drewcassidy/nvidia-texture-tools
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Add external libs for comparisons and benchmarks.

Browse Source
This commit is contained in:
Ignacio
2020-03-23 10:07:38 -07:00
parent 4a33d1ac75
commit 9a16bebf8f
67 changed files with 24230 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

153
extern/CMP_Core/source/CMP_Core.h vendored Normal file
View File

@ -0,0 +1,153 @@
//=====================================================================
// Copyright (c) 2019 Advanced Micro Devices, Inc. All rights reserved.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files(the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and / or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions :
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.
//
/// \file CMP_Core.h
//
//=====================================================================
#ifndef CMP_CORE_H
#define CMP_CORE_H
#include <stdint.h>
#ifdef _WIN32
#define CMP_CDECL __cdecl
#else
#define CMP_CDECL
#endif
//====================================================================================
// API Definitions for Core API
//------------------------------------------------------------------------------------
// All API return 0 on success else error codes > 0
// See Common_Def.h CGU_CORE_ values for the error codes
//=====================================================================================
//======================================================================================================
// Block level setting option: Create and Destroy Reference Pointers
//======================================================================================================
// Context create and destroy to use for BCn codec settings, where n is the set [1,2,3,4,5,6,7]
// All codecs will use default max quality settings, users can create multiple contexts to
// set quality levels, masks , channel mapping, etc...
int CMP_CDECL CreateOptionsBC1(void **optionsBC1);
int CMP_CDECL CreateOptionsBC2(void **optionsBC2);
int CMP_CDECL CreateOptionsBC3(void **optionsBC3);
int CMP_CDECL CreateOptionsBC4(void **optionsBC4);
int CMP_CDECL CreateOptionsBC5(void **optionsBC5);
int CMP_CDECL CreateOptionsBC6(void **optionsBC6);
int CMP_CDECL CreateOptionsBC7(void **optionsBC7);
int CMP_CDECL DestroyOptionsBC1(void *optionsBC1);
int CMP_CDECL DestroyOptionsBC2(void *optionsBC2);
int CMP_CDECL DestroyOptionsBC3(void *optionsBC3);
int CMP_CDECL DestroyOptionsBC4(void *optionsBC4);
int CMP_CDECL DestroyOptionsBC5(void *optionsBC5);
int CMP_CDECL DestroyOptionsBC6(void *optionsBC6);
int CMP_CDECL DestroyOptionsBC7(void *optionsBC7);
//======================================================================================================
// Block level settings using the options Reference Pointers
//======================================================================================================
// Setting channel Weights : Applies to BC1, BC2 and BC3 valid ranges are [0..1.0f] Default is {1.0f, 1.0f , 1.0f}
// Use channel weightings. With swizzled formats the weighting applies to the data within the specified channel not the channel itself.
int CMP_CDECL SetChannelWeightsBC1(void *options, float WeightRed, float WeightGreen, float WeightBlue);
int CMP_CDECL SetChannelWeightsBC2(void *options, float WeightRed, float WeightGreen, float WeightBlue);
int CMP_CDECL SetChannelWeightsBC3(void *options, float WeightRed, float WeightGreen, float WeightBlue);
// True sets mapping CMP_Core BC1, BC2 & BC3 to decode Red,Green,Blue and Alpha as
// RGBA to channels [0,1,2,3] else BGRA maps to [0,1,2,3]
// Default is set to true.
