Update RealDenseFace image reconstruction demo (part 2)

cuda_extensions/third_party/glm/glm/ext.hpp

@@ -0,0 +1,255 @@
+/// @file glm/ext.hpp
+///
+/// @ref core (Dependence)
+
+#include "detail/setup.hpp"
+
+#pragma once
+
+#include "glm.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_MESSAGE_EXT_INCLUDED_DISPLAYED)
+#	define GLM_MESSAGE_EXT_INCLUDED_DISPLAYED
+#	pragma message("GLM: All extensions included (not recommended)")
+#endif//GLM_MESSAGES
+
+#include "./ext/matrix_clip_space.hpp"
+#include "./ext/matrix_common.hpp"
+
+#include "./ext/matrix_double2x2.hpp"
+#include "./ext/matrix_double2x2_precision.hpp"
+#include "./ext/matrix_double2x3.hpp"
+#include "./ext/matrix_double2x3_precision.hpp"
+#include "./ext/matrix_double2x4.hpp"
+#include "./ext/matrix_double2x4_precision.hpp"
+#include "./ext/matrix_double3x2.hpp"
+#include "./ext/matrix_double3x2_precision.hpp"
+#include "./ext/matrix_double3x3.hpp"
+#include "./ext/matrix_double3x3_precision.hpp"
+#include "./ext/matrix_double3x4.hpp"
+#include "./ext/matrix_double3x4_precision.hpp"
+#include "./ext/matrix_double4x2.hpp"
+#include "./ext/matrix_double4x2_precision.hpp"
+#include "./ext/matrix_double4x3.hpp"
+#include "./ext/matrix_double4x3_precision.hpp"
+#include "./ext/matrix_double4x4.hpp"
+#include "./ext/matrix_double4x4_precision.hpp"
+
+#include "./ext/matrix_float2x2.hpp"
+#include "./ext/matrix_float2x2_precision.hpp"
+#include "./ext/matrix_float2x3.hpp"
+#include "./ext/matrix_float2x3_precision.hpp"
+#include "./ext/matrix_float2x4.hpp"
+#include "./ext/matrix_float2x4_precision.hpp"
+#include "./ext/matrix_float3x2.hpp"
+#include "./ext/matrix_float3x2_precision.hpp"
+#include "./ext/matrix_float3x3.hpp"
+#include "./ext/matrix_float3x3_precision.hpp"
+#include "./ext/matrix_float3x4.hpp"
+#include "./ext/matrix_float3x4_precision.hpp"
+#include "./ext/matrix_float4x2.hpp"
+#include "./ext/matrix_float4x2_precision.hpp"
+#include "./ext/matrix_float4x3.hpp"
+#include "./ext/matrix_float4x3_precision.hpp"
+#include "./ext/matrix_float4x4.hpp"
+#include "./ext/matrix_float4x4_precision.hpp"
+
+#include "./ext/matrix_int2x2.hpp"
+#include "./ext/matrix_int2x2_sized.hpp"
+#include "./ext/matrix_int2x3.hpp"
+#include "./ext/matrix_int2x3_sized.hpp"
+#include "./ext/matrix_int2x4.hpp"
+#include "./ext/matrix_int2x4_sized.hpp"
+#include "./ext/matrix_int3x2.hpp"
+#include "./ext/matrix_int3x2_sized.hpp"
+#include "./ext/matrix_int3x3.hpp"
+#include "./ext/matrix_int3x3_sized.hpp"
+#include "./ext/matrix_int3x4.hpp"
+#include "./ext/matrix_int3x4_sized.hpp"
+#include "./ext/matrix_int4x2.hpp"
+#include "./ext/matrix_int4x2_sized.hpp"
+#include "./ext/matrix_int4x3.hpp"
+#include "./ext/matrix_int4x3_sized.hpp"
+#include "./ext/matrix_int4x4.hpp"
+#include "./ext/matrix_int4x4_sized.hpp"
+
+#include "./ext/matrix_uint2x2.hpp"
+#include "./ext/matrix_uint2x2_sized.hpp"
+#include "./ext/matrix_uint2x3.hpp"
+#include "./ext/matrix_uint2x3_sized.hpp"
+#include "./ext/matrix_uint2x4.hpp"
+#include "./ext/matrix_uint2x4_sized.hpp"
+#include "./ext/matrix_uint3x2.hpp"
+#include "./ext/matrix_uint3x2_sized.hpp"
+#include "./ext/matrix_uint3x3.hpp"
+#include "./ext/matrix_uint3x3_sized.hpp"
+#include "./ext/matrix_uint3x4.hpp"
+#include "./ext/matrix_uint3x4_sized.hpp"
+#include "./ext/matrix_uint4x2.hpp"
+#include "./ext/matrix_uint4x2_sized.hpp"
+#include "./ext/matrix_uint4x3.hpp"
+#include "./ext/matrix_uint4x3_sized.hpp"
+#include "./ext/matrix_uint4x4.hpp"
+#include "./ext/matrix_uint4x4_sized.hpp"
+
+#include "./ext/matrix_projection.hpp"
+#include "./ext/matrix_relational.hpp"
+#include "./ext/matrix_transform.hpp"
+
+#include "./ext/quaternion_common.hpp"
+#include "./ext/quaternion_double.hpp"
+#include "./ext/quaternion_double_precision.hpp"
+#include "./ext/quaternion_float.hpp"
+#include "./ext/quaternion_float_precision.hpp"
+#include "./ext/quaternion_exponential.hpp"
+#include "./ext/quaternion_geometric.hpp"
+#include "./ext/quaternion_relational.hpp"
+#include "./ext/quaternion_transform.hpp"
+#include "./ext/quaternion_trigonometric.hpp"
+
+#include "./ext/scalar_common.hpp"
+#include "./ext/scalar_constants.hpp"
+#include "./ext/scalar_integer.hpp"
+#include "./ext/scalar_packing.hpp"
+#include "./ext/scalar_reciprocal.hpp"
+#include "./ext/scalar_relational.hpp"
+#include "./ext/scalar_ulp.hpp"
+
+#include "./ext/scalar_int_sized.hpp"
+#include "./ext/scalar_uint_sized.hpp"
+
+#include "./ext/vector_common.hpp"
+#include "./ext/vector_integer.hpp"
+#include "./ext/vector_packing.hpp"
+#include "./ext/vector_reciprocal.hpp"
+#include "./ext/vector_relational.hpp"
+#include "./ext/vector_ulp.hpp"
+
+#include "./ext/vector_bool1.hpp"
+#include "./ext/vector_bool1_precision.hpp"
+#include "./ext/vector_bool2.hpp"
+#include "./ext/vector_bool2_precision.hpp"
+#include "./ext/vector_bool3.hpp"
+#include "./ext/vector_bool3_precision.hpp"
+#include "./ext/vector_bool4.hpp"
+#include "./ext/vector_bool4_precision.hpp"
+
+#include "./ext/vector_double1.hpp"
+#include "./ext/vector_double1_precision.hpp"
+#include "./ext/vector_double2.hpp"
+#include "./ext/vector_double2_precision.hpp"
+#include "./ext/vector_double3.hpp"
+#include "./ext/vector_double3_precision.hpp"
+#include "./ext/vector_double4.hpp"
+#include "./ext/vector_double4_precision.hpp"
+
+#include "./ext/vector_float1.hpp"
+#include "./ext/vector_float1_precision.hpp"
+#include "./ext/vector_float2.hpp"
+#include "./ext/vector_float2_precision.hpp"
+#include "./ext/vector_float3.hpp"
+#include "./ext/vector_float3_precision.hpp"
+#include "./ext/vector_float4.hpp"
+#include "./ext/vector_float4_precision.hpp"
+
+#include "./ext/vector_int1.hpp"
+#include "./ext/vector_int1_sized.hpp"
+#include "./ext/vector_int2.hpp"
+#include "./ext/vector_int2_sized.hpp"
+#include "./ext/vector_int3.hpp"
+#include "./ext/vector_int3_sized.hpp"
+#include "./ext/vector_int4.hpp"
+#include "./ext/vector_int4_sized.hpp"
+
+#include "./ext/vector_uint1.hpp"
+#include "./ext/vector_uint1_sized.hpp"
+#include "./ext/vector_uint2.hpp"
+#include "./ext/vector_uint2_sized.hpp"
+#include "./ext/vector_uint3.hpp"
+#include "./ext/vector_uint3_sized.hpp"
+#include "./ext/vector_uint4.hpp"
+#include "./ext/vector_uint4_sized.hpp"
+
+#include "./gtc/bitfield.hpp"
+#include "./gtc/color_space.hpp"
+#include "./gtc/constants.hpp"
+#include "./gtc/epsilon.hpp"
+#include "./gtc/integer.hpp"
+#include "./gtc/matrix_access.hpp"
+#include "./gtc/matrix_integer.hpp"
+#include "./gtc/matrix_inverse.hpp"
+#include "./gtc/matrix_transform.hpp"
+#include "./gtc/noise.hpp"
+#include "./gtc/packing.hpp"
+#include "./gtc/quaternion.hpp"
+#include "./gtc/random.hpp"
+#include "./gtc/reciprocal.hpp"
+#include "./gtc/round.hpp"
+#include "./gtc/type_precision.hpp"
+#include "./gtc/type_ptr.hpp"
+#include "./gtc/ulp.hpp"
+#include "./gtc/vec1.hpp"
+#if GLM_CONFIG_ALIGNED_GENTYPES == GLM_ENABLE
+#	include "./gtc/type_aligned.hpp"
+#endif
+
+#ifdef GLM_ENABLE_EXPERIMENTAL
+#include "./gtx/associated_min_max.hpp"
+#include "./gtx/bit.hpp"
+#include "./gtx/closest_point.hpp"
+#include "./gtx/color_encoding.hpp"
+#include "./gtx/color_space.hpp"
+#include "./gtx/color_space_YCoCg.hpp"
+#include "./gtx/compatibility.hpp"
+#include "./gtx/component_wise.hpp"
+#include "./gtx/dual_quaternion.hpp"
+#include "./gtx/euler_angles.hpp"
+#include "./gtx/extend.hpp"
+#include "./gtx/extended_min_max.hpp"
+#include "./gtx/fast_exponential.hpp"
+#include "./gtx/fast_square_root.hpp"
+#include "./gtx/fast_trigonometry.hpp"
+#include "./gtx/functions.hpp"
+#include "./gtx/gradient_paint.hpp"
+#include "./gtx/handed_coordinate_space.hpp"
+#include "./gtx/integer.hpp"
+#include "./gtx/intersect.hpp"
+#include "./gtx/log_base.hpp"
+#include "./gtx/matrix_cross_product.hpp"
+#include "./gtx/matrix_interpolation.hpp"
+#include "./gtx/matrix_major_storage.hpp"
+#include "./gtx/matrix_operation.hpp"
+#include "./gtx/matrix_query.hpp"
+#include "./gtx/mixed_product.hpp"
+#include "./gtx/norm.hpp"
+#include "./gtx/normal.hpp"
+#include "./gtx/normalize_dot.hpp"
+#include "./gtx/number_precision.hpp"
+#include "./gtx/optimum_pow.hpp"
+#include "./gtx/orthonormalize.hpp"
+#include "./gtx/perpendicular.hpp"
+#include "./gtx/polar_coordinates.hpp"
+#include "./gtx/projection.hpp"
+#include "./gtx/quaternion.hpp"
+#include "./gtx/raw_data.hpp"
+#include "./gtx/rotate_vector.hpp"
+#include "./gtx/spline.hpp"
+#include "./gtx/std_based_type.hpp"
+#if !((GLM_COMPILER & GLM_COMPILER_CUDA) || (GLM_COMPILER & GLM_COMPILER_HIP))
+#	include "./gtx/string_cast.hpp"
+#endif
+#include "./gtx/transform.hpp"
+#include "./gtx/transform2.hpp"
+#include "./gtx/vec_swizzle.hpp"
+#include "./gtx/vector_angle.hpp"
+#include "./gtx/vector_query.hpp"
+#include "./gtx/wrap.hpp"
+
+#if GLM_HAS_TEMPLATE_ALIASES
+#	include "./gtx/scalar_multiplication.hpp"
+#endif
+
+#if GLM_HAS_RANGE_FOR
+#	include "./gtx/range.hpp"
+#endif
+#endif//GLM_ENABLE_EXPERIMENTAL

