fvec.h File Reference

#include <xmmintrin.h>
#include <ivec.h>
#include <crtdefs.h>

Go to the source code of this file.

Classes
class	F32vec4
class	F32vec1

Macros
#define	_VEC_ASSERT(_Expression)   (void)( (!!(_Expression)) || (_wassert(_CRT_WIDE(#_Expression), _CRT_WIDE(__FILE__), __LINE__), 0) )
#define	_f32vec4_abs_mask   ((F32vec4)__f32vec4_abs_mask_cheat.m)
#define	Fvec32s4_COMP(op)   friend F32vec4 cmp##op (const F32vec4 &a, const F32vec4 &b) { return _mm_cmp##op##_ps(a,b); }
#define	Fvec32s4_SELECT(op)
#define	Fvec32s1_COMP(op)   friend F32vec1 cmp##op (const F32vec1 &a, const F32vec1 &b) { return _mm_cmp##op##_ss(a,b); }
#define	Fvec32s1_SELECT(op)

Functions
_CRTIMP void __cdecl	_wassert (_In_z_ const wchar_t *_Message, _In_z_ const wchar_t *_File, _In_ unsigned _Line)
F32vec4	unpack_low (const F32vec4 &a, const F32vec4 &b)
F32vec4	unpack_high (const F32vec4 &a, const F32vec4 &b)
int	move_mask (const F32vec4 &a)
void	loadu (F32vec4 &a, float *p)
void	storeu (float *p, const F32vec4 &a)
void	store_nta (float *p, const F32vec4 &a)
	Fvec32s4_SELECT (eq) Fvec32s4_SELECT(lt) Fvec32s4_SELECT(le) Fvec32s4_SELECT(gt) Fvec32s4_SELECT(ge) Fvec32s4_SELECT(neq) Fvec32s4_SELECT(nlt) Fvec32s4_SELECT(nle) Fvec32s4_SELECT(ngt) Fvec32s4_SELECT(nge) inline Is16vec4 simd_max(const Is16vec4 &a
const Is16vec4 &b return	_m_pmaxsw (a, b)
Is16vec4	simd_min (const Is16vec4 &a, const Is16vec4 &b)
Iu8vec8	simd_max (const Iu8vec8 &a, const Iu8vec8 &b)
Iu8vec8	simd_min (const Iu8vec8 &a, const Iu8vec8 &b)
Iu16vec4	simd_avg (const Iu16vec4 &a, const Iu16vec4 &b)
Iu8vec8	simd_avg (const Iu8vec8 &a, const Iu8vec8 &b)
int	move_mask (const I8vec8 &a)
Iu16vec4	mul_high (const Iu16vec4 &a, const Iu16vec4 &b)
void	mask_move (const I8vec8 &a, const I8vec8 &b, char *addr)
void	store_nta (__m64 *p, const M64 &a)
int	F32vec4ToInt (const F32vec4 &a)
Is32vec2	F32vec4ToIs32vec2 (const F32vec4 &a)
F32vec4	IntToF32vec4 (const F32vec4 &a, int i)
F32vec4	Is32vec2ToF32vec4 (const F32vec4 &a, const Is32vec2 &b)
	Fvec32s1_SELECT (eq) Fvec32s1_SELECT(lt) Fvec32s1_SELECT(le) Fvec32s1_SELECT(gt) Fvec32s1_SELECT(ge) Fvec32s1_SELECT(neq) Fvec32s1_SELECT(nlt) Fvec32s1_SELECT(nle) Fvec32s1_SELECT(ngt) Fvec32s1_SELECT(nge) inline int F32vec1ToInt(const F32vec1 &a)

Variables
union {
int   i [4]
__m128   m
}	__f32vec4_abs_mask_cheat = {0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff}

Macro Definition Documentation

#define _f32vec4_abs_mask   ((F32vec4)__f32vec4_abs_mask_cheat.m)

#define _VEC_ASSERT

(

_Expression

)

(void)( (!!(_Expression)) || (_wassert(_CRT_WIDE(#_Expression), _CRT_WIDE(__FILE__), __LINE__), 0) )

#define Fvec32s1_COMP

(

op

)

friend F32vec1 cmp##op (const F32vec1 &a, const F32vec1 &b) { return _mm_cmp##op##_ss(a,b); }

#define Fvec32s1_SELECT

(

op

)

