added SSE version of vectormath

This commit is contained in:
ejcoumans
2007-07-24 23:18:11 +00:00
parent f998250c58
commit 744f287935
8 changed files with 7258 additions and 0 deletions

View File

@@ -0,0 +1,247 @@
/*
Copyright (C) 2006, 2007 Sony Computer Entertainment Inc.
All rights reserved.
Redistribution and use in source and binary forms,
with or without modification, are permitted provided that the
following conditions are met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the Sony Computer Entertainment Inc nor the names
of its contributors may be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef _BOOLINVEC_H
#define _BOOLINVEC_H
#include <math.h>
namespace Vectormath {
class floatInVec;
//--------------------------------------------------------------------------------------------------
// boolInVec class
//
class boolInVec
{
private:
__m128 mData;
inline boolInVec(__m128 vec);
public:
inline boolInVec() {}
// matches standard type conversions
//
inline boolInVec(const floatInVec &vec);
// explicit cast from bool
//
explicit inline boolInVec(bool scalar);
#ifdef _VECTORMATH_NO_SCALAR_CAST
// explicit cast to bool
//
inline bool getAsBool() const;
#else
// implicit cast to bool
//
inline operator bool() const;
#endif
// get vector data
// bool value is splatted across all word slots of vector as 0 (false) or -1 (true)
//
inline __m128 get128() const;
// operators
//
inline const boolInVec operator ! () const;
inline boolInVec& operator = (const boolInVec &vec);
inline boolInVec& operator &= (const boolInVec &vec);
inline boolInVec& operator ^= (const boolInVec &vec);
inline boolInVec& operator |= (const boolInVec &vec);
// friend functions
//
friend inline const boolInVec operator == (const boolInVec &vec0, const boolInVec &vec1);
friend inline const boolInVec operator != (const boolInVec &vec0, const boolInVec &vec1);
friend inline const boolInVec operator < (const floatInVec &vec0, const floatInVec &vec1);
friend inline const boolInVec operator <= (const floatInVec &vec0, const floatInVec &vec1);
friend inline const boolInVec operator > (const floatInVec &vec0, const floatInVec &vec1);
friend inline const boolInVec operator >= (const floatInVec &vec0, const floatInVec &vec1);
friend inline const boolInVec operator == (const floatInVec &vec0, const floatInVec &vec1);
friend inline const boolInVec operator != (const floatInVec &vec0, const floatInVec &vec1);
friend inline const boolInVec operator & (const boolInVec &vec0, const boolInVec &vec1);
friend inline const boolInVec operator ^ (const boolInVec &vec0, const boolInVec &vec1);
friend inline const boolInVec operator | (const boolInVec &vec0, const boolInVec &vec1);
friend inline const boolInVec select(const boolInVec &vec0, const boolInVec &vec1, const boolInVec &select_vec1);
};
//--------------------------------------------------------------------------------------------------
// boolInVec functions
//
// operators
//
inline const boolInVec operator == (const boolInVec &vec0, const boolInVec &vec1);
inline const boolInVec operator != (const boolInVec &vec0, const boolInVec &vec1);
inline const boolInVec operator & (const boolInVec &vec0, const boolInVec &vec1);
inline const boolInVec operator ^ (const boolInVec &vec0, const boolInVec &vec1);
inline const boolInVec operator | (const boolInVec &vec0, const boolInVec &vec1);
// select between vec0 and vec1 using boolInVec.