cuda_extensions/third_party/glm/glm/ext/vector_int3_sized.hpp

@@ -0,0 +1,49 @@
+/// @ref ext_vector_int3_sized
+/// @file glm/ext/vector_int3_sized.hpp
+///
+/// @defgroup ext_vector_int3_sized GLM_EXT_vector_int3_sized
+/// @ingroup ext
+///
+/// Exposes sized signed integer vector of 3 components type.
+///
+/// Include <glm/ext/vector_int3_sized.hpp> to use the features of this extension.
+///
+/// @see ext_scalar_int_sized
+/// @see ext_vector_uint3_sized
+
+#pragma once
+
+#include "../ext/vector_int3.hpp"
+#include "../ext/scalar_int_sized.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_int3_sized extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_int3_sized
+	/// @{
+
+	/// 8 bit signed integer vector of 3 components type.
+	///
+	/// @see ext_vector_int3_sized
+	typedef vec<3, int8, defaultp>		i8vec3;
+
+	/// 16 bit signed integer vector of 3 components type.
+	///
+	/// @see ext_vector_int3_sized
+	typedef vec<3, int16, defaultp>		i16vec3;
+
+	/// 32 bit signed integer vector of 3 components type.
+	///
+	/// @see ext_vector_int3_sized
+	typedef vec<3, int32, defaultp>		i32vec3;
+
+	/// 64 bit signed integer vector of 3 components type.
+	///
+	/// @see ext_vector_int3_sized
+	typedef vec<3, int64, defaultp>		i64vec3;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_int4.hpp

@@ -0,0 +1,18 @@
+/// @ref core
+/// @file glm/ext/vector_int4.hpp
+
+#pragma once
+#include "../detail/type_vec4.hpp"
+
+namespace glm
+{
+	/// @addtogroup core_vector
+	/// @{
+
+	/// 4 components vector of signed integer numbers.
+	///
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 4.1.5 Vectors</a>
+	typedef vec<4, int, defaultp>		ivec4;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_int4_sized.hpp

@@ -0,0 +1,49 @@
+/// @ref ext_vector_int4_sized
+/// @file glm/ext/vector_int4_sized.hpp
+///
+/// @defgroup ext_vector_int4_sized GLM_EXT_vector_int4_sized
+/// @ingroup ext
+///
+/// Exposes sized signed integer vector of 4 components type.
+///
+/// Include <glm/ext/vector_int4_sized.hpp> to use the features of this extension.
+///
+/// @see ext_scalar_int_sized
+/// @see ext_vector_uint4_sized
+
+#pragma once
+
+#include "../ext/vector_int4.hpp"
+#include "../ext/scalar_int_sized.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_int4_sized extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_int4_sized
+	/// @{
+
+	/// 8 bit signed integer vector of 4 components type.
+	///
+	/// @see ext_vector_int4_sized
+	typedef vec<4, int8, defaultp>		i8vec4;
+
+	/// 16 bit signed integer vector of 4 components type.
+	///
+	/// @see ext_vector_int4_sized
+	typedef vec<4, int16, defaultp>		i16vec4;
+
+	/// 32 bit signed integer vector of 4 components type.
+	///
+	/// @see ext_vector_int4_sized
+	typedef vec<4, int32, defaultp>		i32vec4;
+
+	/// 64 bit signed integer vector of 4 components type.
+	///
+	/// @see ext_vector_int4_sized
+	typedef vec<4, int64, defaultp>		i64vec4;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_integer.hpp

@@ -0,0 +1,149 @@
+/// @ref ext_vector_integer
+/// @file glm/ext/vector_integer.hpp
+///
+/// @see core (dependence)
+/// @see ext_vector_integer (dependence)
+///
+/// @defgroup ext_vector_integer GLM_EXT_vector_integer
+/// @ingroup ext
+///
+/// Include <glm/ext/vector_integer.hpp> to use the features of this extension.
+
+#pragma once
+
+// Dependencies
+#include "../detail/setup.hpp"
+#include "../detail/qualifier.hpp"
+#include "../detail/_vectorize.hpp"
+#include "../vector_relational.hpp"
+#include "../common.hpp"
+#include <limits>
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_integer extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_integer
+	/// @{
+
+	/// Return true if the value is a power of two number.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, bool, Q> isPowerOfTwo(vec<L, T, Q> const& v);
+
+	/// Return the power of two number which value is just higher the input value,
+	/// round up to a power of two.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> nextPowerOfTwo(vec<L, T, Q> const& v);
+
+	/// Return the power of two number which value is just lower the input value,
+	/// round down to a power of two.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> prevPowerOfTwo(vec<L, T, Q> const& v);
+
+	/// Return true if the 'Value' is a multiple of 'Multiple'.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, bool, Q> isMultiple(vec<L, T, Q> const& v, T Multiple);
+
+	/// Return true if the 'Value' is a multiple of 'Multiple'.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, bool, Q> isMultiple(vec<L, T, Q> const& v, vec<L, T, Q> const& Multiple);
+
+	/// Higher multiple number of Source.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @param v Source values to which is applied the function
+	/// @param Multiple Must be a null or positive value
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> nextMultiple(vec<L, T, Q> const& v, T Multiple);
+
+	/// Higher multiple number of Source.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @param v Source values to which is applied the function
+	/// @param Multiple Must be a null or positive value
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> nextMultiple(vec<L, T, Q> const& v, vec<L, T, Q> const& Multiple);
+
+	/// Lower multiple number of Source.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @param v Source values to which is applied the function
+	/// @param Multiple Must be a null or positive value
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> prevMultiple(vec<L, T, Q> const& v, T Multiple);
+
+	/// Lower multiple number of Source.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed or unsigned integer scalar types.
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @param v Source values to which is applied the function
+	/// @param Multiple Must be a null or positive value
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> prevMultiple(vec<L, T, Q> const& v, vec<L, T, Q> const& Multiple);
+
+	/// Returns the bit number of the Nth significant bit set to
+	/// 1 in the binary representation of value.
+	/// If value bitcount is less than the Nth significant bit, -1 will be returned.
+	///
+	/// @tparam L An integer between 1 and 4 included that qualify the dimension of the vector.
+	/// @tparam T Signed or unsigned integer scalar types.
+	///
+	/// @see ext_vector_integer
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, int, Q> findNSB(vec<L, T, Q> const& Source, vec<L, int, Q> SignificantBitCount);
+
+	/// @}
+} //namespace glm
+
+#include "vector_integer.inl"

cuda_extensions/third_party/glm/glm/ext/vector_integer.inl

@@ -0,0 +1,85 @@
+#include "scalar_integer.hpp"
+
+namespace glm
+{
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, bool, Q> isPowerOfTwo(vec<L, T, Q> const& Value)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'isPowerOfTwo' only accept integer inputs");
+
+		vec<L, T, Q> const Result(abs(Value));
+		return equal(Result & (Result - vec<L, T, Q>(1)), vec<L, T, Q>(0));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> nextPowerOfTwo(vec<L, T, Q> const& v)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'nextPowerOfTwo' only accept integer inputs");
+
+		return detail::compute_ceilPowerOfTwo<L, T, Q, std::numeric_limits<T>::is_signed>::call(v);
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> prevPowerOfTwo(vec<L, T, Q> const& v)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'prevPowerOfTwo' only accept integer inputs");
+
+		return detail::functor1<vec, L, T, T, Q>::call(prevPowerOfTwo, v);
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, bool, Q> isMultiple(vec<L, T, Q> const& Value, T Multiple)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'isMultiple' only accept integer inputs");
+
+		return equal(Value % Multiple, vec<L, T, Q>(0));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, bool, Q> isMultiple(vec<L, T, Q> const& Value, vec<L, T, Q> const& Multiple)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'isMultiple' only accept integer inputs");
+
+		return equal(Value % Multiple, vec<L, T, Q>(0));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> nextMultiple(vec<L, T, Q> const& Source, T Multiple)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'nextMultiple' only accept integer inputs");
+
+		return detail::functor2<vec, L, T, Q>::call(nextMultiple, Source, vec<L, T, Q>(Multiple));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> nextMultiple(vec<L, T, Q> const& Source, vec<L, T, Q> const& Multiple)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'nextMultiple' only accept integer inputs");
+
+		return detail::functor2<vec, L, T, Q>::call(nextMultiple, Source, Multiple);
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> prevMultiple(vec<L, T, Q> const& Source, T Multiple)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'prevMultiple' only accept integer inputs");
+
+		return detail::functor2<vec, L, T, Q>::call(prevMultiple, Source, vec<L, T, Q>(Multiple));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> prevMultiple(vec<L, T, Q> const& Source, vec<L, T, Q> const& Multiple)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'prevMultiple' only accept integer inputs");
+
+		return detail::functor2<vec, L, T, Q>::call(prevMultiple, Source, Multiple);
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, int, Q> findNSB(vec<L, T, Q> const& Source, vec<L, int, Q> SignificantBitCount)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'findNSB' only accept integer inputs");
+
+		return detail::functor2_vec_int<L, T, Q>::call(findNSB, Source, SignificantBitCount);
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_packing.hpp