int CMP_CDECL SetDecodeChannelMapping(void *options, bool mapRGBA);
int CMP_CDECL SetQualityBC1(void *options, float fquality);
int CMP_CDECL SetQualityBC2(void *options, float fquality);
int CMP_CDECL SetQualityBC3(void *options, float fquality);
int CMP_CDECL SetQualityBC4(void *options, float fquality);
int CMP_CDECL SetQualityBC5(void *options, float fquality);
int CMP_CDECL SetQualityBC6(void *options, float fquality);
int CMP_CDECL SetQualityBC7(void *options, float fquality);
int CMP_CDECL SetAlphaThresholdBC1(void *options, unsigned char alphaThreshold);
int CMP_CDECL SetMaskBC6(void *options, unsigned int mask);
int CMP_CDECL SetMaskBC7(void *options, unsigned char mask);
int CMP_CDECL SetAlphaOptionsBC7(void *options, bool imageNeedsAlpha, bool colourRestrict, bool alphaRestrict);
int CMP_CDECL SetErrorThresholdBC7(void *options, float minThreshold, float maxThreshold);
//======================================================================================================
// (4x4) Block level 4 channel source CompressBlock and DecompressBlock API for BCn Codecs
//======================================================================================================
// The options parameter for these API can be set to null in the calls if defaults settings is sufficient
// Example: CompressBlockBC1(srcBlock,16,cmpBlock,NULL); For "C" call
// CompressBlockBC1(srcBlock,16,cmpBlock); For "C++" calls
//
// To use this parameter first create the options context using the CreateOptions call
// then use the Set Options to set various codec settings and pass them to the appropriate
// Compress or Decompress API.
// The source (srcBlock) channel format is expected to be RGBA:8888 by default for LDR Codecs
// for BC6H the format is RGBA Half float (16 bits per channel)
//------------------------------------------------------------------------------------------------------
#ifdef __cplusplus
#define CMP_DEFAULTNULL =NULL
#else
#define CMP_DEFAULTNULL
#endif
//=========================================================================================================
// 4 channel Sources, default format RGBA:8888 is processed as a 4x4 block starting at srcBlock location
// where each row of the block is calculated from srcStride
//=========================================================================================================
int CMP_CDECL CompressBlockBC1(const unsigned char *srcBlock, unsigned int srcStrideInBytes, unsigned char cmpBlock[8 ], const void *options CMP_DEFAULTNULL);
int CMP_CDECL CompressBlockBC2(const unsigned char *srcBlock, unsigned int srcStrideInBytes, unsigned char cmpBlock[16], const void *options CMP_DEFAULTNULL);
int CMP_CDECL CompressBlockBC3(const unsigned char *srcBlock, unsigned int srcStrideInBytes, unsigned char cmpBlock[16], const void *options CMP_DEFAULTNULL);
int CMP_CDECL CompressBlockBC7(const unsigned char *srcBlock, unsigned int srcStrideInBytes, unsigned char cmpBlock[16], const void *options CMP_DEFAULTNULL);
int CMP_CDECL DecompressBlockBC1(const unsigned char cmpBlock[8 ], unsigned char srcBlock[64], const void *options CMP_DEFAULTNULL);
int CMP_CDECL DecompressBlockBC2(const unsigned char cmpBlock[16], unsigned char srcBlock[64], const void *options CMP_DEFAULTNULL);
int CMP_CDECL DecompressBlockBC3(const unsigned char cmpBlock[16], unsigned char srcBlock[64], const void *options CMP_DEFAULTNULL);
int CMP_CDECL DecompressBlockBC7(const unsigned char cmpBlock[16], unsigned char srcBlock[64], const void *options CMP_DEFAULTNULL);
//================================================
// 1 channel Source 4x4 8 bits per block
//================================================
int CMP_CDECL CompressBlockBC4(const unsigned char *srcBlock, unsigned int srcStrideInBytes, unsigned char cmpBlock[8], const void *options CMP_DEFAULTNULL);
int CMP_CDECL DecompressBlockBC4(const unsigned char cmpBlock[8], unsigned char srcBlock[16], const void *options CMP_DEFAULTNULL);
//================================================
// 2 channel Source 2x(4x4 8 bits)
//================================================
int CMP_CDECL CompressBlockBC5(const unsigned char *srcBlock1, unsigned int srcStrideInBytes1,
const unsigned char *srcBlock2, unsigned int srcStrideInBytes2,
unsigned char cmpBlock[16], const void *options CMP_DEFAULTNULL);
int CMP_CDECL DecompressBlockBC5(const unsigned char cmpBlock[16], unsigned char srcBlock1[16], unsigned char srcBlock2[16], const void *options CMP_DEFAULTNULL);
//========================================================================================
// For 3 channel Source RGB_16, Note srcStride is in unsigned short steps (2 bytes each)
//========================================================================================
int CMP_CDECL CompressBlockBC6(const unsigned short *srcBlock, unsigned int srcStrideInShorts, unsigned char cmpBlock[16], const void *options CMP_DEFAULTNULL);
int CMP_CDECL DecompressBlockBC6(const unsigned char cmpBlock[16], unsigned short srcBlock[48], const void *options CMP_DEFAULTNULL);
#endif // CMP_CORE

417
extern/CMP_Core/source/cmp_math_vec4.h vendored Normal file
View File

@ -0,0 +1,417 @@
//=====================================================================
// Copyright 2019 (c), Advanced Micro Devices, Inc. All rights reserved.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files(the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and / or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions :
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.
