sync with private repository.

CMakeLists.txt

@@ -39,7 +39,7 @@ IF(MSVC)
 	ADD_DEFINITIONS(-D__SSE2__ -D__SSE__ -D__MMX__)
 ENDIF(MSVC)
 
-#SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=pentium4")
+SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=pentium4")
 #SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mcpu=powerpc")
 
 # ibook G4:
@@ -48,3 +48,6 @@ ENDIF(MSVC)
 
 ADD_SUBDIRECTORY(src)
 
+IF(WIN32)
+	ADD_SUBDIRECTORY(gnuwin32)
+ENDIF(WIN32)

cmake/FindCUDA.cmake

@@ -19,6 +19,7 @@ IF (WIN32)
 ELSE(WIN32)
 	FIND_PROGRAM (CUDA_COMPILER nvcc
 		$ENV{CUDA_BIN_PATH}
+		/usr/local/cuda/bin
 		DOC "The CUDA Compiler")
 ENDIF(WIN32)
 
@@ -37,6 +38,14 @@ FIND_PATH (CUDA_INCLUDE_PATH cuda_runtime.h
 	DOC "The directory where CUDA headers reside")
     
 FIND_LIBRARY (CUDA_LIBRARY
+	NAMES cuda
+	PATHS
+	$ENV{CUDA_LIB_PATH}
+	${CUDA_COMPILER_SUPER_DIR}/lib
+	${CUDA_COMPILER_DIR}
+	DOC "The CUDA library")
+
+FIND_LIBRARY (CUDA_RUNTIME_LIBRARY
 	NAMES cudart
 	PATHS
 	$ENV{CUDA_LIB_PATH}
@@ -45,21 +54,23 @@ FIND_LIBRARY (CUDA_LIBRARY
 	DOC "The CUDA runtime library")
 
 
-IF (CUDA_INCLUDE_PATH)
+IF (CUDA_INCLUDE_PATH AND CUDA_LIBRARY AND CUDA_RUNTIME_LIBRARY)
 	SET (CUDA_FOUND 1 CACHE STRING "Set to 1 if CUDA is found, 0 otherwise")
-ELSE (CUDA_INCLUDE_PATH)
+ELSE (CUDA_INCLUDE_PATH AND CUDA_LIBRARY AND CUDA_RUNTIME_LIBRARY)
 	SET (CUDA_FOUND 0 CACHE STRING "Set to 1 if CUDA is found, 0 otherwise")
-ENDIF (CUDA_INCLUDE_PATH)
+ENDIF (CUDA_INCLUDE_PATH AND CUDA_LIBRARY AND CUDA_RUNTIME_LIBRARY)
+
+SET (CUDA_LIBRARIES ${CUDA_LIBRARY} ${CUDA_RUNTIME_LIBRARY})
 
-MARK_AS_ADVANCED (CUDA_FOUND CUDA_COMPILER)
+MARK_AS_ADVANCED (CUDA_FOUND CUDA_COMPILER CUDA_LIBRARY CUDA_RUNTIME_LIBRARY)
 
 
 #SET(CUDA_OPTIONS "-ncfe")
 SET(CUDA_OPTIONS "")
 
-IF (CUDA_EMU)
+IF (CUDA_EMULATION)
 	SET (CUDA_OPTIONS "${CUDA_OPTIONS} -deviceemu")
-ENDIF (CUDA_EMU)
+ENDIF (CUDA_EMULATION)
 
 
 # Get include directories.
@@ -89,15 +100,24 @@ MACRO (GET_CUFILE_DEPENDENCIES dependencies file)
 	
 	#  parse file for dependencies
 	FILE(READ "${file}" CONTENTS)
-	STRING(REGEX MATCHALL "#[ \t]*include[ \t]+[<\"][^>\"]*" DEPS "${CONTENTS}")
 	
 	SET(${dependencies})
 	