// false selects vec0, true selects vec1
//
inline const boolInVec select(const boolInVec &vec0, const boolInVec &vec1, const boolInVec &select_vec1);
} // namespace Vectormath
//--------------------------------------------------------------------------------------------------
// boolInVec implementation
//
#include "floatInVec.h"
namespace Vectormath {
inline
boolInVec::boolInVec(__m128 vec)
{
mData = vec;
}
inline
boolInVec::boolInVec(const floatInVec &vec)
{
*this = (vec != floatInVec(0.0f));
}
inline
boolInVec::boolInVec(bool scalar)
{
unsigned int mask = -(int)scalar;
mData = _mm_set1_ps(*(float *)&mask); // TODO: Union
}
#ifdef _VECTORMATH_NO_SCALAR_CAST
inline
bool
boolInVec::getAsBool() const
#else
inline
boolInVec::operator bool() const
#endif
{
return *(bool *)&mData;
}
inline
__m128
boolInVec::get128() const
{
return mData;
}
inline
const boolInVec
boolInVec::operator ! () const
{
return boolInVec(_mm_andnot_ps(mData, _mm_cmpneq_ps(_mm_setzero_ps(),_mm_setzero_ps())));
}
inline
boolInVec&
boolInVec::operator = (const boolInVec &vec)
{
mData = vec.mData;
return *this;
}
inline
boolInVec&
boolInVec::operator &= (const boolInVec &vec)
{
*this = *this & vec;
return *this;
}
inline
boolInVec&
boolInVec::operator ^= (const boolInVec &vec)
{
*this = *this ^ vec;
return *this;
}
inline
boolInVec&
boolInVec::operator |= (const boolInVec &vec)
{
*this = *this | vec;
return *this;
}
inline
const boolInVec
operator == (const boolInVec &vec0, const boolInVec &vec1)
{
return boolInVec(_mm_cmpeq_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
operator != (const boolInVec &vec0, const boolInVec &vec1)
{
return boolInVec(_mm_cmpneq_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
operator & (const boolInVec &vec0, const boolInVec &vec1)
{
return boolInVec(_mm_and_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
operator | (const boolInVec &vec0, const boolInVec &vec1)
{
return boolInVec(_mm_or_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
operator ^ (const boolInVec &vec0, const boolInVec &vec1)
{
return boolInVec(_mm_xor_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
select(const boolInVec &vec0, const boolInVec &vec1, const boolInVec &select_vec1)
{
return boolInVec(vec_sel(vec0.get128(), vec1.get128(), select_vec1.get128()));
}
} // namespace Vectormath
#endif // boolInVec_h

View File

@@ -0,0 +1,340 @@
/*
Copyright (C) 2006, 2007 Sony Computer Entertainment Inc.
All rights reserved.
Redistribution and use in source and binary forms,
with or without modification, are permitted provided that the
following conditions are met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the Sony Computer Entertainment Inc nor the names
of its contributors may be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef _FLOATINVEC_H
#define _FLOATINVEC_H
#include <math.h>
#include <xmmintrin.h>
namespace Vectormath {
class boolInVec;
//--------------------------------------------------------------------------------------------------
// floatInVec class
//
class floatInVec
{
private:
__m128 mData;
public:
inline floatInVec(__m128 vec);
inline floatInVec() {}
// matches standard type conversions
//
inline floatInVec(const boolInVec &vec);
// construct from a slot of __m128
//
inline floatInVec(__m128 vec, int slot);
// explicit cast from float
//
explicit inline floatInVec(float scalar);
#ifdef _VECTORMATH_NO_SCALAR_CAST
// explicit cast to float
//
inline float getAsFloat() const;
#else
// implicit cast to float
//
inline operator float() const;
#endif
// get vector data
// float value is splatted across all word slots of vector
//
inline __m128 get128() const;
// operators
//
inline const floatInVec operator ++ (int);
inline const floatInVec operator -- (int);
inline floatInVec& operator ++ ();
inline floatInVec& operator -- ();
inline const floatInVec operator - () const;
inline floatInVec& operator = (const floatInVec &vec);
inline floatInVec& operator *= (const floatInVec &vec);
inline floatInVec& operator /= (const floatInVec &vec);
inline floatInVec& operator += (const floatInVec &vec);
inline floatInVec& operator -= (const floatInVec &vec);
// friend functions
//
friend inline const floatInVec operator * (const floatInVec &vec0, const floatInVec &vec1);
friend inline const floatInVec operator / (const floatInVec &vec0, const floatInVec &vec1);
friend inline const floatInVec operator + (const floatInVec &vec0, const floatInVec &vec1);
friend inline const floatInVec operator - (const floatInVec &vec0, const floatInVec &vec1);
friend inline const floatInVec select(const floatInVec &vec0, const floatInVec &vec1, boolInVec select_vec1);
};
//--------------------------------------------------------------------------------------------------
// floatInVec functions
//
// operators
//
inline const floatInVec operator * (const floatInVec &vec0, const floatInVec &vec1);
inline const floatInVec operator / (const floatInVec &vec0, const floatInVec &vec1);
inline const floatInVec operator + (const floatInVec &vec0, const floatInVec &vec1);
inline const floatInVec operator - (const floatInVec &vec0, const floatInVec &vec1);
inline const boolInVec operator < (const floatInVec &vec0, const floatInVec &vec1);
inline const boolInVec operator <= (const floatInVec &vec0, const floatInVec &vec1);
inline const boolInVec operator > (const floatInVec &vec0, const floatInVec &vec1);
inline const boolInVec operator >= (const floatInVec &vec0, const floatInVec &vec1);
inline const boolInVec operator == (const floatInVec &vec0, const floatInVec &vec1);
inline const boolInVec operator != (const floatInVec &vec0, const floatInVec &vec1);
// select between vec0 and vec1 using boolInVec.