//
//=====================================================================
#ifndef CMP_MATH_VEC4_H
#define CMP_MATH_VEC4_H
//====================================================
// Vector Class definitions for CPU & Intrinsics
//====================================================
#if defined (_LINUX) || defined (_WIN32)
//============================================= VEC2 ==================================================
template<class T>
class Vec2
{
public:
T x;
T y;
// *****************************************
// Constructors
// *****************************************
/// Default constructor
Vec2() : x((T)0), y((T)0) {};
/// Value constructor
Vec2(const T& vx, const T& vy) : x(vx), y(vy) {};
/// Copy constructor
Vec2(const Vec2<T>& val) : x(val.x), y(val.y) {};
/// Single value constructor. Sets all components to the given value
Vec2(const T& v) : x(v), y(v) {};
// *****************************************
// Conversions/Assignment/Indexing
// *****************************************
/// cast to T*
operator const T* () const { return (const T*)this; };
/// cast to T*
operator T* () { return (T*)this; };
/// Indexing
const T& operator[](int i) const { return ((const T*)this)[i]; };
T& operator[](int i) { return ((T*)this)[i]; };
/// Assignment
const Vec2<T>& operator=(const Vec2<T>& rhs) { x = rhs.x; y = rhs.y; return *this; };
// *****************************************
// Comparison
// *****************************************
/// Equality comparison
bool operator==(const Vec2<T>& rhs) const { return (x == rhs.x && y == rhs.y); };
/// Inequality comparision
bool operator!=(const Vec2<T>& rhs) const { return (x != rhs.x || y != rhs.y); };
// *****************************************
// Arithmetic
// *****************************************
/// Addition
const Vec2<T> operator+(const Vec2<T>& rhs) const { return Vec2<T>(x + rhs.x, y + rhs.y); };
/// Subtraction
const Vec2<T> operator-(const Vec2<T>& rhs) const { return Vec2<T>(x - rhs.x, y - rhs.y); };
/// Multiply by scalar
const Vec2<T> operator*(const T& v) const { return Vec2<T>(x * v, y * v); };
/// Divide by scalar
const Vec2<T> operator/(const T& v) const { return Vec2<T>(x / v, y / v); };
/// Addition in-place
Vec2<T>& operator+= (const Vec2<T>& rhs) { x += rhs.x; y += rhs.y; return *this; };
/// Subtract in-place
Vec2<T>& operator-= (const Vec2<T>& rhs) { x -= rhs.x; y -= rhs.y; return *this; };
/// Scalar multiply in-place
Vec2<T>& operator*= (const T& v) { x *= v; y *= v; return *this; };
/// Scalar divide in-place
Vec2<T>& operator/= (const T& v) { x /= v; y /= v; return *this; };
};
typedef Vec2<float> CMP_Vec2f;
typedef Vec2<float> CGU_Vec2f;
typedef Vec2<float> CGV_Vec2f;
typedef Vec2<double> CMP_Vec2d;
typedef Vec2<int> CMP_Vec2i;
//}
//============================================= VEC3 ==================================================
template<class T>
class Vec3
{
public:
T x;
T y;
T z;
// *****************************************
// Constructors
// *****************************************
/// Default constructor
Vec3() : x((T)0), y((T)0), z((T)0) {};
/// Value constructor
Vec3(const T& vx, const T& vy, const T& vz) : x(vx), y(vy), z(vz) {};
/// Copy constructor
Vec3(const Vec3<T>& val) : x(val.x), y(val.y), z(val.z) {};
/// Single value constructor. Sets all components to the given value