 	FOREACH(DEP ${DEPS})
-		STRING(REGEX REPLACE "#[ \t]*include[ \t]+[<\"]" "" DEP "${DEP}")
-		SET(${dependencies} ${${dependencies}} ${filepath}/${DEP})
+		STRING(REGEX REPLACE "#[ \t]*include[ \t]+\"" "" DEP "${DEP}")
+
+		FIND_PATH(PATH_OF_${DEP} ${DEP}
+			${filepath})
+
+		IF(NOT ${PATH_OF_${DEP}} STREQUAL PATH_OF_${DEP}-NOTFOUND)
+			MESSAGE("${filepath} ${PATH_OF_${DEP}}/${DEP}")
+			SET(${dependencies} ${${dependencies}} ${PATH_OF_${DEP}}/${DEP})
+		ENDIF(NOT ${PATH_OF_${DEP}} STREQUAL PATH_OF_${DEP}-NOTFOUND)
+		
+		# reset path.
+		#SET(PATH_OF_${DEP})
 	ENDFOREACH(DEP)
-	
+
 ENDMACRO (GET_CUFILE_DEPENDENCIES)
 
 

gnuwin32/CMakeLists.txt

@@ -0,0 +1,7 @@
+
+INSTALL(PROGRAMS 
+	bin/jpeg62.dll
+	bin/libpng13.dll
+	bin/libtiff3.dll
+	bin/zlib1.dll
+	DESTINATION bin)

src/nvcore/CMakeLists.txt

@@ -34,3 +34,7 @@ ENDIF(NVCORE_SHARED)
 
 TARGET_LINK_LIBRARIES(nvcore ${LIBS})
 
+INSTALL(TARGETS nvcore
+	RUNTIME DESTINATION bin
+	LIBRARY DESTINATION lib
+	ARCHIVE DESTINATION lib/static)

src/nvcore/Debug.h

@@ -13,6 +13,8 @@
 #define NV_ABORT_IGNORE		2
 #define NV_ABORT_EXIT		3
 
+#undef assert	// avoid conflicts with assert method.
+
 #define nvNoAssert(exp) \
 	do { \
 		(void)sizeof(exp); \

src/nvcore/StdStream.h

@@ -13,6 +13,7 @@ namespace nv
 // Portable version of fopen.
 inline FILE * fileOpen(const char * fileName, const char * mode)
 {
+	nvCheck(fileName != NULL);
 #if NV_CC_MSVC && _MSC_VER >= 1400
 	FILE * fp;
 	if (fopen_s(&fp, fileName, mode) == 0) {
@@ -48,6 +49,7 @@ public:
 		virtual void seek( int pos )
 		{
 			nvDebugCheck(m_fp != NULL);
+			nvDebugCheck(pos >= 0 && pos < size());
 			fseek(m_fp, pos, SEEK_SET);
 		}
 		
@@ -59,6 +61,7 @@ public:
 		
 		virtual int size() const
 		{
+			nvDebugCheck(m_fp != NULL);
 			int pos = ftell(m_fp);
 			fseek(m_fp, 0, SEEK_END);
 			int end = ftell(m_fp);

src/nvcore/StrLib.cpp

@@ -196,6 +196,12 @@ StringBuilder::StringBuilder( const StringBuilder & s ) : m_size(0), m_str(NULL)
 	copy(s);
 }
 
+/** Copy string. */
+StringBuilder::StringBuilder( const char * s )
+{
+	copy(s);
+}
+
 /** Allocate and copy string. */
 StringBuilder::StringBuilder( int size_hint, const StringBuilder & s) : m_size(size_hint), m_str(NULL)
 {

src/nvcore/StrLib.h

@@ -44,6 +44,7 @@ namespace nv
 	
 		NVCORE_API StringBuilder();
 		NVCORE_API explicit StringBuilder( int size_hint );
+		NVCORE_API StringBuilder( const char * str );
 		NVCORE_API StringBuilder( const StringBuilder & );
 		NVCORE_API StringBuilder( int size_hint, const StringBuilder & );	
 		NVCORE_API StringBuilder( const char * format, ... ) __attribute__((format (printf, 2, 3)));
@@ -69,7 +70,7 @@ namespace nv
 		NVCORE_API StringBuilder & toUpper();
 		