// false selects vec0, true selects vec1
//
inline const floatInVec select(const floatInVec &vec0, const floatInVec &vec1, const boolInVec &select_vec1);
} // namespace Vectormath
//--------------------------------------------------------------------------------------------------
// floatInVec implementation
//
#include "boolInVec.h"
namespace Vectormath {
inline
floatInVec::floatInVec(__m128 vec)
{
mData = vec;
}
inline
floatInVec::floatInVec(const boolInVec &vec)
{
mData = vec_sel(_mm_setzero_ps(), _mm_set1_ps(1.0f), vec.get128());
}
inline
floatInVec::floatInVec(__m128 vec, int slot)
{
SSEFloat v;
v.m128 = vec;
mData = _mm_set1_ps(v.f[slot]);
}
inline
floatInVec::floatInVec(float scalar)
{
mData = _mm_set1_ps(scalar);
}
#ifdef _VECTORMATH_NO_SCALAR_CAST
inline
float
floatInVec::getAsFloat() const
#else
inline
floatInVec::operator float() const
#endif
{
return *((float *)&mData);
}
inline
__m128
floatInVec::get128() const
{
return mData;
}
inline
const floatInVec
floatInVec::operator ++ (int)
{
__m128 olddata = mData;
operator ++();
return floatInVec(olddata);
}
inline
const floatInVec
floatInVec::operator -- (int)
{
__m128 olddata = mData;
operator --();
return floatInVec(olddata);
}
inline
floatInVec&
floatInVec::operator ++ ()
{
*this += floatInVec(_mm_set1_ps(1.0f));
return *this;
}
inline
floatInVec&
floatInVec::operator -- ()
{
*this -= floatInVec(_mm_set1_ps(1.0f));
return *this;
}
inline
const floatInVec
floatInVec::operator - () const
{
return floatInVec(_mm_sub_ps(_mm_setzero_ps(), mData));
}
inline
floatInVec&
floatInVec::operator = (const floatInVec &vec)
{
mData = vec.mData;
return *this;
}
inline
floatInVec&
floatInVec::operator *= (const floatInVec &vec)
{
*this = *this * vec;
return *this;
}
inline
floatInVec&
floatInVec::operator /= (const floatInVec &vec)
{
*this = *this / vec;
return *this;
}
inline
floatInVec&
floatInVec::operator += (const floatInVec &vec)
{
*this = *this + vec;
return *this;
}
inline
floatInVec&
floatInVec::operator -= (const floatInVec &vec)
{
*this = *this - vec;
return *this;
}
inline
const floatInVec
operator * (const floatInVec &vec0, const floatInVec &vec1)
{
return floatInVec(_mm_mul_ps(vec0.get128(), vec1.get128()));
}
inline
const floatInVec
operator / (const floatInVec &num, const floatInVec &den)
{
return floatInVec(_mm_div_ps(num.get128(), den.get128()));
}
inline
const floatInVec
operator + (const floatInVec &vec0, const floatInVec &vec1)
{
return floatInVec(_mm_add_ps(vec0.get128(), vec1.get128()));
}
inline
const floatInVec
operator - (const floatInVec &vec0, const floatInVec &vec1)
{
return floatInVec(_mm_sub_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
operator < (const floatInVec &vec0, const floatInVec &vec1)
{
return boolInVec(_mm_cmpgt_ps(vec1.get128(), vec0.get128()));
}
inline
const boolInVec
operator <= (const floatInVec &vec0, const floatInVec &vec1)
{
return boolInVec(_mm_cmpge_ps(vec1.get128(), vec0.get128()));
}
inline
const boolInVec
operator > (const floatInVec &vec0, const floatInVec &vec1)
{
return boolInVec(_mm_cmpgt_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
operator >= (const floatInVec &vec0, const floatInVec &vec1)
{
return boolInVec(_mm_cmpge_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
operator == (const floatInVec &vec0, const floatInVec &vec1)
{
return boolInVec(_mm_cmpeq_ps(vec0.get128(), vec1.get128()));
}
inline
const boolInVec
operator != (const floatInVec &vec0, const floatInVec &vec1)
{
return boolInVec(_mm_cmpneq_ps(vec0.get128(), vec1.get128()));
}
inline
const floatInVec
select(const floatInVec &vec0, const floatInVec &vec1, const boolInVec &select_vec1)
{
return floatInVec(vec_sel(vec0.get128(), vec1.get128(), select_vec1.get128()));
}
} // namespace Vectormath
#endif // floatInVec_h

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,533 @@
/*
Copyright (C) 2006, 2007 Sony Computer Entertainment Inc.