@@ -0,0 +1,32 @@
+/// @ref ext_vector_packing
+/// @file glm/ext/vector_packing.hpp
+///
+/// @see core (dependence)
+///
+/// @defgroup ext_vector_packing GLM_EXT_vector_packing
+/// @ingroup ext
+///
+/// Include <glm/ext/vector_packing.hpp> to use the features of this extension.
+///
+/// This extension provides a set of function to convert vectors to packed
+/// formats.
+
+#pragma once
+
+// Dependency:
+#include "../detail/qualifier.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_packing extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_packing
+	/// @{
+
+
+	/// @}
+}// namespace glm
+
+#include "vector_packing.inl"

cuda_extensions/third_party/glm/glm/ext/vector_reciprocal.hpp

@@ -0,0 +1,135 @@
+/// @ref ext_vector_reciprocal
+/// @file glm/ext/vector_reciprocal.hpp
+///
+/// @see core (dependence)
+///
+/// @defgroup ext_vector_reciprocal GLM_EXT_vector_reciprocal
+/// @ingroup ext
+///
+/// Include <glm/ext/vector_reciprocal.hpp> to use the features of this extension.
+///
+/// Define secant, cosecant and cotangent functions.
+
+#pragma once
+
+// Dependencies
+#include "../detail/setup.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_reciprocal extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_reciprocal
+	/// @{
+
+	/// Secant function.
+	/// hypotenuse / adjacent or 1 / cos(x)
+	///
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType sec(genType angle);
+
+	/// Cosecant function.
+	/// hypotenuse / opposite or 1 / sin(x)
+	///
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType csc(genType angle);
+
+	/// Cotangent function.
+	/// adjacent / opposite or 1 / tan(x)
+	///
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType cot(genType angle);
+
+	/// Inverse secant function.
+	///
+	/// @return Return an angle expressed in radians.
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType asec(genType x);
+
+	/// Inverse cosecant function.
+	///
+	/// @return Return an angle expressed in radians.
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType acsc(genType x);
+
+	/// Inverse cotangent function.
+	///
+	/// @return Return an angle expressed in radians.
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType acot(genType x);
+
+	/// Secant hyperbolic function.
+	///
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType sech(genType angle);
+
+	/// Cosecant hyperbolic function.
+	///
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType csch(genType angle);
+
+	/// Cotangent hyperbolic function.
+	///
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType coth(genType angle);
+
+	/// Inverse secant hyperbolic function.
+	///
+	/// @return Return an angle expressed in radians.
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType asech(genType x);
+
+	/// Inverse cosecant hyperbolic function.
+	///
+	/// @return Return an angle expressed in radians.
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType acsch(genType x);
+
+	/// Inverse cotangent hyperbolic function.
+	///
+	/// @return Return an angle expressed in radians.
+	/// @tparam genType Floating-point scalar or vector types.
+	///
+	/// @see ext_vector_reciprocal
+	template<typename genType>
+	GLM_FUNC_DECL genType acoth(genType x);
+
+	/// @}
+}//namespace glm
+
+#include "vector_reciprocal.inl"

cuda_extensions/third_party/glm/glm/ext/vector_reciprocal.inl

@@ -0,0 +1,105 @@
+/// @ref ext_vector_reciprocal
+
+#include "../trigonometric.hpp"
+#include <limits>
+
+namespace glm
+{
+	// sec
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> sec(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'sec' only accept floating-point inputs");
+		return static_cast<T>(1) / detail::functor1<vec, L, T, T, Q>::call(cos, x);
+	}
+
+	// csc
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> csc(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'csc' only accept floating-point inputs");
+		return static_cast<T>(1) / detail::functor1<vec, L, T, T, Q>::call(sin, x);
+	}
+
+	// cot
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> cot(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'cot' only accept floating-point inputs");
+		T const pi_over_2 = static_cast<T>(3.1415926535897932384626433832795 / 2.0);
+		return detail::functor1<vec, L, T, T, Q>::call(tan, pi_over_2 - x);
+	}
+
+	// asec
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> asec(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'asec' only accept floating-point inputs");
+		return detail::functor1<vec, L, T, T, Q>::call(acos, static_cast<T>(1) / x);
+	}
+
+	// acsc
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> acsc(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'acsc' only accept floating-point inputs");
+		return detail::functor1<vec, L, T, T, Q>::call(asin, static_cast<T>(1) / x);
+	}
+
+	// acot
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> acot(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'acot' only accept floating-point inputs");
+		T const pi_over_2 = static_cast<T>(3.1415926535897932384626433832795 / 2.0);
+		return pi_over_2 - detail::functor1<vec, L, T, T, Q>::call(atan, x);
+	}
+
+	// sech
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> sech(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'sech' only accept floating-point inputs");
+		return static_cast<T>(1) / detail::functor1<vec, L, T, T, Q>::call(cosh, x);
+	}
+
+	// csch
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> csch(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'csch' only accept floating-point inputs");
+		return static_cast<T>(1) / detail::functor1<vec, L, T, T, Q>::call(sinh, x);
+	}
+
+	// coth
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> coth(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'coth' only accept floating-point inputs");
+		return glm::cosh(x) / glm::sinh(x);
+	}
+
+	// asech
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> asech(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'asech' only accept floating-point inputs");
+		return detail::functor1<vec, L, T, T, Q>::call(acosh, static_cast<T>(1) / x);
+	}
+
+	// acsch
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> acsch(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'acsch' only accept floating-point inputs");
+		return detail::functor1<vec, L, T, T, Q>::call(asinh, static_cast<T>(1) / x);
+	}
+
+	// acoth
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> acoth(vec<L, T, Q> const& x)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_iec559, "'acoth' only accept floating-point inputs");
+		return detail::functor1<vec, L, T, T, Q>::call(atanh, static_cast<T>(1) / x);
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_relational.hpp

@@ -0,0 +1,107 @@
+/// @ref ext_vector_relational
+/// @file glm/ext/vector_relational.hpp
+///
+/// @see core (dependence)
+/// @see ext_scalar_integer (dependence)
+///
+/// @defgroup ext_vector_relational GLM_EXT_vector_relational
+/// @ingroup ext
+///
+/// Exposes comparison functions for vector types that take a user defined epsilon values.
+///
+/// Include <glm/ext/vector_relational.hpp> to use the features of this extension.
+///
+/// @see core_vector_relational
+/// @see ext_scalar_relational
+/// @see ext_matrix_relational
+
+#pragma once
+
+// Dependencies
+#include "../detail/qualifier.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_relational extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_relational
+	/// @{
+
+	/// Returns the component-wise comparison of |x - y| < epsilon.
+	/// True if this expression is satisfied.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point or integer scalar types
+	/// @tparam Q Value from qualifier enum
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<L, bool, Q> equal(vec<L, T, Q> const& x, vec<L, T, Q> const& y, T epsilon);
+
+	/// Returns the component-wise comparison of |x - y| < epsilon.
+	/// True if this expression is satisfied.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point or integer scalar types
+	/// @tparam Q Value from qualifier enum
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<L, bool, Q> equal(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, T, Q> const& epsilon);
+
+	/// Returns the component-wise comparison of |x - y| >= epsilon.
+	/// True if this expression is not satisfied.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point or integer scalar types
+	/// @tparam Q Value from qualifier enum
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<L, bool, Q> notEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, T epsilon);
+
+	/// Returns the component-wise comparison of |x - y| >= epsilon.
+	/// True if this expression is not satisfied.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point or integer scalar types
+	/// @tparam Q Value from qualifier enum
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<L, bool, Q> notEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, T, Q> const& epsilon);
+
+	/// Returns the component-wise comparison between two vectors in term of ULPs.
+	/// True if this expression is satisfied.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<L, bool, Q> equal(vec<L, T, Q> const& x, vec<L, T, Q> const& y, int ULPs);
+
+	/// Returns the component-wise comparison between two vectors in term of ULPs.
+	/// True if this expression is satisfied.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<L, bool, Q> equal(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, int, Q> const& ULPs);
+
+	/// Returns the component-wise comparison between two vectors in term of ULPs.
+	/// True if this expression is not satisfied.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<L, bool, Q> notEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, int ULPs);
+
+	/// Returns the component-wise comparison between two vectors in term of ULPs.
+	/// True if this expression is not satisfied.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<L, bool, Q> notEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, int, Q> const& ULPs);
+
+	/// @}
+}//namespace glm
+
+#include "vector_relational.inl"

cuda_extensions/third_party/glm/glm/ext/vector_relational.inl

@@ -0,0 +1,75 @@
+#include "../vector_relational.hpp"
+#include "../common.hpp"
+#include "../detail/qualifier.hpp"
+#include "../detail/type_float.hpp"
+
+namespace glm
+{
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR vec<L, bool, Q> equal(vec<L, T, Q> const& x, vec<L, T, Q> const& y, T Epsilon)
+	{
+		return equal(x, y, vec<L, T, Q>(Epsilon));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR vec<L, bool, Q> equal(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, T, Q> const& Epsilon)
+	{
+		return lessThanEqual(abs(x - y), Epsilon);
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR vec<L, bool, Q> notEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, T Epsilon)
+	{
+		return notEqual(x, y, vec<L, T, Q>(Epsilon));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR vec<L, bool, Q> notEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, T, Q> const& Epsilon)
+	{
+		return greaterThan(abs(x - y), Epsilon);
+	}
+
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR vec<L, bool, Q> equal(vec<L, T, Q> const& x, vec<L, T, Q> const& y, int MaxULPs)
+	{
+		return equal(x, y, vec<L, int, Q>(MaxULPs));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR vec<L, bool, Q> equal(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, int, Q> const& MaxULPs)
+	{
+		vec<L, bool, Q> Result(false);
+		for(length_t i = 0; i < L; ++i)
+		{
+			detail::float_t<T> const a(x[i]);
+			detail::float_t<T> const b(y[i]);
+
+			// Different signs means they do not match.
+			if(a.negative() != b.negative())
+			{
+				// Check for equality to make sure +0==-0
+				Result[i] = a.mantissa() == b.mantissa() && a.exponent() == b.exponent();
+			}
+			else
+			{
+				// Find the difference in ULPs.
+				typename detail::float_t<T>::int_type const DiffULPs = abs(a.i - b.i);
+				Result[i] = DiffULPs <= MaxULPs[i];
+			}
+		}
+		return Result;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR vec<L, bool, Q> notEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, int MaxULPs)
+	{
+		return notEqual(x, y, vec<L, int, Q>(MaxULPs));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR vec<L, bool, Q> notEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, int, Q> const& MaxULPs)
+	{
+		return not_(equal(x, y, MaxULPs));
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_uint1.hpp

