE_TRAITS_0_TO_4 (char, signed char)

	DEFINE_TYPE_TRAITS_0_TO_4 (double, double)

	DEFINE_TYPE_TRAITS_0_TO_4 (float, float)

	DEFINE_TYPE_TRAITS_0_TO_4 (int, signed int)

	DEFINE_TYPE_TRAITS_0_TO_4 (short, signed short)

	DEFINE_TYPE_TRAITS_0_TO_4 (uchar, unsigned char)

	DEFINE_TYPE_TRAITS_0_TO_4 (uint, unsigned int)

	DEFINE_TYPE_TRAITS_0_TO_4 (ushort, unsigned short)

template<typename T , size_t... I>
__host__ constexpr __device__ T	ExpandSwizzle (const T &v, std::integer_sequence< size_t, I... >)

template<eTensorLayout LAYOUT>
auto	get_sdwrapper (const roccv::Tensor &tensor) -> const StridedDataWrap< layout_get_rank< LAYOUT >()>

template<eTensorLayout LAYOUT>
auto	get_sdwrapper (roccv::Tensor &tensor) -> StridedDataWrap< layout_get_rank< LAYOUT >()>

template<>
auto	get_sdwrapper< TENSOR_LAYOUT_LNHWC > (const roccv::Tensor &tensor) -> const StridedDataWrap< layout_get_rank< TENSOR_LAYOUT_LNHWC >()>

template<>
auto	get_sdwrapper< TENSOR_LAYOUT_NHWC > (const roccv::Tensor &tensor) -> const StridedDataWrap< layout_get_rank< TENSOR_LAYOUT_NHWC >()>

template<>
auto	get_sdwrapper< TENSOR_LAYOUT_NMC > (const roccv::Tensor &tensor) -> const StridedDataWrap< layout_get_rank< TENSOR_LAYOUT_NMC >()>

template<>
auto	get_sdwrapper< TENSOR_LAYOUT_NMD > (const roccv::Tensor &tensor) -> const StridedDataWrap< layout_get_rank< TENSOR_LAYOUT_NMD >()>

template<typename T , typename RT = BaseType<T>, class = std::enable_if_t<HasTypeTraits<T>>>
__host__ __device__ RT &	GetElement (T &v, int idx)

template<eTensorLayout LAYOUT>
constexpr size_t	layout_get_rank ()

template<>
constexpr size_t	layout_get_rank< TENSOR_LAYOUT_HWC > ()

template<>
constexpr size_t	layout_get_rank< TENSOR_LAYOUT_LNHWC > ()

template<>
constexpr size_t	layout_get_rank< TENSOR_LAYOUT_NHWC > ()

template<>
constexpr size_t	layout_get_rank< TENSOR_LAYOUT_NMC > ()

template<>
constexpr size_t	layout_get_rank< TENSOR_LAYOUT_NMD > ()

template<typename T , typename U , class = std::enable_if_t<(HasTypeTraits<T> && HasTypeTraits<U>) && (NumElements<T> <= NumElements<U>)>>
__device__ __host__ T	SaturateCast (U v)
	Performs a saturation cast from one type to another. Each type must have type traits supported. A saturation cast converts one type to another, clamping values down to the minimum/maximum of the desired cast if the value being cast goes out of bounds. For example, casting a float with a value of 256.0f to a uchar would result in a uchar with value 255. More...

template<typename T , typename U , class = std::enable_if_t<(HasTypeTraits<T> && HasTypeTraits<U>) && (!IsCompound<T> && !IsCompound<U>)>>
__device__ __host__ T	ScalarRangeCast (U v)
	ScalarRangeCast is for implementation purposes only. Use RangeCast directly instead. More...

template<typename T , typename U , class = std::enable_if_t<!IsCompound<T> && !IsCompound<U>>>
__device__ __host__ T	ScalarSaturateCast (U v)
	ScalarSaturateCast is for implementation purposes only. Use SaturateCast directly. More...

