SSE Intrinsics各函数介绍

最新推荐文章于 2022-06-21 11:28:20 发布

有点模糊

最新推荐文章于 2022-06-21 11:28:20 发布

阅读量357

点赞数

本文链接：https://blog.csdn.net/hddryjv/article/details/87349598

版权

SIMD相关头文件包括：

//#include <ivec.h>//MMX//#include <fvec.h>//SSE(also include ivec.h)//#include <dvec.h>//SSE2(also include fvec.h)#include <mmintrin.h> //MMX#include <xmmintrin.h> //SSE(include mmintrin.h)#include <emmintrin.h> //SSE2(include xmmintrin.h)#include <pmmintrin.h> //SSE3(include emmintrin.h)#include <tmmintrin.h>//SSSE3(include pmmintrin.h)#include <smmintrin.h>//SSE4.1(include tmmintrin.h)#include <nmmintrin.h>//SSE4.2(include smmintrin.h)#include <wmmintrin.h>//AES(include nmmintrin.h)#include <immintrin.h>//AVX(include wmmintrin.h)#include <intrin.h>//(include immintrin.h)

mmintrin.h为MMX头文件，其中__m64的定义为：

typedef union __declspec(intrin_type) _CRT_ALIGN(8) __m64{    unsigned __int64    m64_u64;    float               m64_f32[2];    __int8              m64_i8[8];    __int16             m64_i16[4];    __int32             m64_i32[2];        __int64             m64_i64;    unsigned __int8     m64_u8[8];    unsigned __int16    m64_u16[4];    unsigned __int32    m64_u32[2];} __m64;

xmmintrin.h为SSE头文件，此头文件里包含MMX头文件，其中__m128的定义为：

typedef union __declspec(intrin_type) _CRT_ALIGN(16) __m128 {     float               m128_f32[4];     unsigned __int64    m128_u64[2];     __int8              m128_i8[16];     __int16             m128_i16[8];     __int32             m128_i32[4];     __int64             m128_i64[2];     unsigned __int8     m128_u8[16];     unsigned __int16    m128_u16[8];     unsigned __int32    m128_u32[4]; } __m128;