@@ -0,0 +1,32 @@
+/// @ref ext_vector_uint1
+/// @file glm/ext/vector_uint1.hpp
+///
+/// @defgroup ext_vector_uint1 GLM_EXT_vector_uint1
+/// @ingroup ext
+///
+/// Exposes uvec1 vector type.
+///
+/// Include <glm/ext/vector_uvec1.hpp> to use the features of this extension.
+///
+/// @see ext_vector_int1 extension.
+/// @see ext_vector_uint1_precision extension.
+
+#pragma once
+
+#include "../detail/type_vec1.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_uint1 extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_uint1
+	/// @{
+
+	/// 1 component vector of unsigned integer numbers.
+	typedef vec<1, unsigned int, defaultp>			uvec1;
+
+	/// @}
+}//namespace glm
+

cuda_extensions/third_party/glm/glm/ext/vector_uint1_sized.hpp

@@ -0,0 +1,49 @@
+/// @ref ext_vector_uint1_sized
+/// @file glm/ext/vector_uint1_sized.hpp
+///
+/// @defgroup ext_vector_uint1_sized GLM_EXT_vector_uint1_sized
+/// @ingroup ext
+///
+/// Exposes sized unsigned integer vector types.
+///
+/// Include <glm/ext/vector_uint1_sized.hpp> to use the features of this extension.
+///
+/// @see ext_scalar_uint_sized
+/// @see ext_vector_int1_sized
+
+#pragma once
+
+#include "../ext/vector_uint1.hpp"
+#include "../ext/scalar_uint_sized.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_uint1_sized extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_uint1_sized
+	/// @{
+
+	/// 8 bit unsigned integer vector of 1 component type.
+	///
+	/// @see ext_vector_uint1_sized
+	typedef vec<1, uint8, defaultp>		u8vec1;
+
+	/// 16 bit unsigned integer vector of 1 component type.
+	///
+	/// @see ext_vector_uint1_sized
+	typedef vec<1, uint16, defaultp>	u16vec1;
+
+	/// 32 bit unsigned integer vector of 1 component type.
+	///
+	/// @see ext_vector_uint1_sized
+	typedef vec<1, uint32, defaultp>	u32vec1;
+
+	/// 64 bit unsigned integer vector of 1 component type.
+	///
+	/// @see ext_vector_uint1_sized
+	typedef vec<1, uint64, defaultp>	u64vec1;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_uint2.hpp

@@ -0,0 +1,18 @@
+/// @ref core
+/// @file glm/ext/vector_uint2.hpp
+
+#pragma once
+#include "../detail/type_vec2.hpp"
+
+namespace glm
+{
+	/// @addtogroup core_vector
+	/// @{
+
+	/// 2 components vector of unsigned integer numbers.
+	///
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 4.1.5 Vectors</a>
+	typedef vec<2, unsigned int, defaultp>		uvec2;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_uint2_sized.hpp

@@ -0,0 +1,49 @@
+/// @ref ext_vector_uint2_sized
+/// @file glm/ext/vector_uint2_sized.hpp
+///
+/// @defgroup ext_vector_uint2_sized GLM_EXT_vector_uint2_sized
+/// @ingroup ext
+///
+/// Exposes sized unsigned integer vector of 2 components type.
+///
+/// Include <glm/ext/vector_uint2_sized.hpp> to use the features of this extension.
+///
+/// @see ext_scalar_uint_sized
+/// @see ext_vector_int2_sized
+
+#pragma once
+
+#include "../ext/vector_uint2.hpp"
+#include "../ext/scalar_uint_sized.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_uint2_sized extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_uint2_sized
+	/// @{
+
+	/// 8 bit unsigned integer vector of 2 components type.
+	///
+	/// @see ext_vector_uint2_sized
+	typedef vec<2, uint8, defaultp>		u8vec2;
+
+	/// 16 bit unsigned integer vector of 2 components type.
+	///
+	/// @see ext_vector_uint2_sized
+	typedef vec<2, uint16, defaultp>	u16vec2;
+
+	/// 32 bit unsigned integer vector of 2 components type.
+	///
+	/// @see ext_vector_uint2_sized
+	typedef vec<2, uint32, defaultp>	u32vec2;
+
+	/// 64 bit unsigned integer vector of 2 components type.
+	///
+	/// @see ext_vector_uint2_sized
+	typedef vec<2, uint64, defaultp>	u64vec2;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_uint3.hpp

@@ -0,0 +1,18 @@
+/// @ref core
+/// @file glm/ext/vector_uint3.hpp
+
+#pragma once
+#include "../detail/type_vec3.hpp"
+
+namespace glm
+{
+	/// @addtogroup core_vector
+	/// @{
+
+	/// 3 components vector of unsigned integer numbers.
+	///
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 4.1.5 Vectors</a>
+	typedef vec<3, unsigned int, defaultp>		uvec3;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_uint3_sized.hpp

@@ -0,0 +1,49 @@
+/// @ref ext_vector_uint3_sized
+/// @file glm/ext/vector_uint3_sized.hpp
+///
+/// @defgroup ext_vector_uint3_sized GLM_EXT_vector_uint3_sized
+/// @ingroup ext
+///
+/// Exposes sized unsigned integer vector of 3 components type.
+///
+/// Include <glm/ext/vector_uint3_sized.hpp> to use the features of this extension.
+///
+/// @see ext_scalar_uint_sized
+/// @see ext_vector_int3_sized
+
+#pragma once
+
+#include "../ext/vector_uint3.hpp"
+#include "../ext/scalar_uint_sized.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_uint3_sized extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_uint3_sized
+	/// @{
+
+	/// 8 bit unsigned integer vector of 3 components type.
+	///
+	/// @see ext_vector_uint3_sized
+	typedef vec<3, uint8, defaultp>		u8vec3;
+
+	/// 16 bit unsigned integer vector of 3 components type.
+	///
+	/// @see ext_vector_uint3_sized
+	typedef vec<3, uint16, defaultp>	u16vec3;
+
+	/// 32 bit unsigned integer vector of 3 components type.
+	///
+	/// @see ext_vector_uint3_sized
+	typedef vec<3, uint32, defaultp>	u32vec3;
+
+	/// 64 bit unsigned integer vector of 3 components type.
+	///
+	/// @see ext_vector_uint3_sized
+	typedef vec<3, uint64, defaultp>	u64vec3;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_uint4.hpp

@@ -0,0 +1,18 @@
+/// @ref core
+/// @file glm/ext/vector_uint4.hpp
+
+#pragma once
+#include "../detail/type_vec4.hpp"
+
+namespace glm
+{
+	/// @addtogroup core_vector
+	/// @{
+
+	/// 4 components vector of unsigned integer numbers.
+	///
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 4.1.5 Vectors</a>
+	typedef vec<4, unsigned int, defaultp>		uvec4;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_uint4_sized.hpp

@@ -0,0 +1,49 @@
+/// @ref ext_vector_uint4_sized
+/// @file glm/ext/vector_uint4_sized.hpp
+///
+/// @defgroup ext_vector_uint4_sized GLM_EXT_vector_uint4_sized
+/// @ingroup ext
+///
+/// Exposes sized unsigned integer vector of 4 components type.
+///
+/// Include <glm/ext/vector_uint4_sized.hpp> to use the features of this extension.
+///
+/// @see ext_scalar_uint_sized
+/// @see ext_vector_int4_sized
+
+#pragma once
+
+#include "../ext/vector_uint4.hpp"
+#include "../ext/scalar_uint_sized.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_uint4_sized extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_uint4_sized
+	/// @{
+
+	/// 8 bit unsigned integer vector of 4 components type.
+	///
+	/// @see ext_vector_uint4_sized
+	typedef vec<4, uint8, defaultp>		u8vec4;
+
+	/// 16 bit unsigned integer vector of 4 components type.
+	///
+	/// @see ext_vector_uint4_sized
+	typedef vec<4, uint16, defaultp>	u16vec4;
+
+	/// 32 bit unsigned integer vector of 4 components type.
+	///
+	/// @see ext_vector_uint4_sized
+	typedef vec<4, uint32, defaultp>	u32vec4;
+
+	/// 64 bit unsigned integer vector of 4 components type.
+	///
+	/// @see ext_vector_uint4_sized
+	typedef vec<4, uint64, defaultp>	u64vec4;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/ext/vector_ulp.hpp

@@ -0,0 +1,112 @@
+/// @ref ext_vector_ulp
+/// @file glm/ext/vector_ulp.hpp
+///
+/// @defgroup ext_vector_ulp GLM_EXT_vector_ulp
+/// @ingroup ext
+///
+/// Allow the measurement of the accuracy of a function against a reference
+/// implementation. This extension works on floating-point data and provide results
+/// in ULP.
+///
+/// Include <glm/ext/vector_ulp.hpp> to use the features of this extension.
+///
+/// @see ext_scalar_ulp
+/// @see ext_scalar_relational
+/// @see ext_vector_relational
+
+#pragma once
+
+// Dependencies
+#include "../ext/scalar_ulp.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_EXT_vector_ulp extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup ext_vector_ulp
+	/// @{
+
+	/// Return the next ULP value(s) after the input value(s).
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_scalar_ulp
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> nextFloat(vec<L, T, Q> const& x);
+
+	/// Return the value(s) ULP distance after the input value(s).
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_scalar_ulp
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> nextFloat(vec<L, T, Q> const& x, int ULPs);
+
+	/// Return the value(s) ULP distance after the input value(s).
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_scalar_ulp
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> nextFloat(vec<L, T, Q> const& x, vec<L, int, Q> const& ULPs);
+
+	/// Return the previous ULP value(s) before the input value(s).
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_scalar_ulp
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> prevFloat(vec<L, T, Q> const& x);
+
+	/// Return the value(s) ULP distance before the input value(s).
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_scalar_ulp
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> prevFloat(vec<L, T, Q> const& x, int ULPs);
+
+	/// Return the value(s) ULP distance before the input value(s).
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Floating-point
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_scalar_ulp
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> prevFloat(vec<L, T, Q> const& x, vec<L, int, Q> const& ULPs);
+
+	/// Return the distance in the number of ULP between 2 single-precision floating-point scalars.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_scalar_ulp
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, int, Q> floatDistance(vec<L, float, Q> const& x, vec<L, float, Q> const& y);
+
+	/// Return the distance in the number of ULP between 2 double-precision floating-point scalars.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_scalar_ulp
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, int64, Q> floatDistance(vec<L, double, Q> const& x, vec<L, double, Q> const& y);
+
+	/// @}
+}//namespace glm
+
+#include "vector_ulp.inl"