All rights reserved.
Redistribution and use in source and binary forms,
with or without modification, are permitted provided that the
following conditions are met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the Sony Computer Entertainment Inc nor the names
of its contributors may be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef _VECTORMATH_QUAT_AOS_CPP_H
#define _VECTORMATH_QUAT_AOS_CPP_H
//-----------------------------------------------------------------------------
// Definitions
#ifndef _VECTORMATH_INTERNAL_FUNCTIONS
#define _VECTORMATH_INTERNAL_FUNCTIONS
#endif
namespace Vectormath {
namespace Aos {
inline Quat::Quat( float _x, float _y, float _z, float _w )
{
mVec128 = _mm_setr_ps(_x, _y, _z, _w);
}
inline Quat::Quat( const floatInVec &_x, const floatInVec &_y, const floatInVec &_z, const floatInVec &_w )
{
mVec128 = _mm_unpacklo_ps(
_mm_unpacklo_ps( _x.get128(), _z.get128() ),
_mm_unpacklo_ps( _y.get128(), _w.get128() ) );
}
inline Quat::Quat( const Vector3 &xyz, float _w )
{
mVec128 = xyz.get128();
_vmathVfSetElement(mVec128, _w, 3);
}
inline Quat::Quat( const Vector3 &xyz, const floatInVec &_w )
{
mVec128 = xyz.get128();
mVec128 = _vmathVfInsert(mVec128, _w.get128(), 3);
}
inline Quat::Quat( const Vector4 &vec )
{
mVec128 = vec.get128();
}
inline Quat::Quat( float scalar )
{
mVec128 = floatInVec(scalar).get128();
}
inline Quat::Quat( const floatInVec &scalar )
{
mVec128 = scalar.get128();
}
inline Quat::Quat( __m128 vf4 )
{
mVec128 = vf4;
}
inline const Quat Quat::identity( )
{
return Quat( _VECTORMATH_UNIT_0001 );
}
inline const Quat lerp( float t, const Quat &quat0, const Quat &quat1 )
{
return lerp( floatInVec(t), quat0, quat1 );
}
inline const Quat lerp( const floatInVec &t, const Quat &quat0, const Quat &quat1 )
{
return ( quat0 + ( ( quat1 - quat0 ) * t ) );
}
inline const Quat slerp( float t, const Quat &unitQuat0, const Quat &unitQuat1 )
{
return slerp( floatInVec(t), unitQuat0, unitQuat1 );
}
inline const Quat slerp( const floatInVec &t, const Quat &unitQuat0, const Quat &unitQuat1 )
{
Quat start;
vec_float4 scales, scale0, scale1, cosAngle, angle, tttt, oneMinusT, angles, sines;
vec_uint4 selectMask;
cosAngle = _vmathVfDot4( unitQuat0.get128(), unitQuat1.get128() );
selectMask = (vec_uint4)vec_cmpgt( _mm_setzero_ps(), cosAngle );
cosAngle = vec_sel( cosAngle, negatef4( cosAngle ), selectMask );
start = Quat( vec_sel( unitQuat0.get128(), negatef4( unitQuat0.get128() ), selectMask ) );
selectMask = (vec_uint4)vec_cmpgt( _mm_set1_ps(_VECTORMATH_SLERP_TOL), cosAngle );
angle = acosf4( cosAngle );
tttt = t.get128();
oneMinusT = vec_sub( _mm_set1_ps(1.0f), tttt );
angles = vec_mergeh( _mm_set1_ps(1.0f), tttt );
angles = vec_mergeh( angles, oneMinusT );
angles = vec_madd( angles, angle, _mm_setzero_ps() );
sines = sinf4( angles );
scales = _mm_div_ps( sines, vec_splat( sines, 0 ) );
scale0 = vec_sel( oneMinusT, vec_splat( scales, 1 ), selectMask );
scale1 = vec_sel( tttt, vec_splat( scales, 2 ), selectMask );
return Quat( vec_madd( start.get128(), scale0, vec_mul( unitQuat1.get128(), scale1 ) ) );
}