xmmintrin.h文件中各函数的介绍：

 /*----------Floating Point Intrinsics Using Streaming SIMD Extensions------------*/ //Arithmetic Operations(Floating Point ):add、sub、mul、div、sqrt、rcp、min、max //---------------------说明：_ps结尾的指令表示对4个单精度浮点数同时进行运算， //_ss结尾的指令表示仅对4个单精度浮点数最低位的浮点数进行运算--------------------- //返回一个__m128的寄存器，仅将寄存器_A和寄存器_B最低对应位置的32bit单精度浮点数相加， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器为r=(_A0+_B0, _A1, _A2, _A3) extern __m128 _mm_add_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数相加， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=_A0+_B0, r1=_A1+_B1, r2=_A2+_B2, r3=_A3+_B3 extern __m128 _mm_add_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，仅将寄存器_A和寄存器_B最低对应位置的32bit单精度浮点数相减， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器为r=(_A0-_B0, _A1, _A2, _A3) extern __m128 _mm_sub_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数相减， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=_A0-_B0, r1=_A1-_B1, r2=_A2-_B2, r3=_A3-_B3 extern __m128 _mm_sub_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，仅将寄存器_A和寄存器_B最低对应位置的32bit单精度浮点数相乘， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器为r=(_A0*_B0, _A1, _A2, _A3) extern __m128 _mm_mul_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数相乘， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=_A0*_B0, r1=_A1*_B1, r2=_A2*_B2, r3=_A3*_B3 extern __m128 _mm_mul_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，仅将寄存器_A和寄存器_B最低对应位置的32bit单精度浮点数相除， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器为r=(_A0/_B0, _A1, _A2, _A3) extern __m128 _mm_div_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数相除， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=_A0/_B0, r1=_A1/_B1, r2=_A2/_B2, r3=_A3/_B3 extern __m128 _mm_div_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，仅将寄存器_A最低对应位置的32bit单精度浮点数开平方， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3) //则返回寄存器为r=(sqrt(_A0), _A1, _A2, _A3) extern __m128 _mm_sqrt_ss(__m128 _A); //返回一个__m128的寄存器，将寄存器_A中4个32bit单精度浮点数开平方， //例如_A=(_A0,_A1,_A2,_A3)，则返回寄存器为 //r=(sqrt(_A0), sqrt(_A1), sqrt(_A2), sqrt(_A3)) extern __m128 _mm_sqrt_ps(__m128 _A); //返回一个__m128的寄存器，仅将寄存器_A最低对应位置的32bit单精度浮点数取倒数， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3) //则返回寄存器为r=(recip(_A0), _A1, _A2, _A3) extern __m128 _mm_rcp_ss(__m128 _A); //返回一个__m128的寄存器，将寄存器_A中4个32bit单精度浮点数取倒数， //例如_A=(_A0,_A1,_A2,_A3)，则返回寄存器为 //r=(recip(_A0), recip(_A1), recip(_A2), recip(_A3)) extern __m128 _mm_rcp_ps(__m128 _A); //返回一个__m128的寄存器，仅将寄存器_A最低对应位置的32bit单精度浮点数取平方根的倒数， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3) //则返回寄存器为r=(recip(sqrt(_A0)), _A1, _A2, _A3) extern __m128 _mm_rsqrt_ss(__m128 _A); //返回一个__m128的寄存器，将寄存器_A中4个32bit单精度浮点数取平方根的倒数， //例如_A=(_A0,_A1,_A2,_A3)，则返回寄存器为 //r=(recip(sqrt(_A0)), recip(sqrt(_A1)), recip(sqrt(_A2)), recip(sqrt(_A3))) extern __m128 _mm_rsqrt_ps(__m128 _A); //返回一个__m128的寄存器，仅将寄存器_A和寄存器_B最低对应位置的32bit单精度浮点数取最小值， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器为r=(min(_A0,_B0), _A1, _A2, _A3) extern __m128 _mm_min_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数取最小值， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=min(_A0,_B0), r1=min(_A1,_B1), r2=min(_A2,_B2), r3=min(_A3,_B3) extern __m128 _mm_min_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，仅将寄存器_A和寄存器_B最低对应位置的32bit单精度浮点数取最大值， //其余位置取寄存器_A中的数据,例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器为r=(max(_A0,_B0), _A1, _A2, _A3) extern __m128 _mm_max_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数取最大值， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=max(_A0,_B0), r1=max(_A1,_B1), r2=max(_A2,_B2), r3=max(_A3,_B3) extern __m128 _mm_max_ps(__m128 _A, __m128 _B); //Logical