cuda_extensions/third_party/glm/glm/ext/vector_ulp.inl

@@ -0,0 +1,74 @@
+namespace glm
+{
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> nextFloat(vec<L, T, Q> const& x)
+	{
+		vec<L, T, Q> Result;
+		for(length_t i = 0, n = Result.length(); i < n; ++i)
+			Result[i] = nextFloat(x[i]);
+		return Result;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> nextFloat(vec<L, T, Q> const& x, int ULPs)
+	{
+		vec<L, T, Q> Result;
+		for(length_t i = 0, n = Result.length(); i < n; ++i)
+			Result[i] = nextFloat(x[i], ULPs);
+		return Result;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> nextFloat(vec<L, T, Q> const& x, vec<L, int, Q> const& ULPs)
+	{
+		vec<L, T, Q> Result;
+		for(length_t i = 0, n = Result.length(); i < n; ++i)
+			Result[i] = nextFloat(x[i], ULPs[i]);
+		return Result;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> prevFloat(vec<L, T, Q> const& x)
+	{
+		vec<L, T, Q> Result;
+		for(length_t i = 0, n = Result.length(); i < n; ++i)
+			Result[i] = prevFloat(x[i]);
+		return Result;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> prevFloat(vec<L, T, Q> const& x, int ULPs)
+	{
+		vec<L, T, Q> Result;
+		for(length_t i = 0, n = Result.length(); i < n; ++i)
+			Result[i] = prevFloat(x[i], ULPs);
+		return Result;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> prevFloat(vec<L, T, Q> const& x, vec<L, int, Q> const& ULPs)
+	{
+		vec<L, T, Q> Result;
+		for(length_t i = 0, n = Result.length(); i < n; ++i)
+			Result[i] = prevFloat(x[i], ULPs[i]);
+		return Result;
+	}
+
+	template<length_t L, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, int, Q> floatDistance(vec<L, float, Q> const& x, vec<L, float, Q> const& y)
+	{
+		vec<L, int, Q> Result;
+		for(length_t i = 0, n = Result.length(); i < n; ++i)
+			Result[i] = floatDistance(x[i], y[i]);
+		return Result;
+	}
+
+	template<length_t L, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, int64, Q> floatDistance(vec<L, double, Q> const& x, vec<L, double, Q> const& y)
+	{
+		vec<L, int64, Q> Result;
+		for(length_t i = 0, n = Result.length(); i < n; ++i)
+			Result[i] = floatDistance(x[i], y[i]);
+		return Result;
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/fwd.hpp

