Defines explicit conversions (with and without saturation), truncation etc. More...

Collaboration diagram for Explicit conversion APIs:

Functions
template<typename T_dst , typename T_src , int N>
__XETLA_API std::enable_if_t<!(is_internal_type< T_dst >::value) &&!(is_internal_type< T_src >::value), xetla_vector< T_dst, N > >	gpu::xetla::xetla_cvt (xetla_vector< T_src, N > src)
	xetla explicit data conversion for standard data types(integer,float,half)

template<typename T_dst , typename T_src , int N>
__XETLA_API std::enable_if_t< std::is_same< T_dst, fp16 >::value &&std::is_same< T_src, int32_t >::value, xetla_vector< T_dst, N > >	gpu::xetla::xetla_cvt (xetla_vector< T_src, N > src, float scaling_value)
	xetpp explicit data conversion with scaling, int32->fp16.

Detailed Description

Defines explicit conversions (with and without saturation), truncation etc.

between XeTLA vector types.

Function Documentation

◆ xetla_cvt() [1/2]

template<typename T_dst , typename T_src , int N>

__XETLA_API std::enable_if_t<!(is_internal_type< T_dst >::value) &&!(is_internal_type< T_src >::value), xetla_vector< T_dst, N > > gpu::xetla::xetla_cvt ( xetla_vector< T_src, N > src )

xetla explicit data conversion for standard data types(integer,float,half)

Template Parameters

T_dst	is the destination data type.
T_src	is the source data type.
N	is the element number in xetla_vector.

◆ xetla_cvt() [2/2]