 		NVCORE_API void reset();
-		NVCORE_API bool empty() const { return m_size == 0; }
+		NVCORE_API bool isNull() const { return m_size == 0; }
 	
 		// const char * accessors
 		operator const char * () const { return m_str; }
@@ -81,22 +82,27 @@ namespace nv
 		StringBuilder & operator=( const StringBuilder & s ) {
 			return copy(s);
 		}
-	
+
+		/// Implement value semantics.
+		StringBuilder & operator=( const char * s ) {
+			return copy(s);
+		}
+
 		/// Equal operator.
 		bool operator==( const StringBuilder & s ) const {
-			nvCheck(m_str != NULL);
-			nvCheck(s.m_str != NULL);
-			return strcmp(s.m_str, m_str) != 0;
+			if (s.isNull()) return isNull();
+			else if (isNull()) return false;
+			else return strcmp(s.m_str, m_str) != 0;
 		}
 		
 		/// Return the exact length.
-		uint length() const { nvCheck(m_str != NULL); return uint(strlen(m_str)); }
+		uint length() const { return isNull() ? 0 : uint(strlen(m_str)); }
 	
 		/// Return the size of the string container.
-		uint capacity() const { nvCheck(m_str != NULL); return m_size; }
+		uint capacity() const { return m_size; }
 	
 		/// Return the hash of the string.
-		uint hash() const { nvCheck(m_str != NULL); return strHash(m_str); }
+		uint hash() const { return isNull() ? 0 : strHash(m_str); }
 	
 		///	Swap strings.
 		friend void swap(StringBuilder & a, StringBuilder & b) {
@@ -104,8 +110,6 @@ namespace nv
 			nv::swap(a.m_str, b.m_str);
 		}
 	
-		static char separator();
-		
 	protected:
 		
 		/// Size of the string container.

src/nvimage/CMakeLists.txt

@@ -55,4 +55,8 @@ ENDIF(NVIMAGE_SHARED)
 
 TARGET_LINK_LIBRARIES(nvimage ${LIBS} nvcore nvmath posh)
 
+INSTALL(TARGETS nvimage
+	RUNTIME DESTINATION bin
+	LIBRARY DESTINATION lib
+	ARCHIVE DESTINATION lib/static)
 

src/nvimage/DirectDrawSurface.cpp

@@ -528,6 +528,13 @@ void DirectDrawSurface::mipmap(Image * img, uint face, uint mipmap)
 	}
 }
 
+static uint8 bitExpand(uint8 c, uint bits)
+{
+	int shifts = 0;
+	uint8 output = c;
+	// @@ TODO!!!
+	
+}
 
 void DirectDrawSurface::readLinearImage(Image * img)
 {

src/nvimage/nvtt/cuda/CompressKernel.cu

@@ -220,7 +220,7 @@ __device__ float evalPermutation4(const float3 * colors, uint permutation, ushor
         alphax_sum += alpha * colors[i];
         betax_sum += beta * colors[i];
     }
-
+
     const float factor = 1.0f / (alpha2_sum * beta2_sum - alphabeta_sum * alphabeta_sum);
 
     float3 a = (alphax_sum * beta2_sum - betax_sum * alphabeta_sum) * factor;
@@ -689,33 +689,33 @@ __global__ void compressWeighted(const uint * permutations, const uint * image,
 