Value:

inline F32vec1 select_##op (const F32vec1 &a, const F32vec1 &b, const F32vec1 &c, const F32vec1 &d)        \
{                                                      \
    F32vec1 mask = _mm_cmp##op##_ss(a,b);                                          \
    return( (mask & c) | F32vec1((_mm_andnot_ps(mask,d))));                                            \
}

#define Fvec32s4_COMP

(

op

)

friend F32vec4 cmp##op (const F32vec4 &a, const F32vec4 &b) { return _mm_cmp##op##_ps(a,b); }

#define Fvec32s4_SELECT

(

op

)

Value:

inline F32vec4 select_##op (const F32vec4 &a, const F32vec4 &b, const F32vec4 &c, const F32vec4 &d)        \
{                                                               \
    F32vec4 mask = _mm_cmp##op##_ps(a,b);                       \
    return( (mask & c) | F32vec4((_mm_andnot_ps(mask,d)))); \
}

Function Documentation

const Is16vec4& b return _m_pmaxsw	(	a	,
		b
	)

_CRTIMP void __cdecl _wassert	(	_In_z_ const wchar_t *	_Message,
		_In_z_ const wchar_t *	_File,
		_In_ unsigned	_Line
	)

int F32vec4ToInt ( const F32vec4 &  a )

inline

{

    return _mm_cvtt_ss2si(a);

}

Is32vec2 F32vec4ToIs32vec2 ( const F32vec4 &  a )

inline

{

    __m64 result;
    result = _mm_cvtt_ps2pi(a);
    return Is32vec2(result);

}

Fvec32s1_SELECT

(

eq

)

const

{
    return _mm_cvtt_ss2si(a);
}

Fvec32s4_SELECT

(

eq

)

const

F32vec4 IntToF32vec4 ( const F32vec4 &  a, int  i  )

inline

{

    __m128 result;
    result = _mm_cvt_si2ss(a,i);
    return F32vec4(result);

}

F32vec4 Is32vec2ToF32vec4 ( const F32vec4 &  a, const Is32vec2 &  b  )

inline

{

    __m128 result;
    result = _mm_cvt_pi2ps(a,b);
    return F32vec4(result);
}

void loadu ( F32vec4 &  a, float *  p  )

inline

{ a = _mm_loadu_ps(p); }

void mask_move ( const I8vec8 &  a, const I8vec8 &  b, char *  addr  )

inline

{ _m_maskmovq(a, b, addr); }

int move_mask ( const F32vec4 &  a )

inline

{ return _mm_movemask_ps(a);}

int move_mask ( const I8vec8 &  a )

inline

{ return _m_pmovmskb(a);}

Iu16vec4 mul_high ( const Iu16vec4 &  a, const Iu16vec4 &  b  )

inline

{ return _m_pmulhuw(a,b); }

Iu16vec4 simd_avg ( const Iu16vec4 &  a, const Iu16vec4 &  b  )

inline

{ return _mm_avg_pu16(a,b); }

Iu8vec8 simd_avg ( const Iu8vec8 &  a, const Iu8vec8 &  b  )

inline

{ return _mm_avg_pu8(a,b); }

Iu8vec8 simd_max ( const Iu8vec8 &  a, const Iu8vec8 &  b  )

inline

{ return _m_pmaxub(a,b);}

Is16vec4 simd_min ( const Is16vec4 &  a, const Is16vec4 &  b  )

inline

{ return _m_pminsw(a,b);}

Iu8vec8 simd_min ( const Iu8vec8 &  a, const Iu8vec8 &  b  )

inline

{ return _m_pminub(a,b);}

void store_nta ( float *  p, const F32vec4 &  a  )

inline

{ _mm_stream_ps(p,a);}

void store_nta ( __m64 *  p, const M64 &  a  )

inline

{ _mm_stream_pi(p,a); }

void storeu ( float *  p, const F32vec4 &  a  )

inline

{ _mm_storeu_ps(p, a); }

F32vec4 unpack_high ( const F32vec4 &  a, const F32vec4 &  b  )

inline

{ return _mm_unpackhi_ps(a, b); }

F32vec4 unpack_low ( const F32vec4 &  a, const F32vec4 &  b  )

inline

{ return _mm_unpacklo_ps(a, b); }

Variable Documentation

const { ... } __f32vec4_abs_mask_cheat

int i[4]

__m128 m