@@ -0,0 +1,1233 @@
+#pragma once
+
+#include "detail/qualifier.hpp"
+
+namespace glm
+{
+#if GLM_HAS_EXTENDED_INTEGER_TYPE
+	typedef std::int8_t				int8;
+	typedef std::int16_t			int16;
+	typedef std::int32_t			int32;
+	typedef std::int64_t			int64;
+
+	typedef std::uint8_t			uint8;
+	typedef std::uint16_t			uint16;
+	typedef std::uint32_t			uint32;
+	typedef std::uint64_t			uint64;
+#else
+	typedef signed char				int8;
+	typedef signed short			int16;
+	typedef signed int				int32;
+	typedef detail::int64			int64;
+
+	typedef unsigned char			uint8;
+	typedef unsigned short			uint16;
+	typedef unsigned int			uint32;
+	typedef detail::uint64			uint64;
+#endif
+
+	// Scalar int
+
+	typedef int8					lowp_i8;
+	typedef int8					mediump_i8;
+	typedef int8					highp_i8;
+	typedef int8					i8;
+
+	typedef int8					lowp_int8;
+	typedef int8					mediump_int8;
+	typedef int8					highp_int8;
+
+	typedef int8					lowp_int8_t;
+	typedef int8					mediump_int8_t;
+	typedef int8					highp_int8_t;
+	typedef int8					int8_t;
+
+	typedef int16					lowp_i16;
+	typedef int16					mediump_i16;
+	typedef int16					highp_i16;
+	typedef int16					i16;
+
+	typedef int16					lowp_int16;
+	typedef int16					mediump_int16;
+	typedef int16					highp_int16;
+
+	typedef int16					lowp_int16_t;
+	typedef int16					mediump_int16_t;
+	typedef int16					highp_int16_t;
+	typedef int16					int16_t;
+
+	typedef int32					lowp_i32;
+	typedef int32					mediump_i32;
+	typedef int32					highp_i32;
+	typedef int32					i32;
+
+	typedef int32					lowp_int32;
+	typedef int32					mediump_int32;
+	typedef int32					highp_int32;
+
+	typedef int32					lowp_int32_t;
+	typedef int32					mediump_int32_t;
+	typedef int32					highp_int32_t;
+	typedef int32					int32_t;
+
+	typedef int64					lowp_i64;
+	typedef int64					mediump_i64;
+	typedef int64					highp_i64;
+	typedef int64					i64;
+
+	typedef int64					lowp_int64;
+	typedef int64					mediump_int64;
+	typedef int64					highp_int64;
+
+	typedef int64					lowp_int64_t;
+	typedef int64					mediump_int64_t;
+	typedef int64					highp_int64_t;
+	typedef int64					int64_t;
+
+	// Scalar uint
+
+	typedef unsigned int			uint;
+
+	typedef uint8					lowp_u8;
+	typedef uint8					mediump_u8;
+	typedef uint8					highp_u8;
+	typedef uint8					u8;
+
+	typedef uint8					lowp_uint8;
+	typedef uint8					mediump_uint8;
+	typedef uint8					highp_uint8;
+
+	typedef uint8					lowp_uint8_t;
+	typedef uint8					mediump_uint8_t;
+	typedef uint8					highp_uint8_t;
+	typedef uint8					uint8_t;
+
+	typedef uint16					lowp_u16;
+	typedef uint16					mediump_u16;
+	typedef uint16					highp_u16;
+	typedef uint16					u16;
+
+	typedef uint16					lowp_uint16;
+	typedef uint16					mediump_uint16;
+	typedef uint16					highp_uint16;
+
+	typedef uint16					lowp_uint16_t;
+	typedef uint16					mediump_uint16_t;
+	typedef uint16					highp_uint16_t;
+	typedef uint16					uint16_t;
+
+	typedef uint32					lowp_u32;
+	typedef uint32					mediump_u32;
+	typedef uint32					highp_u32;
+	typedef uint32					u32;
+
+	typedef uint32					lowp_uint32;
+	typedef uint32					mediump_uint32;
+	typedef uint32					highp_uint32;
+
+	typedef uint32					lowp_uint32_t;
+	typedef uint32					mediump_uint32_t;
+	typedef uint32					highp_uint32_t;
+	typedef uint32					uint32_t;
+
+	typedef uint64					lowp_u64;
+	typedef uint64					mediump_u64;
+	typedef uint64					highp_u64;
+	typedef uint64					u64;
+
+	typedef uint64					lowp_uint64;
+	typedef uint64					mediump_uint64;
+	typedef uint64					highp_uint64;
+
+	typedef uint64					lowp_uint64_t;
+	typedef uint64					mediump_uint64_t;
+	typedef uint64					highp_uint64_t;
+	typedef uint64					uint64_t;
+
+	// Scalar float
+
+	typedef float					lowp_f32;
+	typedef float					mediump_f32;
+	typedef float					highp_f32;
+	typedef float					f32;
+
+	typedef float					lowp_float32;
+	typedef float					mediump_float32;
+	typedef float					highp_float32;
+	typedef float					float32;
+
+	typedef float					lowp_float32_t;
+	typedef float					mediump_float32_t;
+	typedef float					highp_float32_t;
+	typedef float					float32_t;
+
+
+	typedef double					lowp_f64;
+	typedef double					mediump_f64;
+	typedef double					highp_f64;
+	typedef double					f64;
+
+	typedef double					lowp_float64;
+	typedef double					mediump_float64;
+	typedef double					highp_float64;
+	typedef double					float64;
+
+	typedef double					lowp_float64_t;
+	typedef double					mediump_float64_t;
+	typedef double					highp_float64_t;
+	typedef double					float64_t;
+
+	// Vector bool
+
+	typedef vec<1, bool, lowp>		lowp_bvec1;
+	typedef vec<2, bool, lowp>		lowp_bvec2;
+	typedef vec<3, bool, lowp>		lowp_bvec3;
+	typedef vec<4, bool, lowp>		lowp_bvec4;
+
+	typedef vec<1, bool, mediump>	mediump_bvec1;
+	typedef vec<2, bool, mediump>	mediump_bvec2;
+	typedef vec<3, bool, mediump>	mediump_bvec3;
+	typedef vec<4, bool, mediump>	mediump_bvec4;
+
+	typedef vec<1, bool, highp>		highp_bvec1;
+	typedef vec<2, bool, highp>		highp_bvec2;
+	typedef vec<3, bool, highp>		highp_bvec3;
+	typedef vec<4, bool, highp>		highp_bvec4;
+
+	typedef vec<1, bool, defaultp>	bvec1;
+	typedef vec<2, bool, defaultp>	bvec2;
+	typedef vec<3, bool, defaultp>	bvec3;
+	typedef vec<4, bool, defaultp>	bvec4;
+
+	// Vector int
+
+	typedef vec<1, int, lowp>		lowp_ivec1;
+	typedef vec<2, int, lowp>		lowp_ivec2;
+	typedef vec<3, int, lowp>		lowp_ivec3;
+	typedef vec<4, int, lowp>		lowp_ivec4;
+
+	typedef vec<1, int, mediump>	mediump_ivec1;
+	typedef vec<2, int, mediump>	mediump_ivec2;
+	typedef vec<3, int, mediump>	mediump_ivec3;
+	typedef vec<4, int, mediump>	mediump_ivec4;
+
+	typedef vec<1, int, highp>		highp_ivec1;
+	typedef vec<2, int, highp>		highp_ivec2;
+	typedef vec<3, int, highp>		highp_ivec3;
+	typedef vec<4, int, highp>		highp_ivec4;
+
+	typedef vec<1, int, defaultp>	ivec1;
+	typedef vec<2, int, defaultp>	ivec2;
+	typedef vec<3, int, defaultp>	ivec3;
+	typedef vec<4, int, defaultp>	ivec4;
+
+	typedef vec<1, i8, lowp>		lowp_i8vec1;
+	typedef vec<2, i8, lowp>		lowp_i8vec2;
+	typedef vec<3, i8, lowp>		lowp_i8vec3;
+	typedef vec<4, i8, lowp>		lowp_i8vec4;
+
+	typedef vec<1, i8, mediump>		mediump_i8vec1;
+	typedef vec<2, i8, mediump>		mediump_i8vec2;
+	typedef vec<3, i8, mediump>		mediump_i8vec3;
+	typedef vec<4, i8, mediump>		mediump_i8vec4;
+
+	typedef vec<1, i8, highp>		highp_i8vec1;
+	typedef vec<2, i8, highp>		highp_i8vec2;
+	typedef vec<3, i8, highp>		highp_i8vec3;
+	typedef vec<4, i8, highp>		highp_i8vec4;
+
+	typedef vec<1, i8, defaultp>	i8vec1;
+	typedef vec<2, i8, defaultp>	i8vec2;
+	typedef vec<3, i8, defaultp>	i8vec3;
+	typedef vec<4, i8, defaultp>	i8vec4;
+
+	typedef vec<1, i16, lowp>		lowp_i16vec1;
+	typedef vec<2, i16, lowp>		lowp_i16vec2;
+	typedef vec<3, i16, lowp>		lowp_i16vec3;
+	typedef vec<4, i16, lowp>		lowp_i16vec4;
+
+	typedef vec<1, i16, mediump>	mediump_i16vec1;
+	typedef vec<2, i16, mediump>	mediump_i16vec2;
+	typedef vec<3, i16, mediump>	mediump_i16vec3;
+	typedef vec<4, i16, mediump>	mediump_i16vec4;
+
+	typedef vec<1, i16, highp>		highp_i16vec1;
+	typedef vec<2, i16, highp>		highp_i16vec2;
+	typedef vec<3, i16, highp>		highp_i16vec3;
+	typedef vec<4, i16, highp>		highp_i16vec4;
+
+	typedef vec<1, i16, defaultp>	i16vec1;
+	typedef vec<2, i16, defaultp>	i16vec2;
+	typedef vec<3, i16, defaultp>	i16vec3;
+	typedef vec<4, i16, defaultp>	i16vec4;
+
+	typedef vec<1, i32, lowp>		lowp_i32vec1;
+	typedef vec<2, i32, lowp>		lowp_i32vec2;
+	typedef vec<3, i32, lowp>		lowp_i32vec3;
+	typedef vec<4, i32, lowp>		lowp_i32vec4;
+
+	typedef vec<1, i32, mediump>	mediump_i32vec1;
+	typedef vec<2, i32, mediump>	mediump_i32vec2;
+	typedef vec<3, i32, mediump>	mediump_i32vec3;
+	typedef vec<4, i32, mediump>	mediump_i32vec4;
+
+	typedef vec<1, i32, highp>		highp_i32vec1;
+	typedef vec<2, i32, highp>		highp_i32vec2;
+	typedef vec<3, i32, highp>		highp_i32vec3;
+	typedef vec<4, i32, highp>		highp_i32vec4;
+
+	typedef vec<1, i32, defaultp>	i32vec1;
+	typedef vec<2, i32, defaultp>	i32vec2;
+	typedef vec<3, i32, defaultp>	i32vec3;
+	typedef vec<4, i32, defaultp>	i32vec4;
+
+	typedef vec<1, i64, lowp>		lowp_i64vec1;
+	typedef vec<2, i64, lowp>		lowp_i64vec2;
+	typedef vec<3, i64, lowp>		lowp_i64vec3;
+	typedef vec<4, i64, lowp>		lowp_i64vec4;
+
+	typedef vec<1, i64, mediump>	mediump_i64vec1;
+	typedef vec<2, i64, mediump>	mediump_i64vec2;
+	typedef vec<3, i64, mediump>	mediump_i64vec3;
+	typedef vec<4, i64, mediump>	mediump_i64vec4;
+
+	typedef vec<1, i64, highp>		highp_i64vec1;
+	typedef vec<2, i64, highp>		highp_i64vec2;
+	typedef vec<3, i64, highp>		highp_i64vec3;
+	typedef vec<4, i64, highp>		highp_i64vec4;
+
+	typedef vec<1, i64, defaultp>	i64vec1;
+	typedef vec<2, i64, defaultp>	i64vec2;
+	typedef vec<3, i64, defaultp>	i64vec3;