template<typename T , typename Base = BaseType<T>, class = std::enable_if_t<HasTypeTraits<T>>>
__device__ constexpr __host__ T	SetAll (Base v)
	Create a vectorized type with all values set to v. More...

template<eSwizzle Pattern, typename T >
__host__ constexpr __device__ T	Swizzle (const T &v)
	Rearranges the components of a vector according to a given swizzle pattern. More...

template<typename T , size_t... Indices>
__host__ constexpr __device__ T	SwizzleIndicesImpl (const T &v)

Variables
template<typename T >
constexpr bool	HasTypeTraits = HasTypeTraits_t<T>::value
	Returns whether the datatype passed in has type traits associated with it. More...

template<typename T >
constexpr bool	IsCompound = TypeTraits<T>::components != 0

template<typename T >
constexpr int	NumComponents = TypeTraits<T>::components
	Returns the number of components in a given type. Scalar types will have 0 components, while vectorized types will have 1-4 components typically. More...

template<typename T >
constexpr int	NumElements = TypeTraits<T>::elements
	Returns the number of elements in a HIP vectorized type. For example: uchar3 will return 3, int2 will return 2. More...

Detailed Description

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Typedef Documentation

◆ BaseType

template<typename T >

using roccv::detail::BaseType = typedef typename TypeTraits<T>::base_type

Returns the base type of a given HIP vectorized type.

Template Parameters

T	A HIP vectorized type.

◆ MakeType

template<typename T , int C>

using roccv::detail::MakeType = typedef MakeType_t<T, C>::type

Function Documentation

◆ DEFINE_TYPE_TRAITS_0_TO_4() [1/8]

roccv::detail::DEFINE_TYPE_TRAITS_0_TO_4	(	char	,
		signed char
	)

◆ DEFINE_TYPE_TRAITS_0_TO_4() [2/8]

roccv::detail::DEFINE_TYPE_TRAITS_0_TO_4	(	double	,
		double
	)

◆ DEFINE_TYPE_TRAITS_0_TO_4() [3/8]

roccv::detail::DEFINE_TYPE_TRAITS_0_TO_4	(	float	,
		float
	)

◆ DEFINE_TYPE_TRAITS_0_TO_4() [4/8]

roccv::detail::DEFINE_TYPE_TRAITS_0_TO_4	(	int	,
		signed int
	)

◆ DEFINE_TYPE_TRAITS_0_TO_4() [5/8]

roccv::detail::DEFINE_TYPE_TRAITS_0_TO_4	(	short	,
		signed short
	)

◆ DEFINE_TYPE_TRAITS_0_TO_4() [6/8]

roccv::detail::DEFINE_TYPE_TRAITS_0_TO_4	(	uchar	,
		unsigned char
	)

◆ DEFINE_TYPE_TRAITS_0_TO_4() [7/8]

roccv::detail::DEFINE_TYPE_TRAITS_0_TO_4	(	uint	,
		unsigned int
	)

◆ DEFINE_TYPE_TRAITS_0_TO_4() [8/8]

roccv::detail::DEFINE_TYPE_TRAITS_0_TO_4	(	ushort	,
		unsigned short
	)

◆ ExpandSwizzle()

template<typename T , size_t... I>

__host__ constexpr __device__ T roccv::detail::ExpandSwizzle	(	const T &	v,
		std::integer_sequence< size_t, I... >
	)

constexpr

◆ get_sdwrapper() [1/2]

template<eTensorLayout LAYOUT>

auto roccv::detail::get_sdwrapper ( const roccv::Tensor & tensor ) -> const StridedDataWrap< layout_get_rank< LAYOUT >()>

inline

◆ get_sdwrapper() [2/2]

template<eTensorLayout LAYOUT>

auto roccv::detail::get_sdwrapper ( roccv::Tensor & tensor ) -> StridedDataWrap<layout_get_rank<LAYOUT>()>

inline

◆ get_sdwrapper< TENSOR_LAYOUT_LNHWC >()

template<>

auto roccv::detail::get_sdwrapper< TENSOR_LAYOUT_LNHWC > ( const roccv::Tensor & tensor ) -> const StridedDataWrap<layout_get_rank<TENSOR_LAYOUT_LNHWC>()>