Vec3(const T& v) : x(v), y(v), z(v) {};
/// Array constructor. Assumes a 3-component array
Vec3(const T* v) : x(v[0]), y(v[1]), z(v[2]) {};
// *****************************************
// Conversions/Assignment/Indexing
// *****************************************
/// cast to T*
operator const T* () const { return (const T*)this; };
/// cast to T*
operator T* () { return (T*)this; };
/// Assignment
const Vec3<T>& operator=(const Vec3<T>& rhs) { x = rhs.x; y = rhs.y; z = rhs.z; return *this; };
// *****************************************
// Comparison
// *****************************************
/// Equality comparison
bool operator==(const Vec3<T>& rhs) const { return (x == rhs.x && y == rhs.y && z == rhs.z); };
/// Inequality comparision
bool operator!=(const Vec3<T>& rhs) const { return (x != rhs.x || y != rhs.y || z != rhs.z); };
// *****************************************
// Arithmetic
// *****************************************
/// Addition
const Vec3<T> operator+(const Vec3<T>& rhs) const { return Vec3<T>(x + rhs.x, y + rhs.y, z + rhs.z); };
/// Subtraction
const Vec3<T> operator-(const Vec3<T>& rhs) const { return Vec3<T>(x - rhs.x, y - rhs.y, z - rhs.z); };
/// Multiply by scalar
const Vec3<T> operator*(const T& v) const { return Vec3<T>(x * v, y * v, z * v); };
/// Divide by scalar
const Vec3<T> operator/(const T& v) const { return Vec3<T>(x / v, y / v, z / v); };
/// Divide by vector
const Vec3<T> operator/(const Vec3<T>& rhs) const { return Vec3<T>(x / rhs.x, y / rhs.y, z / rhs.z); };
/// Addition in-place
Vec3<T>& operator+= (const Vec3<T>& rhs) { x += rhs.x; y += rhs.y; z += rhs.z; return *this; };
/// Subtract in-place
Vec3<T>& operator-= (const Vec3<T>& rhs) { x -= rhs.x; y -= rhs.y; z -= rhs.z; return *this; };
/// Scalar multiply in-place
Vec3<T>& operator*= (const T& v) { x *= v; y *= v; z *= v; return *this; };
/// Scalar divide in-place
Vec3<T>& operator/= (const T& v) { x /= v; y /= v; z /= v; return *this; };
};
typedef Vec3<float> CGU_Vec3f;
typedef Vec3<float> CGV_Vec3f;
typedef Vec3<unsigned char> CGU_Vec3uc;
typedef Vec3<unsigned char> CGV_Vec3uc;
typedef Vec3<float> CMP_Vec3f;
typedef Vec3<double> CMP_Vec3d;
typedef Vec3<int> CMP_Vec3i;
typedef Vec3<unsigned char> CMP_Vec3uc;
//============================================= VEC4 ==================================================
template<class T>
class Vec4
{
public:
T x;
T y;
T z;
T w;
// *****************************************
// Constructors
// *****************************************
/// Default constructor
Vec4() : x((T)0), y((T)0), z((T)0), w((T)0) {};
/// Value constructor
Vec4(const T& vx, const T& vy, const T& vz, const T& vw) : x(vx), y(vy), z(vz), w(vw) {};
/// Copy constructor
Vec4(const Vec4<T>& val) : x(val.x), y(val.y), z(val.z), w(val.w) {};
/// Single value constructor. Sets all components to the given value
Vec4(const T& v) : x(v), y(v), z(v), w(v) {};
/// Array constructor. Assumes a 4-component array
Vec4(const T* v) : x(v[0]), y(v[1]), z(v[2]), w(v[3]) {};
// *****************************************
// Conversions/Assignment/Indexing
// *****************************************
/// cast to T*
operator const T* () const { return (const T*)this; };
/// cast to T*
operator T* () { return (T*)this; };