Operations(SSE)：and、andnot、or、xor //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数分别进行按位与运算， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=_A0 & _B0, r1=_A1 & _B1, r2=_A2 & _B2, r3=_A3 & _B3 extern __m128 _mm_and_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A对应位置的32bit单精度浮点数的非和寄存器_B对应位置的32bit //单精度浮点数分别进行按位与运算，例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=~_A0 & _B0, r1=~_A1 & _B1, r2=~_A2 & _B2, r3=~_A3 & _B3 extern __m128 _mm_andnot_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数分别进行按位或运算， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=_A0 | _B0, r1=_A1 | _B1, r2=_A2 | _B2, r3=_A3 | _B3 extern __m128 _mm_or_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，将寄存器_A和_B的对应位置的32bit单精度浮点数分别进行按位异或运算， //例如_A=(_A0,_A1,_A2,_A3), _B=(_B0,_B1,_B2,_B3), //则返回寄存器r0=_A0 ^ _B0, r1=_A1 ^ _B1, r2=_A2 ^ _B2, r3=_A3 ^ _B3 extern __m128 _mm_xor_ps(__m128 _A, __m128 _B); //Comparison Intrinsics(SSE):==、<、<=、>、>=、!=、不小于、不小于等于、不大于、不大于等于 //返回一个__m128的寄存器，Compares for equality, //r0=(_A0 == _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpeq_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for equality, //r0=(_A0 == _B0) ? 0xffffffff : 0x0, r1=(_A1 == _B1) ? 0xffffffff : 0x0,  //r2=(_A2 == _B2) ? 0xffffffff : 0x0, r3=(_A3 == _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpeq_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for less than, //r0=(_A0 < _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmplt_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for less than, //r0=(_A0 < _B0) ? 0xffffffff : 0x0, r1=(_A1 < _B1) ? 0xffffffff : 0x0,  //r2=(_A2 < _B2) ? 0xffffffff : 0x0, r3=(_A3 < _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmplt_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for less than or equal, //r0=(_A0 <= _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmple_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for less than or equal, //r0=(_A0 <= _B0) ? 0xffffffff : 0x0, r1=(_A1 <= _B1) ? 0xffffffff : 0x0,  //r2=(_A2 <= _B2) ? 0xffffffff : 0x0, r3=(_A3 <= _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmple_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for greater than, //r0=(_A0 > _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpgt_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for greater than, //r0=(_A0 > _B0) ? 0xffffffff : 0x0, r1=(_A1 > _B1) ? 0xffffffff : 0x0,  //r2=(_A2 > _B2) ? 0xffffffff : 0x0, r3=(_A3 > _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpgt_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for greater than or equal, //r0=(_A0 >= _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpge_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for greater than or equal, //r0=(_A0 >= _B0) ? 0xffffffff : 0x0, r1=(_A1 >= _B1) ? 0xffffffff : 0x0,  //r2=(_A2 >= _B2) ? 0xffffffff : 0x0, r3=(_A3 >= _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpge_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for inequality, //r0=(_A0 != _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpneq_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for inequality, //r0=(_A0 != _B0) ? 0xffffffff : 0x0, r1=(_A1 != _B1) ? 0xffffffff : 0x0,  //r2=(_A2 != _B2) ? 0xffffffff : 0x0, r3=(_A3 != _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpneq_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for not less than, //r0= !(_A0 < _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpnlt_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for not less than, //r0=!(_A0 < _B0) ? 0xffffffff : 0x0, r1=!(_A1 < _B1) ? 0xffffffff : 0x0,  //r2=!(_A2 < _B2) ? 0xffffffff : 0x0, r3=!