inline const Quat squad( float t, const Quat &unitQuat0, const Quat &unitQuat1, const Quat &unitQuat2, const Quat &unitQuat3 )
{
return squad( floatInVec(t), unitQuat0, unitQuat1, unitQuat2, unitQuat3 );
}
inline const Quat squad( const floatInVec &t, const Quat &unitQuat0, const Quat &unitQuat1, const Quat &unitQuat2, const Quat &unitQuat3 )
{
return slerp( ( ( floatInVec(2.0f) * t ) * ( floatInVec(1.0f) - t ) ), slerp( t, unitQuat0, unitQuat3 ), slerp( t, unitQuat1, unitQuat2 ) );
}
inline __m128 Quat::get128( ) const
{
return mVec128;
}
inline Quat & Quat::operator =( const Quat &quat )
{
mVec128 = quat.mVec128;
return *this;
}
inline Quat & Quat::setXYZ( const Vector3 &vec )
{
__declspec(align(16)) unsigned int sw[4] = {0, 0, 0, 0xffffffff};
mVec128 = vec_sel( vec.get128(), mVec128, sw );
return *this;
}
inline const Vector3 Quat::getXYZ( ) const
{
return Vector3( mVec128 );
}
inline Quat & Quat::setX( float _x )
{
_vmathVfSetElement(mVec128, _x, 0);
return *this;
}
inline Quat & Quat::setX( const floatInVec &_x )
{
mVec128 = _vmathVfInsert(mVec128, _x.get128(), 0);
return *this;
}
inline const floatInVec Quat::getX( ) const
{
return floatInVec( mVec128, 0 );
}
inline Quat & Quat::setY( float _y )
{
_vmathVfSetElement(mVec128, _y, 1);
return *this;
}
inline Quat & Quat::setY( const floatInVec &_y )
{
mVec128 = _vmathVfInsert(mVec128, _y.get128(), 1);
return *this;
}
inline const floatInVec Quat::getY( ) const
{
return floatInVec( mVec128, 1 );
}
inline Quat & Quat::setZ( float _z )
{
_vmathVfSetElement(mVec128, _z, 2);
return *this;
}
inline Quat & Quat::setZ( const floatInVec &_z )
{
mVec128 = _vmathVfInsert(mVec128, _z.get128(), 2);
return *this;
}
inline const floatInVec Quat::getZ( ) const
{
return floatInVec( mVec128, 2 );
}
inline Quat & Quat::setW( float _w )
{
_vmathVfSetElement(mVec128, _w, 3);
return *this;
}
inline Quat & Quat::setW( const floatInVec &_w )
{
mVec128 = _vmathVfInsert(mVec128, _w.get128(), 3);
return *this;
}
inline const floatInVec Quat::getW( ) const
{
return floatInVec( mVec128, 3 );
}
inline Quat & Quat::setElem( int idx, float value )
{
_vmathVfSetElement(mVec128, value, idx);
return *this;
}
inline Quat & Quat::setElem( int idx, const floatInVec &value )
{
mVec128 = _vmathVfInsert(mVec128, value.get128(), idx);
return *this;
}
inline const floatInVec Quat::getElem( int idx ) const
{
return floatInVec( mVec128, idx );
}
inline VecIdx Quat::operator []( int idx )
{
return VecIdx( mVec128, idx );
}
inline const floatInVec Quat::operator []( int idx ) const
{
return floatInVec( mVec128, idx );
}
inline const Quat Quat::operator +( const Quat &quat ) const
{
return Quat( _mm_add_ps( mVec128, quat.mVec128 ) );
}
inline const Quat Quat::operator -( const Quat &quat ) const
{
return Quat( _mm_sub_ps( mVec128, quat.mVec128 ) );
}
inline const Quat Quat::operator *( float scalar ) const
{
return *this * floatInVec(scalar);
}
inline const Quat Quat::operator *( const floatInVec &scalar ) const
{
return Quat( _mm_mul_ps( mVec128, scalar.get128() ) );
}
inline Quat & Quat::operator +=( const Quat &quat )
{
*this = *this + quat;
return *this;
}
inline Quat & Quat::operator -=( const Quat &quat )
{
*this = *this - quat;