 __device__ float computeError(const float weights[16], uchar a0, uchar a1)
 {
-	float palette[6];
-	palette[0] = (6.0f/7.0f * a0 + 1.0f/7.0f * a1);
-	palette[1] = (5.0f/7.0f * a0 + 2.0f/7.0f * a1);
-	palette[2] = (4.0f/7.0f * a0 + 3.0f/7.0f * a1);
-	palette[3] = (3.0f/7.0f * a0 + 4.0f/7.0f * a1);
-	palette[4] = (2.0f/7.0f * a0 + 5.0f/7.0f * a1);
-	palette[5] = (1.0f/7.0f * a0 + 6.0f/7.0f * a1);
-
-	float total = 0.0f;
-
-	for (uint i = 0; i < 16; i++)
-	{
-		float alpha = weights[i];
-
-		float error = a0 - alpha;
-		error = min(error, palette[0] - alpha);
-		error = min(error, palette[1] - alpha);
-		error = min(error, palette[2] - alpha);
-		error = min(error, palette[3] - alpha);
-		error = min(error, palette[4] - alpha);
-		error = min(error, palette[5] - alpha);
-		error = min(error, a1 - alpha);
-		
-		total += error;
-	}
-	
-	return total;
+	float palette[6];
+	palette[0] = (6.0f/7.0f * a0 + 1.0f/7.0f * a1);
+	palette[1] = (5.0f/7.0f * a0 + 2.0f/7.0f * a1);
+	palette[2] = (4.0f/7.0f * a0 + 3.0f/7.0f * a1);
+	palette[3] = (3.0f/7.0f * a0 + 4.0f/7.0f * a1);
+	palette[4] = (2.0f/7.0f * a0 + 5.0f/7.0f * a1);
+	palette[5] = (1.0f/7.0f * a0 + 6.0f/7.0f * a1);
+
+	float total = 0.0f;
+
+	for (uint i = 0; i < 16; i++)
+	{
+		float alpha = weights[i];
+
+		float error = a0 - alpha;
+		error = min(error, palette[0] - alpha);
+		error = min(error, palette[1] - alpha);
+		error = min(error, palette[2] - alpha);
+		error = min(error, palette[3] - alpha);
+		error = min(error, palette[4] - alpha);
+		error = min(error, palette[5] - alpha);
+		error = min(error, a1 - alpha);
+		
+		total += error;
+	}
+	
+	return total;
 }
 
 inline __device__ uchar roundAndExpand(float a)
@@ -726,35 +726,35 @@ inline __device__ uchar roundAndExpand(float a)
 /*
 __device__ void optimizeAlpha8(const float alphas[16], uchar & a0, uchar & a1)
 {
-	float alpha2_sum = 0;
-	float beta2_sum = 0;
-	float alphabeta_sum = 0;
-	float alphax_sum = 0;
-	float betax_sum = 0;
-
-	for (int i = 0; i < 16; i++)
-	{
-		uint idx = index[i];
-		float alpha;
-		if (idx < 2) alpha = 1.0f - idx;
-		else alpha = (8.0f - idx) / 7.0f;
-		
-		float beta = 1 - alpha;
-
-        alpha2_sum += alpha * alpha;
-        beta2_sum += beta * beta;
-        alphabeta_sum += alpha * beta;
-        alphax_sum += alpha * alphas[i];
-        betax_sum += beta * alphas[i];
-	}
-
-    const float factor = 1.0f / (alpha2_sum * beta2_sum - alphabeta_sum * alphabeta_sum);
-
-    float a = (alphax_sum * beta2_sum - betax_sum * alphabeta_sum) * factor;
-	float b = (betax_sum * alpha2_sum - alphax_sum * alphabeta_sum) * factor;
-
-	a0 = roundAndExpand(a);
-	a1 = roundAndExpand(b);
+	float alpha2_sum = 0;
+	float beta2_sum = 0;
+	float alphabeta_sum = 0;
+	float alphax_sum = 0;
+	float betax_sum = 0;
+
+	for (int i = 0; i < 16; i++)
+	{
+		uint idx = index[i];
+		float alpha;
+		if (idx < 2) alpha = 1.0f - idx;
+		else alpha = (8.0f - idx) / 7.0f;
+		
+		float beta = 1 - alpha;
+
+        alpha2_sum += alpha * alpha;
+        beta2_sum += beta * beta;
+        alphabeta_sum += alpha * beta;
+        alphax_sum += alpha * alphas[i];
+        betax_sum += beta * alphas[i];
+	}
+
+    const float factor = 1.0f / (alpha2_sum * beta2_sum - alphabeta_sum * alphabeta_sum);
+
+    float a = (alphax_sum * beta2_sum - betax_sum * alphabeta_sum) * factor;
+	float b = (betax_sum * alpha2_sum - alphax_sum * alphabeta_sum) * factor;
+
+	a0 = roundAndExpand(a);
+	a1 = roundAndExpand(b);
 }
 */
 