inline

◆ get_sdwrapper< TENSOR_LAYOUT_NHWC >()

template<>

auto roccv::detail::get_sdwrapper< TENSOR_LAYOUT_NHWC > ( const roccv::Tensor & tensor ) -> const StridedDataWrap<layout_get_rank<TENSOR_LAYOUT_NHWC>()>

inline

◆ get_sdwrapper< TENSOR_LAYOUT_NMC >()

template<>

auto roccv::detail::get_sdwrapper< TENSOR_LAYOUT_NMC > ( const roccv::Tensor & tensor ) -> const StridedDataWrap<layout_get_rank<TENSOR_LAYOUT_NMC>()>

inline

◆ get_sdwrapper< TENSOR_LAYOUT_NMD >()

template<>

auto roccv::detail::get_sdwrapper< TENSOR_LAYOUT_NMD > ( const roccv::Tensor & tensor ) -> const StridedDataWrap<layout_get_rank<TENSOR_LAYOUT_NMD>()>

inline

◆ GetElement()

template<typename T , typename RT = BaseType<T>, class = std::enable_if_t<HasTypeTraits<T>>>

__host__ __device__ RT& roccv::detail::GetElement	(	T &	v,
		int	idx
	)

◆ layout_get_rank()

template<eTensorLayout LAYOUT>

constexpr size_t roccv::detail::layout_get_rank ( )

constexpr

◆ layout_get_rank< TENSOR_LAYOUT_HWC >()

template<>

constexpr size_t roccv::detail::layout_get_rank< TENSOR_LAYOUT_HWC > ( )

constexpr

◆ layout_get_rank< TENSOR_LAYOUT_LNHWC >()

template<>

constexpr size_t roccv::detail::layout_get_rank< TENSOR_LAYOUT_LNHWC > ( )

constexpr

◆ layout_get_rank< TENSOR_LAYOUT_NHWC >()

template<>

constexpr size_t roccv::detail::layout_get_rank< TENSOR_LAYOUT_NHWC > ( )

constexpr

◆ layout_get_rank< TENSOR_LAYOUT_NMC >()

template<>

constexpr size_t roccv::detail::layout_get_rank< TENSOR_LAYOUT_NMC > ( )

constexpr

◆ layout_get_rank< TENSOR_LAYOUT_NMD >()

template<>

constexpr size_t roccv::detail::layout_get_rank< TENSOR_LAYOUT_NMD > ( )

constexpr

◆ SaturateCast()

template<typename T , typename U , class = std::enable_if_t<(HasTypeTraits<T> && HasTypeTraits) && (NumElements<T> <= NumElements)>>

__device__ __host__ T roccv::detail::SaturateCast ( U v )

Performs a saturation cast from one type to another. Each type must have type traits supported. A saturation cast converts one type to another, clamping values down to the minimum/maximum of the desired cast if the value being cast goes out of bounds. For example, casting a float with a value of 256.0f to a uchar would result in a uchar with value 255.

Template Parameters

T	The type to cast <v> to. The number of elements in this type must be <= to that of v.

Parameters

[in] v The value to cast.

Returns: The values in v saturate casted to type T.

◆ ScalarRangeCast()

template<typename T , typename U , class = std::enable_if_t<(HasTypeTraits<T> && HasTypeTraits) && (!IsCompound<T> && !IsCompound)>>

__device__ __host__ T roccv::detail::ScalarRangeCast ( U v )

ScalarRangeCast is for implementation purposes only. Use RangeCast directly instead.

◆ ScalarSaturateCast()

template<typename T , typename U , class = std::enable_if_t<!IsCompound<T> && !IsCompound>>

__device__ __host__ T roccv::detail::ScalarSaturateCast ( U v )

ScalarSaturateCast is for implementation purposes only. Use SaturateCast directly.

◆ SetAll()

template<typename T , typename Base = BaseType<T>, class = std::enable_if_t<HasTypeTraits<T>>>

__device__ constexpr __host__ T roccv::detail::SetAll ( Base v )

constexpr

Create a vectorized type with all values set to v.