/// Assignment
const Vec4<T>& operator=(const Vec4<T>& rhs) { x = rhs.x; y = rhs.y; z = rhs.z; w = rhs.w; return *this; };
// *****************************************
// Comparison
// *****************************************
/// Equality comparison
bool operator==(const Vec4<T>& rhs) const { return (x == rhs.x && y == rhs.y && z == rhs.z && w == rhs.w); };
/// Inequality comparision
bool operator!=(const Vec4<T>& rhs) const { return (x != rhs.x || y != rhs.y || z != rhs.z || w != rhs.w); };
// *****************************************
// Arithmetic
// *****************************************
/// Addition
const Vec4<T> operator+(const Vec4<T>& rhs) const { return Vec4<T>(x + rhs.x, y + rhs.y, z + rhs.z, w + rhs.w); };
/// Subtraction
const Vec4<T> operator-(const Vec4<T>& rhs) const { return Vec4<T>(x - rhs.x, y - rhs.y, z - rhs.z, w - rhs.w); };
/// Multiply by scalar
const Vec4<T> operator*(const T& v) const { return Vec4<T>(x * v, y * v, z * v, w * v); };
/// Divide by scalar
const Vec4<T> operator/(const T& v) const { return Vec4<T>(x / v, y / v, z / v, w / v); };
/// Divide by vector
const Vec4<T> operator/(const Vec4<T>& rhs) const { return Vec4<T>(x / rhs.x, y / rhs.y, z / rhs.z, w / rhs.w); };
/// Addition in-place
Vec4<T>& operator+= (const Vec4<T>& rhs) { x += rhs.x; y += rhs.y; z += rhs.z; w += rhs.w; return *this; };
/// Subtract in-place
Vec4<T>& operator-= (const Vec4<T>& rhs) { x -= rhs.x; y -= rhs.y; z -= rhs.z; w -= rhs.w; return *this; };
/// Scalar multiply in-place
Vec4<T>& operator*= (const T& v) { x *= v; y *= v; z *= v; w *= v; return *this; };
/// Scalar divide in-place
Vec4<T>& operator/= (const T& v) { x /= v; y /= v; z /= v; w /= v; return *this; };
};
#include <stdio.h>
#include "xmmintrin.h"
#include <math.h>
#include <float.h>
// SSE Vec4
#ifdef _LINUX
class CMP_SSEVec4f
#else
#include "intrin.h"
class __declspec(align(16)) CMP_SSEVec4f
#endif
{
public:
union
{
__m128 vec128; // float Vector 128 bits in total (16 Bytes) = array of 4 floats
#ifdef _LINUX
float f32[4];
#endif
};
// constructors
inline CMP_SSEVec4f() {};
inline CMP_SSEVec4f(float x, float y, float z, float w) : vec128(_mm_setr_ps(x, y, z, w)) {};
inline CMP_SSEVec4f(__m128 vec) : vec128(vec) {}
inline CMP_SSEVec4f(const float* data) : vec128(_mm_load_ps(data)) {};
inline CMP_SSEVec4f(float scalar) : vec128(_mm_load1_ps(&scalar)) {};
// copy and assignment
inline CMP_SSEVec4f(const CMP_SSEVec4f& init) : vec128(init.vec128) {};
inline const CMP_SSEVec4f& operator=(const CMP_SSEVec4f& lhs) { vec128 = lhs.vec128; return *this; };
// conversion to m128 type for direct use in _mm intrinsics
inline operator __m128() { return vec128; };
inline operator const __m128() const { return vec128; };
// indexing
#ifdef _LINUX
inline const float& operator[](int i) const { return f32[i]; };
inline float& operator[](int i) { return f32[i]; };
#else
inline const float& operator[](int i) const { return vec128.m128_f32[i]; };
inline float& operator[](int i) { return vec128.m128_f32[i]; };
#endif
// addition
inline CMP_SSEVec4f operator+(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_add_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f& operator+=(const CMP_SSEVec4f& rhs) { vec128 = _mm_add_ps(vec128, rhs.vec128); return *this; };
// multiplication