src/nvimage/nvtt/tools/compress.cpp

@@ -238,7 +238,7 @@ int main(int argc, char *argv[])
 		}
 	}
 
-	if (input.empty())
+	if (input.isNull())
 	{
 		printf("NVIDIA Texture Tools - Copyright NVIDIA Corporation 2007\n\n");
 		

src/nvmath/CMakeLists.txt

@@ -10,7 +10,8 @@ SET(MATH_SRCS
 	Fitting.h Fitting.cpp
 	Montecarlo.h Montecarlo.cpp
 	Random.h Random.cpp
-	SphericalHarmonic.h SphericalHarmonic.cpp)
+	SphericalHarmonic.h SphericalHarmonic.cpp
+	Basis.h Basis.cpp)
 
 INCLUDE_DIRECTORIES(${CMAKE_CURRENT_SOURCE_DIR})
 
@@ -25,3 +26,7 @@ ENDIF(NVMATH_SHARED)
 
 TARGET_LINK_LIBRARIES(nvmath ${LIBS} nvcore)
 
+INSTALL(TARGETS nvmath
+	RUNTIME DESTINATION bin
+	LIBRARY DESTINATION lib
+	ARCHIVE DESTINATION lib/static)

src/nvmath/Matrix.h

@@ -33,12 +33,12 @@ public:
 	Vector4 row(uint i) const;
 	Vector4 column(uint i) const;
 	
-	void scale(double s);
+	void scale(scalar s);
 	void scale(Vector3::Arg s);
 	void translate(Vector3::Arg t);
 	void rotate(scalar theta, scalar v0, scalar v1, scalar v2);
     Matrix inverse();
-    double determinant();
+    scalar determinant();
 	
 	void apply(Matrix::Arg m);
 
@@ -110,7 +110,7 @@ inline Vector4 Matrix::column(uint i) const
 }
 
 /// Apply scale.
-inline void Matrix::scale(double s)
+inline void Matrix::scale(scalar s)
 {
 	m_data[0] *= s; m_data[1] *= s; m_data[2] *= s; m_data[3] *= s;
 	m_data[4] *= s; m_data[5] *= s; m_data[6] *= s; m_data[7] *= s;
@@ -150,7 +150,7 @@ inline void Matrix::apply(Matrix::Arg m)
 	nvDebugCheck(this != &m);
 	
 	for(int i = 0; i < 4; i++) {
-		const float ai0 = get(i,0), ai1 = get(i,1), ai2 = get(i,2), ai3 = get(i,3);
+		const scalar ai0 = get(i,0), ai1 = get(i,1), ai2 = get(i,2), ai3 = get(i,3);
 		m_data[0 + i] = ai0 * m(0,0) + ai1 * m(1,0) + ai2 * m(2,0) + ai3 * m(3,0);
 		m_data[4 + i] = ai0 * m(0,1) + ai1 * m(1,1) + ai2 * m(2,1) + ai3 * m(3,1);
 		m_data[8 + i] = ai0 * m(0,2) + ai1 * m(1,2) + ai2 * m(2,2) + ai3 * m(3,2);
@@ -305,7 +305,7 @@ inline Matrix perspective(scalar fovy, scalar aspect, scalar zNear)
 }
 
 /// Get matrix determinant.
-inline double Matrix::determinant()
+inline scalar Matrix::determinant()
 {
    return m_data[3] * m_data[6] * m_data[9] * m_data[12]-m_data[2] * m_data[7] * m_data[9] * m_data[12]-m_data[3] * m_data[5] * m_data[10] * m_data[12]+m_data[1] * m_data[7]    * m_data[10] * m_data[12]+
    m_data[2] * m_data[5] * m_data[11] * m_data[12]-m_data[1] * m_data[6] * m_data[11] * m_data[12]-m_data[3] * m_data[6] * m_data[8] * m_data[13]+m_data[2] * m_data[7]    * m_data[8] * m_data[13]+
@@ -315,9 +315,19 @@ inline double Matrix::determinant()
    m_data[2] * m_data[4] * m_data[9] * m_data[15]-m_data[0] * m_data[6] * m_data[9] * m_data[15]-m_data[1] * m_data[4] * m_data[10] * m_data[15]+m_data[0] * m_data[5]    * m_data[10] * m_data[15];
 }
 