return *this;
}
inline Quat & Quat::operator *=( float scalar )
{
*this = *this * scalar;
return *this;
}
inline Quat & Quat::operator *=( const floatInVec &scalar )
{
*this = *this * scalar;
return *this;
}
inline const Quat Quat::operator /( float scalar ) const
{
return *this / floatInVec(scalar);
}
inline const Quat Quat::operator /( const floatInVec &scalar ) const
{
return Quat( _mm_div_ps( mVec128, scalar.get128() ) );
}
inline Quat & Quat::operator /=( float scalar )
{
*this = *this / scalar;
return *this;
}
inline Quat & Quat::operator /=( const floatInVec &scalar )
{
*this = *this / scalar;
return *this;
}
inline const Quat Quat::operator -( ) const
{
return Quat(_mm_sub_ps( _mm_setzero_ps(), mVec128 ) );
}
inline const Quat operator *( float scalar, const Quat &quat )
{
return floatInVec(scalar) * quat;
}
inline const Quat operator *( const floatInVec &scalar, const Quat &quat )
{
return quat * scalar;
}
inline const floatInVec dot( const Quat &quat0, const Quat &quat1 )
{
return floatInVec( _vmathVfDot4( quat0.get128(), quat1.get128() ), 0 );
}
inline const floatInVec norm( const Quat &quat )
{
return floatInVec( _vmathVfDot4( quat.get128(), quat.get128() ), 0 );
}
inline const floatInVec length( const Quat &quat )
{
return floatInVec( _mm_sqrt_ps(_vmathVfDot4( quat.get128(), quat.get128() )), 0 );
}
inline const Quat normalize( const Quat &quat )
{
return Quat( _mm_mul_ps( quat.get128(), _mm_rsqrt_ps( _vmathVfDot4( quat.get128(), quat.get128() ) ) ) );
}
inline const Quat Quat::rotation( const Vector3 &unitVec0, const Vector3 &unitVec1 )
{
Vector3 crossVec;
__m128 cosAngle, cosAngleX2Plus2, recipCosHalfAngleX2, cosHalfAngleX2, res;
cosAngle = _vmathVfDot3( unitVec0.get128(), unitVec1.get128() );
cosAngleX2Plus2 = vec_madd( cosAngle, _mm_set1_ps(2.0f), _mm_set1_ps(2.0f) );
recipCosHalfAngleX2 = _mm_rsqrt_ps( cosAngleX2Plus2 );
cosHalfAngleX2 = vec_mul( recipCosHalfAngleX2, cosAngleX2Plus2 );
crossVec = cross( unitVec0, unitVec1 );
res = vec_mul( crossVec.get128(), recipCosHalfAngleX2 );
__declspec(align(16)) unsigned int sw[4] = {0, 0, 0, 0xffffffff};
res = vec_sel( res, vec_mul( cosHalfAngleX2, _mm_set1_ps(0.5f) ), sw );
return Quat( res );
}
inline const Quat Quat::rotation( float radians, const Vector3 &unitVec )
{
return rotation( floatInVec(radians), unitVec );
}
inline const Quat Quat::rotation( const floatInVec &radians, const Vector3 &unitVec )
{
__m128 s, c, angle, res;
angle = vec_mul( radians.get128(), _mm_set1_ps(0.5f) );
sincosf4( angle, &s, &c );
__declspec(align(16)) unsigned int sw[4] = {0, 0, 0, 0xffffffff};
res = vec_sel( vec_mul( unitVec.get128(), s ), c, sw );
return Quat( res );
}
inline const Quat Quat::rotationX( float radians )
{
return rotationX( floatInVec(radians) );
}
inline const Quat Quat::rotationX( const floatInVec &radians )
{
__m128 s, c, angle, res;
angle = vec_mul( radians.get128(), _mm_set1_ps(0.5f) );
sincosf4( angle, &s, &c );
__declspec(align(16)) unsigned int xsw[4] = {0xffffffff, 0, 0, 0};
__declspec(align(16)) unsigned int wsw[4] = {0, 0, 0, 0xffffffff};
res = vec_sel( _mm_setzero_ps(), s, xsw );
res = vec_sel( res, c, wsw );