(_A3 < _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpnlt_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for not less than or equal //r0= !(_A0 <= _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpnle_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for not less than or equal //r0=!(_A0 <= _B0) ? 0xffffffff : 0x0, r1=!(_A1 <= _B1) ? 0xffffffff : 0x0,  //r2=!(_A2 <= _B2) ? 0xffffffff : 0x0, r3=!(_A3 <= _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpnle_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for not greater than, //r0=!(_A0 > _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpngt_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for not greater than, //r0=!(_A0 > _B0) ? 0xffffffff : 0x0, r1=!(_A1 > _B1) ? 0xffffffff : 0x0,  //r2=!(_A2 > _B2) ? 0xffffffff : 0x0, r3=!(_A3 > _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpngt_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for not greater than or equal, //r0=!(_A0 >= _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpnge_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for not greater than or equal, //r0=!(_A0 >= _B0) ? 0xffffffff : 0x0, r1=!(_A1 >= _B1) ? 0xffffffff : 0x0,  //r2=!(_A2 >= _B2) ? 0xffffffff : 0x0, r3=!(_A3 >= _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpnge_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for ordered, //r0=(_A0 ord? _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cmpord_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for ordered, //r0=(_A0 ord? _B0) ? 0xffffffff : 0x0, r1=(_A1 ord? _B1) ? 0xffffffff : 0x0,  //r2=(_A2 ord? _B2) ? 0xffffffff : 0x0, r3=(_A3 ord? _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpord_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for unordered, //r0=(_A0 unord? _B0) ? 0xffffffff : 0x0, r1=_A1, r2=_A2, r3=_A3  extern __m128 _mm_cmpunord_ss(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Compares for unordered, //r0=(_A0 unord? _B0) ? 0xffffffff : 0x0, r1=(_A1 unord? _B1) ? 0xffffffff : 0x0,  //r2=(_A2 unord? _B2) ? 0xffffffff : 0x0, r3=(_A3 unord? _B3) ? 0xffffffff : 0x0 extern __m128 _mm_cmpunord_ps(__m128 _A, __m128 _B); //返回一个0或1的整数，Compares the lower single-precision, floating-point value of //a and b for a equal to b,If a and b are equal, 1 is returned. Otherwise, //0 is returned. If a or b is a NaN, 1 is returned //r=(_A0 == _B0) ? 0x1 : 0x0 extern int _mm_comieq_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a is less than b, 1 is returned. Otherwise,  //0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 < _B0) ? 0x1 : 0x0 extern int _mm_comilt_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a is less than or equal to b, 1 is returned.  //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 <= _B0) ? 0x1 : 0x0 extern int _mm_comile_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a is greater than b, 1 is returned. //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 > _B0) ? 0x1 : 0x0 extern int _mm_comigt_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a is greater than or equal to b, 1 is returned.  //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 >= _B0) ? 0x1 : 0x0 extern int _mm_comige_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a and b are not equal, 1 is returned.  //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 != _B0) ? 0x1 : 0x0 extern int _mm_comineq_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a and b are equal, 1 is returned.  //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 == _B0) ? 0x1 : 0x0  extern int _mm_ucomieq_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a is less than b , 1 is returned.  //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 < _B0) ? 0x1 : 0x0 extern int _mm_ucomilt_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a is less than or equal to b, 1 is returned.  //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 <= _B0) ? 0x1 : 0x0 extern int _mm_ucomile_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a is greater than b, 1 is returned.  //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 > _B0) ? 0x1 : 0x0 extern int _mm_ucomigt_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a is greater than or equal to b, 1 is returned. //Otherwise, 0 is returned,r=(_A0 >= _B0) ? 0x1 : 0x0 extern int _mm_ucomige_ss(__m128 _A, __m128 _B); //返回一个0或1的整数，If a and b are not equal, 1 is returned.  //Otherwise, 0 is returned. If a or b is a NaN, 1 is returned, //r=(_A0 != _B0) ? 0x1 : 0x0 extern int _mm_ucomineq_ss(__m128 _A, __m128 _B); //Conversion Operations(SSE) //返回一个32bit的整数，Converts the lower single-precision, floating-point value //of a to a 32-bit integer according to the current rounding mode, r=(int)_A0 extern int _mm_cvt_ss2si(__m128 _A);//=_mm_cvtss_si32 //返回一个__m64寄存器，Converts the two lower single-precision, floating-point  //values of a to two 32-bit integers according to the current rounding mode,  //returning the integers in packed form, r0=(int)_A0, r1=(int)_A1 extern __m64 _mm_cvt_ps2pi(__m128 _A);//=_mm_cvtps_pi32 //返回一个32bit的整数，Converts the lower single-precision, floating-point value //of a to a 32-bit integer with truncation, r=(int)_A0 extern int _mm_cvtt_ss2si(__m128 _A);//=_mm_cvttss_si32 //返回一个__m64寄存器，Converts the two lower single-precision, floating-point  //values of a to two 32-bit integer with truncation, returning the integers  //in packed form, r0=(int)_A0, r1=(int)_A1 extern __m64 _mm_cvtt_ps2pi(__m128 _A);//=_mm_cvttps_pi32 //返回一个__m128的寄存器，Converts the 32-bit integer value b to an single-precision, //floating-point value; the upper three single-precision, floating-point values are //passed through from a, r0=(float)_B, r1=_A1, r2=_A2, r3=_A3 extern __m128 _mm_cvt_si2ss(__m128 _A, int _B);//=_mm_cvtsi32_ss  //返回一个__m128的寄存器，Converts the two 32-bit integer values in packed form in b //to two single-precision, floating-point values; the upper two single-precision,  //floating-point values are passed through from a //r0=(float)_B0, r1=(float)_B1, r2=_A2, r3=_A3 extern __m128 _mm_cvt_pi2ps(__m128 _A, __m64 _B);//=_mm_cvtpi32_ps //返回一个__m128的寄存器，Converts the four 16-bit signed integer values in a to  //four single-precision, floating-point values //r0=(float)_A0, r1=(float)_A1, r2=(float)_A2, r3=(float)_A3 __inline __m128 _mm_cvtpi16_ps(__m64 _A); //返回一个__m128的寄存器，Converts the four 16-bit unsigned integer values in a //to four single-precision, floating-point values //r0=(float)_A0, r1=(float)_A1, r2=(float)_A2, r3=(float)_A3 __inline __m128 _mm_cvtpu16_ps(__m64 _A); //返回一个__m64的寄存器，Converts the four single-precision, floating-point values //in a to four signed 16-bit integer values //r0=(short)_A0, r1=(short)_A1, r2=(short)_A2, r3=(short)_A3 __inline __m64 _mm_cvtps_pi16(__m128 _A); //返回一个__m128的寄存器，Converts the lower four 8-bit signed integer values in a  //to four single-precision, floating-point values //r0=(float)_A0, r1=(float)_A1, r2=(float)_A2, r3=(float)_A3 __inline __m128 _mm_cvtpi8_ps(__m64 _A); //返回一个__m128的寄存器，Converts the lower four 8-bit unsigned integer values in a //to four single-precision, floating-point values //r0=(float)_A0, r1=(float)_A1, r2=(float)_A2, r3=(float)_A3 __inline __m128 _mm_cvtpu8_ps(__m64 _A); //返回一个__m64的寄存器，Converts the four single-precision, floating-point values  //in a to the lower four