Half

Namespaces
namespace	inner

Data Structures
struct	number_traits< float16_t >
	Feature of float16 datatype. More...
struct	vector_traits< float16_t >
	float16 vector descrition when no vector architecture More...

Description

---

Data Structure Documentation

number_traits< float16_t >

struct number_traits< float16_t >

Feature of float16 datatype.

Public Types
typedef float16_t	accumulator
	Accumulator datatype.
typedef _Float16	compute_type
	Compute datatype.
Static Public Member Functions
static constexpr float16_t	one ()
	One value.
Static Public Attributes
static constexpr bool	is_float = true
	It is a float number.
static constexpr bool	is_fixed = false
	It is not a fixed point number.

Member Function Documentation

one()

static constexpr float16_t one ( )

inlinestaticconstexpr

One value.

Returns

One value in f16

vector_traits< float16_t >

struct vector_traits< float16_t >

float16 vector descrition when no vector architecture

Public Types
typedef float16_t	type
	Float16 datatype.
typedef float16_t	storage_type
	Float16 storage type.
typedef bool	vector
	Dummy type when no vector instruction is supported.
typedef bool	temp_accumulator
	Dummy type when no vector instruction is supported.
typedef uint32_t	predicate_t
	Dummy type when no vector instruction is supported.
Static Public Attributes
static constexpr bool	has_vector = false
	No vector instruction.
static constexpr bool	is_float = true
	Is float.
static constexpr bool	is_fixed = false
	Not fixed point.
static constexpr bool	has_predicate = false
	Has predicated loop.