Tuple

Namespaces
namespace	inner

Data Structures
struct	number_traits< std::tuple< E... > >
	Number description for a tuple of numbers. More...
struct	vector_traits< std::tuple< E... >, arch >
	Tuple of compatible vectors. More...

Description

Tuples of numbers or expressions used for unrolling

---

Data Structure Documentation

number_traits< std::tuple< E... > >

struct number_traits< std::tuple< E... > >

template<typename ... E>
struct number_traits< std::tuple< E... > >

Number description for a tuple of numbers.

Template Parameters

E	Datatype for all numbers

Public Types
typedef std::tuple< typename number_traits< E >::accumulator... >	accumulator
	Tuple of accumulator datatype for the accumulator type.
typedef std::tuple< typename number_traits< E >::compute_type... >	compute_type
	Tuple of compute datatype for the compute type.
Static Public Member Functions
static std::tuple< typename number_traits< E >::accumulator... >	one ()
	Return of tuples of one values.
Static Public Attributes
static constexpr bool	is_float = false
	It is not a float number.
static constexpr bool	is_fixed = false
	It is not a fixed point number.

Member Function Documentation

one()

static std::tuple< typename number_traits< E >::accumulator... > one ( )

inlinestatic

Return of tuples of one values.

Returns

Tuples of one values with different datatypes

vector_traits< std::tuple< E... >, arch >

struct vector_traits< std::tuple< E... >, arch >

template<typename arch, typename ... E>
struct vector_traits< std::tuple< E... >, arch >

Tuple of compatible vectors.

Template Parameters

arch	Current architecture
E	List of vector dataypes

The vector datatypes must be coherent : have same number of lanes or same lane datatype

Public Types
using	RefScalar = typename std::tuple_element< 0, std::tuple< E... > >::type
	First element of tuple defines the scalar datatype.
typedef std::tuple< typename vector_traits< E, arch >::temp_accumulator... >	temp_accumulator
	Temporary accumulator datatype.
typedef std::tuple< typename vector_traits< E, arch >::vector... >	vector
	Vector datatype.
typedef std::tuple< typename vector_traits< E, arch >::real_vector... >	real_vector
	Real vector datatype.
typedef std::tuple< typename vector_traits< E, arch >::predicate_t... >	predicate_t
	Predicate datatype.
Static Public Member Functions
static temp_accumulator	temp_acc_zero ()
	Zero represented with temp accumulator datatype.
Static Public Attributes
static constexpr int	nb_lanes = vector_traits<RefScalar,arch>::nb_lanes
	Number of lanes (from RefScalar)
static constexpr bool	has_vector = vector_traits<RefScalar,arch>::has_vector
	Has vector instructions.
static constexpr bool	is_float = vector_traits<RefScalar,arch>::is_float
	Is a float.
static constexpr bool	is_fixed = vector_traits<RefScalar,arch>::is_fixed
	Is fixed point.
static constexpr bool	has_predicate = vector_traits<RefScalar,arch>::has_predicate
	Has predicated loop.

Member Function Documentation

temp_acc_zero()

static temp_accumulator temp_acc_zero ( )

inlinestatic

Zero represented with temp accumulator datatype.

Returns

Zero represented with temp accumulator datatype