signed 8-bit integer values of the result //r0=(char)_A0, r1=(char)_A1, r2=(char)_A2, r3=(char)_A3 __inline __m64 _mm_cvtps_pi8(__m128 _A); //返回一个__m128的寄存器，Converts the two 32-bit signed integer values in a and the //two 32-bit signed integer values in b to four single-precision, floating-point values //r0=(float)_A0, r1=(float)_A1, r2=(float)_B0, r3=(float)_B1 __inline __m128 _mm_cvtpi32x2_ps(__m64 _A, __m64 _B); //返回一个32bit浮点数，Extracts the lower order floating point value from the parameter //r=_A0 extern float _mm_cvtss_f32(__m128 _A); //Miscellaneous Instructions That Use Streaming SIMD Extensions: //返回一个__m128的寄存器，Selects four specific single-precision, floating-point  //values from a and b, based on the mask i extern __m128 _mm_shuffle_ps(__m128 _A, __m128 _B, unsigned int _Imm8); //返回一个__m128的寄存器，Selects and interleaves the upper two single-precision, //floating-point values from a and b //r0=_A2, r1=_B2, r2=_A3, r3=_B3 extern __m128 _mm_unpackhi_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Selects and interleaves the lower two single-precision, //floating-point values from a and b //r0=_A0, r1=_B0, r2=_A1, r3=_B1 extern __m128 _mm_unpacklo_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Sets the upper two single-precision, floating-point  //values with 64 bits of data loaded from the address p; the lower two values //are passed through from a //r0=_A0, r1=_A1, r2=*_P0, r3=*_P1 extern __m128 _mm_loadh_pi(__m128 _A, __m64 const* _P); //返回一个__m128的寄存器，Moves the upper two single-precision, floating-point //values of b to the lower two single-precision, floating-point values of the result //r3=_A3, r2=_A2, r1=_B3, r0=_B2 extern __m128 _mm_movehl_ps(__m128 _A, __m128 _B); //返回一个__m128的寄存器，Moves the lower two single-precision, floating-point  //values of b to the upper two single-precision, floating-point values of the result //r3=_B1, r2=_B0, r1=_A1, r0=_A0 extern __m128 _mm_movelh_ps(__m128 _A, __m128 _B); //返回为空，Stores the upper two single-precision, floating-point values of a  //to the address p, *_P0=_A2, *_P1=_A3 extern void _mm_storeh_pi(__m64 *_P, __m128 _A); //返回一个__m128的寄存器，Sets the lower two single-precision, floating-point //values with 64 bits of data loaded from the address p; the upper two values //are passed through from a //r0=*_P0, r1=*_P1, r2=_A2, r3=_A3  extern __m128 _mm_loadl_pi(__m128 _A, __m64 const* _P); //返回为空，Stores the lower two single-precision, floating-point values of a //to the address p, *_P0=_A0, *_P1=_A1 extern void _mm_storel_pi(__m64 *_P, __m128 _A); //返回一个整数，Creates a 4-bit mask from the most significant bits of the //four single-precision, floating-point values  //r=sign(_A3)<<3 | sign(_A2)<<2 | sign(_A1)<<1 | sign(_A0) extern int _mm_movemask_ps(__m128 _A); //返回一个无符号整数，Returns the contents of the control register extern unsigned int _mm_getcsr(void); //返回为空，Sets the control register to the value specified extern void _mm_setcsr(unsigned int); //Memory and Initialization Using Streaming SIMD Extensions //Load Operations(SSE) //返回一个__m128的寄存器，Loads an single-precision, floating-point value into //the low word and clears the upper three words //r0=*_P, r1=0.0, r2=0.0, r3=0.0 extern __m128 _mm_load_ss(float const* _P); //返回一个__m128的寄存器，Loads a single single-precision, floating-point value, //copying it into all four words //r0=*_P0, r1=*_P1, r2=*_P2, r3=*_P3 extern __m128 _mm_load_ps1(float const* _P);//=_mm_load1_ps //返回一个__m128的寄存器，Loads four single-precision, floating-point values //The address must be 16-byte aligned //r0=_P[0], r1=_P[1], r2=_P[2], r3=_P[3] extern __m128 _mm_load_ps(float const* _P); //返回一个__m128的寄存器，Loads four