return Quat( res );
}
inline const Quat Quat::rotationY( float radians )
{
return rotationY( floatInVec(radians) );
}
inline const Quat Quat::rotationY( const floatInVec &radians )
{
__m128 s, c, angle, res;
angle = vec_mul( radians.get128(), _mm_set1_ps(0.5f) );
sincosf4( angle, &s, &c );
__declspec(align(16)) unsigned int ysw[4] = {0, 0xffffffff, 0, 0};
__declspec(align(16)) unsigned int wsw[4] = {0, 0, 0, 0xffffffff};
res = vec_sel( _mm_setzero_ps(), s, ysw );
res = vec_sel( res, c, wsw );
return Quat( res );
}
inline const Quat Quat::rotationZ( float radians )
{
return rotationZ( floatInVec(radians) );
}
inline const Quat Quat::rotationZ( const floatInVec &radians )
{
__m128 s, c, angle, res;
angle = vec_mul( radians.get128(), _mm_set1_ps(0.5f) );
sincosf4( angle, &s, &c );
__declspec(align(16)) unsigned int zsw[4] = {0, 0, 0xffffffff, 0};
__declspec(align(16)) unsigned int wsw[4] = {0, 0, 0, 0xffffffff};
res = vec_sel( _mm_setzero_ps(), s, zsw );
res = vec_sel( res, c, wsw );
return Quat( res );
}
inline const Quat Quat::operator *( const Quat &quat ) const
{
__m128 ldata, rdata, qv, tmp0, tmp1, tmp2, tmp3;
__m128 product, l_wxyz, r_wxyz, xy, qw;
ldata = mVec128;
rdata = quat.mVec128;
tmp0 = _mm_shuffle_ps( ldata, ldata, _MM_SHUFFLE(3,0,2,1) );
tmp1 = _mm_shuffle_ps( rdata, rdata, _MM_SHUFFLE(3,1,0,2) );
tmp2 = _mm_shuffle_ps( ldata, ldata, _MM_SHUFFLE(3,1,0,2) );
tmp3 = _mm_shuffle_ps( rdata, rdata, _MM_SHUFFLE(3,0,2,1) );
qv = vec_mul( vec_splat( ldata, 3 ), rdata );
qv = vec_madd( vec_splat( rdata, 3 ), ldata, qv );
qv = vec_madd( tmp0, tmp1, qv );
qv = vec_nmsub( tmp2, tmp3, qv );
product = vec_mul( ldata, rdata );
l_wxyz = vec_sld( ldata, ldata, 12 );
r_wxyz = vec_sld( rdata, rdata, 12 );
qw = vec_nmsub( l_wxyz, r_wxyz, product );
xy = vec_madd( l_wxyz, r_wxyz, product );
qw = vec_sub( qw, vec_sld( xy, xy, 8 ) );
__declspec(align(16)) unsigned int sw[4] = {0, 0, 0, 0xffffffff};
return Quat( vec_sel( qv, qw, sw ) );
}
inline Quat & Quat::operator *=( const Quat &quat )
{
*this = *this * quat;
return *this;
}
inline const Vector3 rotate( const Quat &quat, const Vector3 &vec )
{ __m128 qdata, vdata, product, tmp0, tmp1, tmp2, tmp3, wwww, qv, qw, res;
qdata = quat.get128();
vdata = vec.get128();
tmp0 = _mm_shuffle_ps( qdata, qdata, _MM_SHUFFLE(3,0,2,1) );
tmp1 = _mm_shuffle_ps( vdata, vdata, _MM_SHUFFLE(3,1,0,2) );
tmp2 = _mm_shuffle_ps( qdata, qdata, _MM_SHUFFLE(3,1,0,2) );
tmp3 = _mm_shuffle_ps( vdata, vdata, _MM_SHUFFLE(3,0,2,1) );
wwww = vec_splat( qdata, 3 );
qv = vec_mul( wwww, vdata );
qv = vec_madd( tmp0, tmp1, qv );
qv = vec_nmsub( tmp2, tmp3, qv );
product = vec_mul( qdata, vdata );
qw = vec_madd( vec_sld( qdata, qdata, 4 ), vec_sld( vdata, vdata, 4 ), product );
qw = vec_add( vec_sld( product, product, 8 ), qw );
tmp1 = _mm_shuffle_ps( qv, qv, _MM_SHUFFLE(3,1,0,2) );
tmp3 = _mm_shuffle_ps( qv, qv, _MM_SHUFFLE(3,0,2,1) );
res = vec_mul( vec_splat( qw, 0 ), qdata );
res = vec_madd( wwww, qv, res );
res = vec_madd( tmp0, tmp1, res );
res = vec_nmsub( tmp2, tmp3, res );
return Vector3( res );
}
inline const Quat conj( const Quat &quat )
{
