gpu::xetla::subgroup::detail Namespace Reference

Classes
struct	check_load_type
struct	check_prefetch_type
struct	check_store_type
struct	gcd
struct	gcd< a, 0 >
struct	NextPowerOf2
	Compute next power of 2 of a constexpr with guaranteed compile-time evaluation. More...
struct	NextPowerOf2< N, K, false >
struct	NextPowerOf2< N, K, true >

Enumerations
enum class	process_flag : uint8_t { load = 0 , store = 1 }

Functions
template<uint32_t N>
constexpr uint32_t	getNextPowerOf2 ()
	Get the Next Power Of2 object.
template<>
constexpr uint32_t	getNextPowerOf2< 0 > ()
	Get the Next Power Of2<0> object.
template<uint32_t remained_len, uint32_t base_len, process_flag flag, cache_hint L1, cache_hint L2, typename payload_t , typename tile_t >
__XETLA_API std::enable_if_t< base_len==0 >	process_1d_tail (tile_t &tile, payload_t &payload, uint32_t offset)
template<uint32_t remained_len, uint32_t base_len, process_flag flag, cache_hint L1, cache_hint L2, typename payload_t , typename tile_t >
__XETLA_API std::enable_if_t< base_len !=0 &&payload_t::memory_space==mem_space::global >	process_1d_tail (tile_t &tile, payload_t &payload, uint32_t offset)
template<uint32_t remained_len, uint32_t base_len, process_flag flag, cache_hint L1, cache_hint L2, typename payload_t , typename tile_t >
__XETLA_API std::enable_if_t< base_len !=0 &&payload_t::memory_space==mem_space::local >	process_1d_tail (tile_t &tile, payload_t &payload, uint32_t offset)
template<uint32_t remained_len, uint32_t base_len, cache_hint L1, cache_hint L2, typename payload_t >
__XETLA_API std::enable_if_t<(base_len< 8)>	process_1d_tail (payload_t &payload, uint32_t offset)

Variables
__XETLA_API uint32_t	size_x
__XETLA_API uint32_t uint32_t	size_y
__XETLA_API uint32_t uint32_t uint32_t	scale_factor
__XETLA_API uint32_t uint32_t uint32_t uint8_t	arr_len

Enumeration Type Documentation

process_flag

enum class gpu::xetla::subgroup::detail::process_flag : uint8_t

strong

Enumerator
load
store

Function Documentation

getNextPowerOf2()

template<uint32_t N>

constexpr uint32_t gpu::xetla::subgroup::detail::getNextPowerOf2 ( )

constexpr

Get the Next Power Of2 object.

Template Parameters

N

Returns

constexpr uint32_t

getNextPowerOf2< 0 >()

template<>

constexpr uint32_t gpu::xetla::subgroup::detail::getNextPowerOf2< 0 > ( )

constexpr

Get the Next Power Of2<0> object.

Template Parameters

return constexpr uint32_t

process_1d_tail() [1/4]

template<uint32_t remained_len, uint32_t base_len, cache_hint L1, cache_hint L2, typename payload_t >

__XETLA_API std::enable_if_t<(base_len< 8)> gpu::xetla::subgroup::detail::process_1d_tail	(	payload_t &	payload,
		uint32_t	offset
	)

process_1d_tail() [2/4]

template<uint32_t remained_len, uint32_t base_len, process_flag flag, cache_hint L1, cache_hint L2, typename payload_t , typename tile_t >

__XETLA_API std::enable_if_t< base_len==0 > gpu::xetla::subgroup::detail::process_1d_tail	(	tile_t &	tile,
		payload_t &	payload,
		uint32_t	offset
	)

process_1d_tail() [3/4]

template<uint32_t remained_len, uint32_t base_len, process_flag flag, cache_hint L1, cache_hint L2, typename payload_t , typename tile_t >

__XETLA_API std::enable_if_t< base_len !=0 &&payload_t::memory_space==mem_space::global > gpu::xetla::subgroup::detail::process_1d_tail	(	tile_t &	tile,
		payload_t &	payload,
		uint32_t	offset
	)

process_1d_tail() [4/4]

template<uint32_t remained_len, uint32_t base_len, process_flag flag, cache_hint L1, cache_hint L2, typename payload_t , typename tile_t >

__XETLA_API std::enable_if_t< base_len !=0 &&payload_t::memory_space==mem_space::local > gpu::xetla::subgroup::detail::process_1d_tail	(	tile_t &	tile,
		payload_t &	payload,
		uint32_t	offset
	)

Variable Documentation

arr_len

__XETLA_API uint32_t uint32_t uint32_t uint8_t gpu::xetla::subgroup::detail::arr_len

scale_factor

__XETLA_API uint32_t uint32_t uint32_t gpu::xetla::subgroup::detail::scale_factor

size_x

__XETLA_API uint32_t gpu::xetla::subgroup::detail::size_x

size_y

__XETLA_API uint32_t uint32_t gpu::xetla::subgroup::detail::size_y