Template Parameters

T	The vectorized type.

Parameters

v	The value to set. Value must be the same as the base type of type T.

Returns: A vectorized type with all values set to v.

◆ Swizzle()

template<eSwizzle Pattern, typename T >

__host__ constexpr __device__ T roccv::detail::Swizzle ( const T & v )

constexpr

Rearranges the components of a vector according to a given swizzle pattern.

Template Parameters

Pattern	The pattern used to rearrange vector components.
T	The vector type.

Parameters

v	The vector to swizzle.

Returns: v with its components rearranged according to Pattern.

Example:: uchar3 vec = make_uchar3(1, 2, 3);

uchar3 vecSwizzled = Swizzle<eSwizzle::ZYXW>(vec);

// vecSwizzled = {3, 2, 1}

◆ SwizzleIndicesImpl()

template<typename T , size_t... Indices>

__host__ constexpr __device__ T roccv::detail::SwizzleIndicesImpl ( const T & v )

constexpr

Variable Documentation

◆ HasTypeTraits

template<typename T >

constexpr bool roccv::detail::HasTypeTraits = HasTypeTraits_t<T>::value

constexpr

Returns whether the datatype passed in has type traits associated with it.

Template Parameters

T

◆ IsCompound

template<typename T >

constexpr bool roccv::detail::IsCompound = TypeTraits<T>::components != 0

constexpr

◆ NumComponents

template<typename T >

constexpr int roccv::detail::NumComponents = TypeTraits<T>::components

constexpr

Returns the number of components in a given type. Scalar types will have 0 components, while vectorized types will have 1-4 components typically.

Template Parameters

T

◆ NumElements

template<typename T >

constexpr int roccv::detail::NumElements = TypeTraits<T>::elements

constexpr

Returns the number of elements in a HIP vectorized type. For example: uchar3 will return 3, int2 will return 2.

Template Parameters

T

detail Namespace Reference

detail Namespace Reference#

Namespaces

Classes

Typedefs

Functions

Variables

Detailed Description

Typedef Documentation

◆ BaseType

◆ MakeType

Function Documentation

◆ DEFINE_TYPE_TRAITS_0_TO_4() [1/8]

◆ DEFINE_TYPE_TRAITS_0_TO_4() [2/8]

◆ DEFINE_TYPE_TRAITS_0_TO_4() [3/8]

◆ DEFINE_TYPE_TRAITS_0_TO_4() [4/8]

◆ DEFINE_TYPE_TRAITS_0_TO_4() [5/8]

◆ DEFINE_TYPE_TRAITS_0_TO_4() [6/8]

◆ DEFINE_TYPE_TRAITS_0_TO_4() [7/8]

◆ DEFINE_TYPE_TRAITS_0_TO_4() [8/8]

◆ ExpandSwizzle()

◆ get_sdwrapper() [1/2]

◆ get_sdwrapper() [2/2]

◆ get_sdwrapper< TENSOR_LAYOUT_LNHWC >()

◆ get_sdwrapper< TENSOR_LAYOUT_NHWC >()

◆ get_sdwrapper< TENSOR_LAYOUT_NMC >()

◆ get_sdwrapper< TENSOR_LAYOUT_NMD >()

◆ GetElement()

◆ layout_get_rank()

◆ layout_get_rank< TENSOR_LAYOUT_HWC >()

◆ layout_get_rank< TENSOR_LAYOUT_LNHWC >()

◆ layout_get_rank< TENSOR_LAYOUT_NHWC >()

◆ layout_get_rank< TENSOR_LAYOUT_NMC >()

◆ layout_get_rank< TENSOR_LAYOUT_NMD >()

◆ SaturateCast()

◆ ScalarRangeCast()

◆ ScalarSaturateCast()

◆ SetAll()

◆ Swizzle()

◆ SwizzleIndicesImpl()

Variable Documentation

◆ HasTypeTraits

◆ IsCompound

◆ NumComponents

◆ NumElements