__declspec(align(16)) unsigned int sw[4] = {0x80000000,0x80000000,0x80000000,0};
return Quat( vec_xor( quat.get128(), _mm_load_ps((float *)sw) ) );
}
inline const Quat select( const Quat &quat0, const Quat &quat1, bool select1 )
{
return select( quat0, quat1, boolInVec(select1) );
}
inline const Quat select( const Quat &quat0, const Quat &quat1, const boolInVec &select1 )
{
return Quat( vec_sel( quat0.get128(), quat1.get128(), select1.get128() ) );
}
#ifdef _VECTORMATH_DEBUG
inline void print( const Quat &quat )
{
union { __m128 v; float s[4]; } tmp;
tmp.v = quat.get128();
printf( "( %f %f %f %f )\n", tmp.s[0], tmp.s[1], tmp.s[2], tmp.s[3] );
}
inline void print( const Quat &quat, const char * name )
{
union { __m128 v; float s[4]; } tmp;
tmp.v = quat.get128();
printf( "%s: ( %f %f %f %f )\n", name, tmp.s[0], tmp.s[1], tmp.s[2], tmp.s[3] );
}
#endif
} // namespace Aos
} // namespace Vectormath
#endif

View File

@@ -0,0 +1 @@
This folder contains an SSE implementation of the c++ aos component of the vectormath library as used by PSSG.

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,79 @@
/*
Copyright (C) 2006, 2007 Sony Computer Entertainment Inc.
All rights reserved.
Redistribution and use in source and binary forms,
with or without modification, are permitted provided that the
following conditions are met:
* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in the
documentation and/or other materials provided with the distribution.
* Neither the name of the Sony Computer Entertainment Inc nor the names
of its contributors may be used to endorse or promote products derived
from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.
*/
#ifndef _VECTORMATH_VECIDX_AOS_H
#define _VECTORMATH_VECIDX_AOS_H
#include "floatInVec.h"
namespace Vectormath {
namespace Aos {
//-----------------------------------------------------------------------------
// VecIdx
// Used in setting elements of Vector3, Vector4, Point3, or Quat with the
// subscripting operator.
//
class VecIdx
{
private:
__m128 &ref __attribute__ ((aligned(16)));
int i __attribute__ ((aligned(16)));
public:
inline VecIdx( __m128& vec, int idx ): ref(vec) { i = idx; }
// implicitly casts to float unless _VECTORMATH_NO_SCALAR_CAST defined
// in which case, implicitly casts to floatInVec, and one must call
// getAsFloat to convert to float.
//
#ifdef _VECTORMATH_NO_SCALAR_CAST
inline operator floatInVec() const;
inline float getAsFloat() const;
#else
inline operator float() const;
#endif
inline float operator =( float scalar );
inline floatInVec operator =( const floatInVec &scalar );
inline floatInVec operator =( const VecIdx& scalar );
inline floatInVec operator *=( float scalar );
inline floatInVec operator *=( const floatInVec &scalar );
inline floatInVec operator /=( float scalar );
inline floatInVec operator /=( const floatInVec &scalar );
inline floatInVec operator +=( float scalar );
inline floatInVec operator +=( const floatInVec &scalar );
inline floatInVec operator -=( float scalar );
inline floatInVec operator -=( const floatInVec &scalar );
};
} // namespace Aos
} // namespace Vectormath
#endif

File diff suppressed because it is too large Load Diff