inline CMP_SSEVec4f operator*(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_mul_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f& operator*=(const CMP_SSEVec4f& rhs) { vec128 = _mm_mul_ps(vec128, rhs.vec128); return *this; };
// scalar multiplication
//inline CMP_SSEVec4f operator*( float rhs ) const { return CMP_SSEVec4f( _mm_mul_ps(vec128, _mm_load1_ps(&rhs)) ); };
//inline CMP_SSEVec4f& operator*=( float rhs ) { vec128 = _mm_mul_ps(vec128, _mm_load1_ps(&rhs)); return *this; };
// subtraction
inline CMP_SSEVec4f operator-(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_sub_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f& operator-= (const CMP_SSEVec4f& rhs) { vec128 = _mm_sub_ps(vec128, rhs.vec128); return *this; };
// division
inline CMP_SSEVec4f operator/(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_div_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f& operator/= (const CMP_SSEVec4f& rhs) { vec128 = _mm_div_ps(vec128, rhs.vec128); return *this; };
// scalar division
inline CMP_SSEVec4f operator/(float rhs) const { return CMP_SSEVec4f(_mm_div_ps(vec128, _mm_load1_ps(&rhs))); };
inline CMP_SSEVec4f& operator/=(float rhs) { vec128 = _mm_div_ps(vec128, _mm_load1_ps(&rhs)); return *this; };
// comparison
// these return 0 or 0xffffffff in each component
inline CMP_SSEVec4f operator< (const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_cmplt_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f operator> (const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_cmpgt_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f operator<=(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_cmple_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f operator>=(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_cmpge_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f operator==(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_cmpeq_ps(vec128, rhs.vec128)); };
// bitwise operators
inline CMP_SSEVec4f operator|(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_or_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f operator&(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_and_ps(vec128, rhs.vec128)); };
inline CMP_SSEVec4f operator^(const CMP_SSEVec4f& rhs) const { return CMP_SSEVec4f(_mm_xor_ps(vec128, rhs.vec128)); };
inline const CMP_SSEVec4f& operator|=(const CMP_SSEVec4f& rhs) { vec128 = _mm_or_ps(vec128, rhs.vec128); return *this; };
inline const CMP_SSEVec4f& operator&=(const CMP_SSEVec4f& rhs) { vec128 = _mm_and_ps(vec128, rhs.vec128); return *this; };
// for some horrible reason,there's no bitwise not instruction for SSE,
// so we have to do xor with 0xfffffff in order to fake it.
// TO get a 0xffffffff, we execute 0=0
inline CMP_SSEVec4f operator~() const
{
__m128 zero = _mm_setzero_ps();
__m128 is_true = _mm_cmpeq_ps(zero, zero);
return _mm_xor_ps(is_true, vec128);
};
};
typedef Vec4<float> CMP_Vec4f;
typedef Vec4<double> CMP_Vec4d;
typedef Vec4<int> CMP_Vec4i;
typedef Vec4<unsigned int> CMP_Vec4ui; // unsigned 16 bit x,y,x,w
typedef Vec4<unsigned char> CMP_Vec4uc; // unsigned 8 bit x,y,x,w
typedef Vec4<unsigned char> CGU_Vec4uc; // unsigned 8 bit x,y,x,w
typedef Vec4<unsigned char> CGV_Vec4uc; // unsigned 8 bit x,y,x,w
#endif // not ASPM_GPU
#endif // Header Guard