+	typedef vec<4, i64, defaultp>	i64vec4;
+
+	// Vector uint
+
+	typedef vec<1, uint, lowp>		lowp_uvec1;
+	typedef vec<2, uint, lowp>		lowp_uvec2;
+	typedef vec<3, uint, lowp>		lowp_uvec3;
+	typedef vec<4, uint, lowp>		lowp_uvec4;
+
+	typedef vec<1, uint, mediump>	mediump_uvec1;
+	typedef vec<2, uint, mediump>	mediump_uvec2;
+	typedef vec<3, uint, mediump>	mediump_uvec3;
+	typedef vec<4, uint, mediump>	mediump_uvec4;
+
+	typedef vec<1, uint, highp>		highp_uvec1;
+	typedef vec<2, uint, highp>		highp_uvec2;
+	typedef vec<3, uint, highp>		highp_uvec3;
+	typedef vec<4, uint, highp>		highp_uvec4;
+
+	typedef vec<1, uint, defaultp>	uvec1;
+	typedef vec<2, uint, defaultp>	uvec2;
+	typedef vec<3, uint, defaultp>	uvec3;
+	typedef vec<4, uint, defaultp>	uvec4;
+
+	typedef vec<1, u8, lowp>		lowp_u8vec1;
+	typedef vec<2, u8, lowp>		lowp_u8vec2;
+	typedef vec<3, u8, lowp>		lowp_u8vec3;
+	typedef vec<4, u8, lowp>		lowp_u8vec4;
+
+	typedef vec<1, u8, mediump>		mediump_u8vec1;
+	typedef vec<2, u8, mediump>		mediump_u8vec2;
+	typedef vec<3, u8, mediump>		mediump_u8vec3;
+	typedef vec<4, u8, mediump>		mediump_u8vec4;
+
+	typedef vec<1, u8, highp>		highp_u8vec1;
+	typedef vec<2, u8, highp>		highp_u8vec2;
+	typedef vec<3, u8, highp>		highp_u8vec3;
+	typedef vec<4, u8, highp>		highp_u8vec4;
+
+	typedef vec<1, u8, defaultp>	u8vec1;
+	typedef vec<2, u8, defaultp>	u8vec2;
+	typedef vec<3, u8, defaultp>	u8vec3;
+	typedef vec<4, u8, defaultp>	u8vec4;
+
+	typedef vec<1, u16, lowp>		lowp_u16vec1;
+	typedef vec<2, u16, lowp>		lowp_u16vec2;
+	typedef vec<3, u16, lowp>		lowp_u16vec3;
+	typedef vec<4, u16, lowp>		lowp_u16vec4;
+
+	typedef vec<1, u16, mediump>	mediump_u16vec1;
+	typedef vec<2, u16, mediump>	mediump_u16vec2;
+	typedef vec<3, u16, mediump>	mediump_u16vec3;
+	typedef vec<4, u16, mediump>	mediump_u16vec4;
+
+	typedef vec<1, u16, highp>		highp_u16vec1;
+	typedef vec<2, u16, highp>		highp_u16vec2;
+	typedef vec<3, u16, highp>		highp_u16vec3;
+	typedef vec<4, u16, highp>		highp_u16vec4;
+
+	typedef vec<1, u16, defaultp>	u16vec1;
+	typedef vec<2, u16, defaultp>	u16vec2;
+	typedef vec<3, u16, defaultp>	u16vec3;
+	typedef vec<4, u16, defaultp>	u16vec4;
+
+	typedef vec<1, u32, lowp>		lowp_u32vec1;
+	typedef vec<2, u32, lowp>		lowp_u32vec2;
+	typedef vec<3, u32, lowp>		lowp_u32vec3;
+	typedef vec<4, u32, lowp>		lowp_u32vec4;
+
+	typedef vec<1, u32, mediump>	mediump_u32vec1;
+	typedef vec<2, u32, mediump>	mediump_u32vec2;
+	typedef vec<3, u32, mediump>	mediump_u32vec3;
+	typedef vec<4, u32, mediump>	mediump_u32vec4;
+
+	typedef vec<1, u32, highp>		highp_u32vec1;
+	typedef vec<2, u32, highp>		highp_u32vec2;
+	typedef vec<3, u32, highp>		highp_u32vec3;
+	typedef vec<4, u32, highp>		highp_u32vec4;
+
+	typedef vec<1, u32, defaultp>	u32vec1;
+	typedef vec<2, u32, defaultp>	u32vec2;
+	typedef vec<3, u32, defaultp>	u32vec3;
+	typedef vec<4, u32, defaultp>	u32vec4;
+
+	typedef vec<1, u64, lowp>		lowp_u64vec1;
+	typedef vec<2, u64, lowp>		lowp_u64vec2;
+	typedef vec<3, u64, lowp>		lowp_u64vec3;
+	typedef vec<4, u64, lowp>		lowp_u64vec4;
+
+	typedef vec<1, u64, mediump>	mediump_u64vec1;
+	typedef vec<2, u64, mediump>	mediump_u64vec2;
+	typedef vec<3, u64, mediump>	mediump_u64vec3;
+	typedef vec<4, u64, mediump>	mediump_u64vec4;
+
+	typedef vec<1, u64, highp>		highp_u64vec1;
+	typedef vec<2, u64, highp>		highp_u64vec2;
+	typedef vec<3, u64, highp>		highp_u64vec3;
+	typedef vec<4, u64, highp>		highp_u64vec4;
+
+	typedef vec<1, u64, defaultp>	u64vec1;
+	typedef vec<2, u64, defaultp>	u64vec2;
+	typedef vec<3, u64, defaultp>	u64vec3;
+	typedef vec<4, u64, defaultp>	u64vec4;
+
+	// Vector float
+
+	typedef vec<1, float, lowp>			lowp_vec1;
+	typedef vec<2, float, lowp>			lowp_vec2;
+	typedef vec<3, float, lowp>			lowp_vec3;
+	typedef vec<4, float, lowp>			lowp_vec4;
+
+	typedef vec<1, float, mediump>		mediump_vec1;
+	typedef vec<2, float, mediump>		mediump_vec2;
+	typedef vec<3, float, mediump>		mediump_vec3;
+	typedef vec<4, float, mediump>		mediump_vec4;
+
+	typedef vec<1, float, highp>		highp_vec1;
+	typedef vec<2, float, highp>		highp_vec2;
+	typedef vec<3, float, highp>		highp_vec3;
+	typedef vec<4, float, highp>		highp_vec4;
+
+	typedef vec<1, float, defaultp>		vec1;
+	typedef vec<2, float, defaultp>		vec2;
+	typedef vec<3, float, defaultp>		vec3;
+	typedef vec<4, float, defaultp>		vec4;
+
+	typedef vec<1, float, lowp>			lowp_fvec1;
+	typedef vec<2, float, lowp>			lowp_fvec2;
+	typedef vec<3, float, lowp>			lowp_fvec3;
+	typedef vec<4, float, lowp>			lowp_fvec4;
+
+	typedef vec<1, float, mediump>		mediump_fvec1;
+	typedef vec<2, float, mediump>		mediump_fvec2;
+	typedef vec<3, float, mediump>		mediump_fvec3;
+	typedef vec<4, float, mediump>		mediump_fvec4;
+
+	typedef vec<1, float, highp>		highp_fvec1;
+	typedef vec<2, float, highp>		highp_fvec2;
+	typedef vec<3, float, highp>		highp_fvec3;
+	typedef vec<4, float, highp>		highp_fvec4;
+
+	typedef vec<1, f32, defaultp>		fvec1;
+	typedef vec<2, f32, defaultp>		fvec2;
+	typedef vec<3, f32, defaultp>		fvec3;
+	typedef vec<4, f32, defaultp>		fvec4;
+
+	typedef vec<1, f32, lowp>			lowp_f32vec1;
+	typedef vec<2, f32, lowp>			lowp_f32vec2;
+	typedef vec<3, f32, lowp>			lowp_f32vec3;
+	typedef vec<4, f32, lowp>			lowp_f32vec4;
+
+	typedef vec<1, f32, mediump>		mediump_f32vec1;
+	typedef vec<2, f32, mediump>		mediump_f32vec2;
+	typedef vec<3, f32, mediump>		mediump_f32vec3;
+	typedef vec<4, f32, mediump>		mediump_f32vec4;
+
+	typedef vec<1, f32, highp>			highp_f32vec1;
+	typedef vec<2, f32, highp>			highp_f32vec2;
+	typedef vec<3, f32, highp>			highp_f32vec3;
+	typedef vec<4, f32, highp>			highp_f32vec4;
+
+	typedef vec<1, f32, defaultp>		f32vec1;
+	typedef vec<2, f32, defaultp>		f32vec2;
+	typedef vec<3, f32, defaultp>		f32vec3;
+	typedef vec<4, f32, defaultp>		f32vec4;
+
+	typedef vec<1, f64, lowp>			lowp_dvec1;
+	typedef vec<2, f64, lowp>			lowp_dvec2;
+	typedef vec<3, f64, lowp>			lowp_dvec3;
+	typedef vec<4, f64, lowp>			lowp_dvec4;
+
+	typedef vec<1, f64, mediump>		mediump_dvec1;
+	typedef vec<2, f64, mediump>		mediump_dvec2;
+	typedef vec<3, f64, mediump>		mediump_dvec3;
+	typedef vec<4, f64, mediump>		mediump_dvec4;
+
+	typedef vec<1, f64, highp>			highp_dvec1;
+	typedef vec<2, f64, highp>			highp_dvec2;
+	typedef vec<3, f64, highp>			highp_dvec3;
+	typedef vec<4, f64, highp>			highp_dvec4;
+
+	typedef vec<1, f64, defaultp>		dvec1;
+	typedef vec<2, f64, defaultp>		dvec2;
+	typedef vec<3, f64, defaultp>		dvec3;
+	typedef vec<4, f64, defaultp>		dvec4;
+
+	typedef vec<1, f64, lowp>			lowp_f64vec1;
+	typedef vec<2, f64, lowp>			lowp_f64vec2;
+	typedef vec<3, f64, lowp>			lowp_f64vec3;
+	typedef vec<4, f64, lowp>			lowp_f64vec4;
+
+	typedef vec<1, f64, mediump>		mediump_f64vec1;
+	typedef vec<2, f64, mediump>		mediump_f64vec2;
+	typedef vec<3, f64, mediump>		mediump_f64vec3;
+	typedef vec<4, f64, mediump>		mediump_f64vec4;
+
+	typedef vec<1, f64, highp>			highp_f64vec1;
+	typedef vec<2, f64, highp>			highp_f64vec2;
+	typedef vec<3, f64, highp>			highp_f64vec3;
+	typedef vec<4, f64, highp>			highp_f64vec4;
+
+	typedef vec<1, f64, defaultp>		f64vec1;
+	typedef vec<2, f64, defaultp>		f64vec2;
+	typedef vec<3, f64, defaultp>		f64vec3;
+	typedef vec<4, f64, defaultp>		f64vec4;
+
+	// Matrix NxN
+
+	typedef mat<2, 2, f32, lowp>		lowp_mat2;
+	typedef mat<3, 3, f32, lowp>		lowp_mat3;
+	typedef mat<4, 4, f32, lowp>		lowp_mat4;
+
+	typedef mat<2, 2, f32, mediump>		mediump_mat2;
+	typedef mat<3, 3, f32, mediump>		mediump_mat3;
+	typedef mat<4, 4, f32, mediump>		mediump_mat4;
+
+	typedef mat<2, 2, f32, highp>		highp_mat2;
+	typedef mat<3, 3, f32, highp>		highp_mat3;
+	typedef mat<4, 4, f32, highp>		highp_mat4;
+
+	typedef mat<2, 2, f32, defaultp>	mat2;
+	typedef mat<3, 3, f32, defaultp>	mat3;
+	typedef mat<4, 4, f32, defaultp>	mat4;
+
+	typedef mat<2, 2, f32, lowp>		lowp_fmat2;
+	typedef mat<3, 3, f32, lowp>		lowp_fmat3;
+	typedef mat<4, 4, f32, lowp>		lowp_fmat4;
+
+	typedef mat<2, 2, f32, mediump>		mediump_fmat2;
+	typedef mat<3, 3, f32, mediump>		mediump_fmat3;
+	typedef mat<4, 4, f32, mediump>		mediump_fmat4;
+
+	typedef mat<2, 2, f32, highp>		highp_fmat2;
+	typedef mat<3, 3, f32, highp>		highp_fmat3;
+	typedef mat<4, 4, f32, highp>		highp_fmat4;
+
+	typedef mat<2, 2, f32, defaultp>	fmat2;
+	typedef mat<3, 3, f32, defaultp>	fmat3;
+	typedef mat<4, 4, f32, defaultp>	fmat4;
+
+	typedef mat<2, 2, f32, lowp>		lowp_f32mat2;
+	typedef mat<3, 3, f32, lowp>		lowp_f32mat3;
+	typedef mat<4, 4, f32, lowp>		lowp_f32mat4;
+
+	typedef mat<2, 2, f32, mediump>		mediump_f32mat2;
+	typedef mat<3, 3, f32, mediump>		mediump_f32mat3;
+	typedef mat<4, 4, f32, mediump>		mediump_f32mat4;
+
+	typedef mat<2, 2, f32, highp>		highp_f32mat2;