single-precision, floating-point values  //in reverse order, The address must be 16-byte aligned //r0=_P[3], r1=_P[2], r2=_P[1], r3=_P[0] extern __m128 _mm_loadr_ps(float const* _P); //返回一个__m128的寄存器，Loads four single-precision, floating-point values //The address does not need to be 16-byte aligned //r0=_P[0], r1=_P[1], r2=_P[2], r3=_P[3] extern __m128 _mm_loadu_ps(float const* _P); //Set Operations(SSE) //返回一个__m128的寄存器，Sets the low word of an single-precision,  //floating-point value to w and clears the upper three words //r0=_W, r1=r2=r3=0.0 extern __m128 _mm_set_ss(float _W); //返回一个__m128的寄存器，Sets the four single-precision, floating-point values to w //r0=r1=r2=r3=_W extern __m128 _mm_set_ps1(float _W);//=_mm_set1_ps //返回一个__m128的寄存器，Sets the four single-precision, floating-point values to  //the four inputs, r0=_D, r1=_C, r2=_B, r3=_A extern __m128 _mm_set_ps(float _A, float _B, float _C, float _D); //返回一个__m128的寄存器，Sets the four single-precision, floating-point values to //the four inputs in reverse order, r0=_A, r1=_B, r2=_C, r3=_D extern __m128 _mm_setr_ps(float _A, float _B, float _C, float _D); //返回一个__m128的寄存器，Clears the four single-precision, floating-point values //r0=r1=r2=r3=0.0 extern __m128 _mm_setzero_ps(void); //Store Operations(SSE) //返回为空，Stores the lower single-precision, floating-point value，*_V=_A0 extern void _mm_store_ss(float *_V, __m128 _A); //返回为空，Stores the lower single-precision, floating-point value across four words //_V[0]=_A0, _V[1]=_A0, _V[2]=_A0, _V[3]=_A0 extern void _mm_store_ps1(float *_V, __m128 _A);//=_mm_store1_ps //返回为空，Stores four single-precision, floating-point values //The address must be 16-byte aligned //_V[0]=_A0, _V[1]=_A1, _V[2]=_A2, _V[3]=_A3 extern void _mm_store_ps(float *_V, __m128 _A); //返回为空，Stores four single-precision, floating-point values in reverse order //The address must be 16-byte aligned, //_V[0]=_A3, _V[1]=_A2, _V[2]=_A1, _V[3]=_A0 extern void _mm_storer_ps(float *_V, __m128 _A); //返回为空，Stores four single-precision, floating-point values, //The address does not need to be 16-byte aligned //_V[0]=_A0, _V[1]=_A1, _V[2]=_A2, _V[3]=_A3 extern void _mm_storeu_ps(float *_V, __m128 _A); //返回一个__m128的寄存器，Sets the low word to the single-precision, floating-point //value of b,The upper 3 single-precision, floating-point values are passed through  //from a, r0=_B0, r1=_A1, r2=_A2, r3=_A3   extern __m128 _mm_move_ss(__m128 _A, __m128 _B); //Integer Intrinsics Using Streaming SIMD Extensions //返回一个16bit整数，Extracts one of the four words of a， //The selector n must be an immediate, //r=(_Imm == 0) ? _A0 : ((_Imm==1) ? _A1 : ((_Imm==2) ? _A2 : _A3)) extern int _m_pextrw(__m64 _A, int _Imm);//=_mm_extract_pi16 //返回一个__m64的寄存器,Inserts word d into one of four words of a, //The selector n must be an immediate //r0=(_Imm==0)? _D : _A0, r1=(_Imm==1)? _D : _A1, //r2=(_Imm==2)? _D : _A2, r3=(_Imm==3)? _D : _A3 extern __m64 _m_pinsrw(__m64 _A, int _D, int _Imm);//=_mm_insert_pi16 //返回一个__m64的寄存器,Computes the element-wise maximum of the words in a and b, //r0=max(_A0, _B0), r1=max(_A1, _B1), r2=max(_A2, _B2), r3=max(_A3, _B3) extern __m64 _m_pmaxsw(__m64 _A, __m64 _B);//=_mm_max_pi16 //返回一个__m64的寄存器,Computes the element-wise maximum of the unsigned bytes in //a and b, r0=max(_A0, _B0), r1=max(_A1, _B1), ... r7=max(_A7, _B7) extern __m64 _m_pmaxub(__m64 _A, __m64 _B);//=_mm_max_pu8 //返回一个__m64的寄存器,Computes the element-wise minimum of the words in a and b //r0=min(_A0, _B0), r1=min(_A1, _B1), r2=min(_A2, _B2), r3=min(_A3, _B3) extern __m64 _m_pminsw(__m64 _A, __m64 _B);//=_mm_min_pi16 //返回一个__m64的寄存器,Computes the element-wise minimum of the unsigned bytes //in a and