-//inline Matrix transpose(Matrix::Arg m)
-//{
-//}
+inline Matrix transpose(Matrix::Arg m)
+{
+	Matrix r;
+	for (int i = 0; i < 4; i++)
+	{
+		for (int j = 0; j < 4; i++)
+		{
+			r(i, j) = m(j, i);
+		}
+	}
+	return r;
+}
+
 inline Matrix Matrix::inverse()
 {
    Matrix r;

src/nvmath/Vector.h

@@ -30,6 +30,8 @@ public:
 	scalar x() const;
 	scalar y() const;
 	
+	const scalar * ptr() const;
+
 	void set(scalar x, scalar y);
 	
 	Vector2 operator-() const;
@@ -143,6 +145,11 @@ inline const Vector2 & Vector2::operator=(Vector2::Arg v)
 inline scalar Vector2::x() const { return m_x; }
 inline scalar Vector2::y() const { return m_y; }
 
+inline const scalar * Vector2::ptr() const
+{
+	return &m_x;
+}
+
 inline void Vector2::set(scalar x, scalar y)
 {
 	m_x = x;
@@ -397,6 +404,11 @@ inline Vector2 operator*(Vector2::Arg v, scalar s)
 	return scale(v, s);
 }
 
+inline Vector2 operator*(Vector2::Arg v1, Vector2::Arg v2)
+{
+	return Vector2(v1.x()*v2.x(), v1.y()*v2.y());
+}
+
 inline Vector2 operator*(scalar s, Vector2::Arg v)
 {
 	return scale(v, s);
@@ -417,6 +429,11 @@ inline scalar length(Vector2::Arg v)
 	return sqrtf(length_squared(v));
 }
 
+inline bool equal(Vector2::Arg v1, Vector2::Arg v2, float epsilon = NV_EPSILON)
+{
+	return equal(v1.x(), v2.x(), epsilon) && equal(v1.y(), v2.y(), epsilon);
+}
+
 inline Vector2 min(Vector2::Arg a, Vector2::Arg b)
 {
 	return Vector2(min(a.x(), b.x()), min(a.y(), b.y()));
@@ -532,6 +549,11 @@ inline Vector3 normalizeSafe(Vector3::Arg v, Vector3::Arg fallback, float epsilo
 	return scale(v, 1.0f / l);
 }
 
+inline bool equal(Vector3::Arg v1, Vector3::Arg v2, float epsilon = NV_EPSILON)
+{
+	return equal(v1.x(), v2.x(), epsilon) && equal(v1.y(), v2.y(), epsilon) && equal(v1.z(), v2.z(), epsilon);
+}
+
 inline Vector3 min(Vector3::Arg a, Vector3::Arg b)
 {
 	return Vector3(min(a.x(), b.x()), min(a.y(), b.y()), min(a.z(), b.z()));
@@ -643,6 +665,11 @@ inline Vector4 normalizeSafe(Vector4::Arg v, Vector4::Arg fallback, float epsilo
 	return scale(v, 1.0f / l);
 }
 
+inline bool equal(Vector4::Arg v1, Vector4::Arg v2, float epsilon = NV_EPSILON)
+{
+	return equal(v1.x(), v2.x(), epsilon) && equal(v1.y(), v2.y(), epsilon) && equal(v1.z(), v2.z(), epsilon) && equal(v1.w(), v2.w(), epsilon);
+}
+
 inline Vector4 min(Vector4::Arg a, Vector4::Arg b)
 {
 	return Vector4(min(a.x(), b.x()), min(a.y(), b.y()), min(a.z(), b.z()), min(a.w(), b.w()));