b, r0=min(_A0, _B0), r1=min(_A1, _B1), ... r7=min(_A7, _B7) extern __m64 _m_pminub(__m64 _A, __m64 _B);//=_mm_min_pu8 //返回一个整数，Creates an 8-bit mask from the most significant bits of the //bytes in a, r=sign(_A7)<<7 | sign(_A6)<<6 | ... | sign(_A0) extern int _m_pmovmskb(__m64 _A);//=_mm_movemask_pi8 //返回一个__m64的寄存器,Multiplies the unsigned words in a and b, returning the //upper 16 bits of the 32-bit intermediate results, //r0=hiword(_A0, _B0), r1=hiword(_A1, _B1), r2=hiword(_A2, _B2), r3=hiword(_A3, _B3) extern __m64 _m_pmulhuw(__m64 _A, __m64 _B);//=_mm_mulhi_pu16 //返回为空，Conditionally stores byte elements of d to address p,The high bit of  //each byte in the selector _B determines whether the corresponding byte in _A  //will be stored, if (sign(_B0)) _P[0]=_A0, if (sign(_B1)) _P[1]=_A1, ... //if (sign(_B7)) _P[7]=_A7 extern void _m_maskmovq(__m64 _A, __m64 _B, char * _P);//=_mm_maskmove_si64 //返回一个__m64的寄存器,Computes the (rounded) averages of the unsigned bytes  //in a and b, t=(unsigned short)_A0 + (unsigned short)_B0, r0=(t>>1) | (t & 0x01), //..., t=(unsigned short)_A7 + (unsigned short)_B7, r7=(t>>1) | (t & 0x01)  extern __m64 _m_pavgb(__m64 _A, __m64 _B);//=_mm_avg_pu8 //返回一个__m64的寄存器,Computes the (rounded) averages of the unsigned words //in a and b, t=(unsigned short)_A0 + (unsigned short)_B0, r0=(t>>1) | (t & 0x01), //..., t=(unsigned short)_A4 + (unsigned short)_B4, r7=(t>>1) | (t & 0x01) extern __m64 _m_pavgw(__m64 _A, __m64 _B);//=_mm_avg_pu16 //返回一个__m64的寄存器,Computes the sum of the absolute differences of the unsigned //bytes in a and b, returning the value in the lower word //The upper three words are cleared //r0=abs(_A0-_B0) + ... + abs(_A7-_B7), r1=r2=r3=0 extern __m64 _m_psadbw(__m64, __m64);//=_mm_sad_pu8 //返回一个__m64的寄存器,Returns a combination of the four words of a. //The selector _Imm must be an immediate //r0=word(_Imm & 0x03) of _A, r1=word((_Imm>>2) & 0x03) of _A,  //r2=word((_Imm>>4) & 0x03) of _A, r1=word((_Imm>>6) & 0x03) of _A,  extern __m64 _m_pshufw(__m64 _A, int _Imm);//=_mm_shuffle_pi16 //Streaming SIMD Extensions that Support the Cache //返回为空，Loads one cache line of data from address p to a location closer //to the processor, The value _Sel specifies the type of prefetch operation extern void _mm_prefetch(char const*_A, int _Sel); //返回为空，Stores the data in a to the address p without polluting the caches //This intrinsic requires you to empty the multimedia state for the MMX register extern void _mm_stream_pi(__m64 * _P, __m64 _A); //返回为空，Stores the data in a to the address p without polluting the caches, //The address must be 16-byte aligned extern void _mm_stream_ps(float *, __m128 _A); //返回为空，Guarantees that every preceding store is globally visible  //before any subsequent store extern void _mm_sfence(void); /* Alternate intrinsic names definition */ #define _mm_cvtss_si32    _mm_cvt_ss2si #define _mm_cvtps_pi32    _mm_cvt_ps2pi #define _mm_cvttss_si32   _mm_cvtt_ss2si #define _mm_cvttps_pi32   _mm_cvtt_ps2pi #define _mm_cvtsi32_ss    _mm_cvt_si2ss #define _mm_cvtpi32_ps    _mm_cvt_pi2ps #define _mm_extract_pi16  _m_pextrw #define _mm_insert_pi16   _m_pinsrw #define _mm_max_pi16      _m_pmaxsw #define _mm_max_pu8       _m_pmaxub #define _mm_min_pi16      _m_pminsw #define _mm_min_pu8       _m_pminub #define _mm_movemask_pi8  _m_pmovmskb #define _mm_mulhi_pu16    _m_pmulhuw #define _mm_shuffle_pi16  _m_pshufw #define _mm_maskmove_si64 _m_maskmovq #define _mm_avg_pu8       _m_pavgb #define _mm_avg_pu16      _m_pavgw #define _mm_sad_pu8       _m_psadbw #define _mm_set1_ps       _mm_set_ps1 #define _mm_load1_ps      _mm_load_ps1 #define _mm_store1_ps     _mm_store_ps1