+	typedef mat<3, 3, f32, highp>		highp_f32mat3;
+	typedef mat<4, 4, f32, highp>		highp_f32mat4;
+
+	typedef mat<2, 2, f32, defaultp>	f32mat2;
+	typedef mat<3, 3, f32, defaultp>	f32mat3;
+	typedef mat<4, 4, f32, defaultp>	f32mat4;
+
+	typedef mat<2, 2, f64, lowp>		lowp_dmat2;
+	typedef mat<3, 3, f64, lowp>		lowp_dmat3;
+	typedef mat<4, 4, f64, lowp>		lowp_dmat4;
+
+	typedef mat<2, 2, f64, mediump>		mediump_dmat2;
+	typedef mat<3, 3, f64, mediump>		mediump_dmat3;
+	typedef mat<4, 4, f64, mediump>		mediump_dmat4;
+
+	typedef mat<2, 2, f64, highp>		highp_dmat2;
+	typedef mat<3, 3, f64, highp>		highp_dmat3;
+	typedef mat<4, 4, f64, highp>		highp_dmat4;
+
+	typedef mat<2, 2, f64, defaultp>	dmat2;
+	typedef mat<3, 3, f64, defaultp>	dmat3;
+	typedef mat<4, 4, f64, defaultp>	dmat4;
+
+	typedef mat<2, 2, f64, lowp>		lowp_f64mat2;
+	typedef mat<3, 3, f64, lowp>		lowp_f64mat3;
+	typedef mat<4, 4, f64, lowp>		lowp_f64mat4;
+
+	typedef mat<2, 2, f64, mediump>		mediump_f64mat2;
+	typedef mat<3, 3, f64, mediump>		mediump_f64mat3;
+	typedef mat<4, 4, f64, mediump>		mediump_f64mat4;
+
+	typedef mat<2, 2, f64, highp>		highp_f64mat2;
+	typedef mat<3, 3, f64, highp>		highp_f64mat3;
+	typedef mat<4, 4, f64, highp>		highp_f64mat4;
+
+	typedef mat<2, 2, f64, defaultp>	f64mat2;
+	typedef mat<3, 3, f64, defaultp>	f64mat3;
+	typedef mat<4, 4, f64, defaultp>	f64mat4;
+
+	// Matrix MxN
+
+	typedef mat<2, 2, f32, lowp>		lowp_mat2x2;
+	typedef mat<2, 3, f32, lowp>		lowp_mat2x3;
+	typedef mat<2, 4, f32, lowp>		lowp_mat2x4;
+	typedef mat<3, 2, f32, lowp>		lowp_mat3x2;
+	typedef mat<3, 3, f32, lowp>		lowp_mat3x3;
+	typedef mat<3, 4, f32, lowp>		lowp_mat3x4;
+	typedef mat<4, 2, f32, lowp>		lowp_mat4x2;
+	typedef mat<4, 3, f32, lowp>		lowp_mat4x3;
+	typedef mat<4, 4, f32, lowp>		lowp_mat4x4;
+
+	typedef mat<2, 2, f32, mediump>		mediump_mat2x2;
+	typedef mat<2, 3, f32, mediump>		mediump_mat2x3;
+	typedef mat<2, 4, f32, mediump>		mediump_mat2x4;
+	typedef mat<3, 2, f32, mediump>		mediump_mat3x2;
+	typedef mat<3, 3, f32, mediump>		mediump_mat3x3;
+	typedef mat<3, 4, f32, mediump>		mediump_mat3x4;
+	typedef mat<4, 2, f32, mediump>		mediump_mat4x2;
+	typedef mat<4, 3, f32, mediump>		mediump_mat4x3;
+	typedef mat<4, 4, f32, mediump>		mediump_mat4x4;
+
+	typedef mat<2, 2, f32, highp>		highp_mat2x2;
+	typedef mat<2, 3, f32, highp>		highp_mat2x3;
+	typedef mat<2, 4, f32, highp>		highp_mat2x4;
+	typedef mat<3, 2, f32, highp>		highp_mat3x2;
+	typedef mat<3, 3, f32, highp>		highp_mat3x3;
+	typedef mat<3, 4, f32, highp>		highp_mat3x4;
+	typedef mat<4, 2, f32, highp>		highp_mat4x2;
+	typedef mat<4, 3, f32, highp>		highp_mat4x3;
+	typedef mat<4, 4, f32, highp>		highp_mat4x4;
+
+	typedef mat<2, 2, f32, defaultp>	mat2x2;
+	typedef mat<2, 3, f32, defaultp>	mat2x3;
+	typedef mat<2, 4, f32, defaultp>	mat2x4;
+	typedef mat<3, 2, f32, defaultp>	mat3x2;
+	typedef mat<3, 3, f32, defaultp>	mat3x3;
+	typedef mat<3, 4, f32, defaultp>	mat3x4;
+	typedef mat<4, 2, f32, defaultp>	mat4x2;
+	typedef mat<4, 3, f32, defaultp>	mat4x3;
+	typedef mat<4, 4, f32, defaultp>	mat4x4;
+
+	typedef mat<2, 2, f32, lowp>		lowp_fmat2x2;
+	typedef mat<2, 3, f32, lowp>		lowp_fmat2x3;
+	typedef mat<2, 4, f32, lowp>		lowp_fmat2x4;
+	typedef mat<3, 2, f32, lowp>		lowp_fmat3x2;
+	typedef mat<3, 3, f32, lowp>		lowp_fmat3x3;
+	typedef mat<3, 4, f32, lowp>		lowp_fmat3x4;
+	typedef mat<4, 2, f32, lowp>		lowp_fmat4x2;
+	typedef mat<4, 3, f32, lowp>		lowp_fmat4x3;
+	typedef mat<4, 4, f32, lowp>		lowp_fmat4x4;
+
+	typedef mat<2, 2, f32, mediump>		mediump_fmat2x2;
+	typedef mat<2, 3, f32, mediump>		mediump_fmat2x3;
+	typedef mat<2, 4, f32, mediump>		mediump_fmat2x4;
+	typedef mat<3, 2, f32, mediump>		mediump_fmat3x2;
+	typedef mat<3, 3, f32, mediump>		mediump_fmat3x3;
+	typedef mat<3, 4, f32, mediump>		mediump_fmat3x4;
+	typedef mat<4, 2, f32, mediump>		mediump_fmat4x2;
+	typedef mat<4, 3, f32, mediump>		mediump_fmat4x3;
+	typedef mat<4, 4, f32, mediump>		mediump_fmat4x4;
+
+	typedef mat<2, 2, f32, highp>		highp_fmat2x2;
+	typedef mat<2, 3, f32, highp>		highp_fmat2x3;
+	typedef mat<2, 4, f32, highp>		highp_fmat2x4;
+	typedef mat<3, 2, f32, highp>		highp_fmat3x2;
+	typedef mat<3, 3, f32, highp>		highp_fmat3x3;
+	typedef mat<3, 4, f32, highp>		highp_fmat3x4;
+	typedef mat<4, 2, f32, highp>		highp_fmat4x2;
+	typedef mat<4, 3, f32, highp>		highp_fmat4x3;
+	typedef mat<4, 4, f32, highp>		highp_fmat4x4;
+
+	typedef mat<2, 2, f32, defaultp>	fmat2x2;
+	typedef mat<2, 3, f32, defaultp>	fmat2x3;
+	typedef mat<2, 4, f32, defaultp>	fmat2x4;
+	typedef mat<3, 2, f32, defaultp>	fmat3x2;
+	typedef mat<3, 3, f32, defaultp>	fmat3x3;
+	typedef mat<3, 4, f32, defaultp>	fmat3x4;
+	typedef mat<4, 2, f32, defaultp>	fmat4x2;
+	typedef mat<4, 3, f32, defaultp>	fmat4x3;
+	typedef mat<4, 4, f32, defaultp>	fmat4x4;
+
+	typedef mat<2, 2, f32, lowp>		lowp_f32mat2x2;
+	typedef mat<2, 3, f32, lowp>		lowp_f32mat2x3;
+	typedef mat<2, 4, f32, lowp>		lowp_f32mat2x4;
+	typedef mat<3, 2, f32, lowp>		lowp_f32mat3x2;
+	typedef mat<3, 3, f32, lowp>		lowp_f32mat3x3;
+	typedef mat<3, 4, f32, lowp>		lowp_f32mat3x4;
+	typedef mat<4, 2, f32, lowp>		lowp_f32mat4x2;
+	typedef mat<4, 3, f32, lowp>		lowp_f32mat4x3;
+	typedef mat<4, 4, f32, lowp>		lowp_f32mat4x4;
+	
+	typedef mat<2, 2, f32, mediump>		mediump_f32mat2x2;
+	typedef mat<2, 3, f32, mediump>		mediump_f32mat2x3;
+	typedef mat<2, 4, f32, mediump>		mediump_f32mat2x4;
+	typedef mat<3, 2, f32, mediump>		mediump_f32mat3x2;
+	typedef mat<3, 3, f32, mediump>		mediump_f32mat3x3;
+	typedef mat<3, 4, f32, mediump>		mediump_f32mat3x4;
+	typedef mat<4, 2, f32, mediump>		mediump_f32mat4x2;
+	typedef mat<4, 3, f32, mediump>		mediump_f32mat4x3;
+	typedef mat<4, 4, f32, mediump>		mediump_f32mat4x4;
+
+	typedef mat<2, 2, f32, highp>		highp_f32mat2x2;
+	typedef mat<2, 3, f32, highp>		highp_f32mat2x3;
+	typedef mat<2, 4, f32, highp>		highp_f32mat2x4;
+	typedef mat<3, 2, f32, highp>		highp_f32mat3x2;
+	typedef mat<3, 3, f32, highp>		highp_f32mat3x3;
+	typedef mat<3, 4, f32, highp>		highp_f32mat3x4;
+	typedef mat<4, 2, f32, highp>		highp_f32mat4x2;
+	typedef mat<4, 3, f32, highp>		highp_f32mat4x3;
+	typedef mat<4, 4, f32, highp>		highp_f32mat4x4;
+
+	typedef mat<2, 2, f32, defaultp>	f32mat2x2;
+	typedef mat<2, 3, f32, defaultp>	f32mat2x3;
+	typedef mat<2, 4, f32, defaultp>	f32mat2x4;
+	typedef mat<3, 2, f32, defaultp>	f32mat3x2;
+	typedef mat<3, 3, f32, defaultp>	f32mat3x3;
+	typedef mat<3, 4, f32, defaultp>	f32mat3x4;
+	typedef mat<4, 2, f32, defaultp>	f32mat4x2;
+	typedef mat<4, 3, f32, defaultp>	f32mat4x3;
+	typedef mat<4, 4, f32, defaultp>	f32mat4x4;
+
+	typedef mat<2, 2, double, lowp>		lowp_dmat2x2;
+	typedef mat<2, 3, double, lowp>		lowp_dmat2x3;
+	typedef mat<2, 4, double, lowp>		lowp_dmat2x4;
+	typedef mat<3, 2, double, lowp>		lowp_dmat3x2;
+	typedef mat<3, 3, double, lowp>		lowp_dmat3x3;
+	typedef mat<3, 4, double, lowp>		lowp_dmat3x4;
+	typedef mat<4, 2, double, lowp>		lowp_dmat4x2;
+	typedef mat<4, 3, double, lowp>		lowp_dmat4x3;
+	typedef mat<4, 4, double, lowp>		lowp_dmat4x4;
+
+	typedef mat<2, 2, double, mediump>	mediump_dmat2x2;
+	typedef mat<2, 3, double, mediump>	mediump_dmat2x3;
+	typedef mat<2, 4, double, mediump>	mediump_dmat2x4;
+	typedef mat<3, 2, double, mediump>	mediump_dmat3x2;
+	typedef mat<3, 3, double, mediump>	mediump_dmat3x3;
+	typedef mat<3, 4, double, mediump>	mediump_dmat3x4;
+	typedef mat<4, 2, double, mediump>	mediump_dmat4x2;
+	typedef mat<4, 3, double, mediump>	mediump_dmat4x3;
+	typedef mat<4, 4, double, mediump>	mediump_dmat4x4;
+
+	typedef mat<2, 2, double, highp>	highp_dmat2x2;
+	typedef mat<2, 3, double, highp>	highp_dmat2x3;
+	typedef mat<2, 4, double, highp>	highp_dmat2x4;
+	typedef mat<3, 2, double, highp>	highp_dmat3x2;
+	typedef mat<3, 3, double, highp>	highp_dmat3x3;
+	typedef mat<3, 4, double, highp>	highp_dmat3x4;
+	typedef mat<4, 2, double, highp>	highp_dmat4x2;
+	typedef mat<4, 3, double, highp>	highp_dmat4x3;
+	typedef mat<4, 4, double, highp>	highp_dmat4x4;
+
+	typedef mat<2, 2, double, defaultp>	dmat2x2;
+	typedef mat<2, 3, double, defaultp>	dmat2x3;
+	typedef mat<2, 4, double, defaultp>	dmat2x4;
+	typedef mat<3, 2, double, defaultp>	dmat3x2;
+	typedef mat<3, 3, double, defaultp>	dmat3x3;
+	typedef mat<3, 4, double, defaultp>	dmat3x4;
+	typedef mat<4, 2, double, defaultp>	dmat4x2;
+	typedef mat<4, 3, double, defaultp>	dmat4x3;
+	typedef mat<4, 4, double, defaultp>	dmat4x4;
+
+	typedef mat<2, 2, f64, lowp>		lowp_f64mat2x2;
+	typedef mat<2, 3, f64, lowp>		lowp_f64mat2x3;
+	typedef mat<2, 4, f64, lowp>		lowp_f64mat2x4;
+	typedef mat<3, 2, f64, lowp>		lowp_f64mat3x2;
+	typedef mat<3, 3, f64, lowp>		lowp_f64mat3x3;
+	typedef mat<3, 4, f64, lowp>		lowp_f64mat3x4;
+	typedef mat<4, 2, f64, lowp>		lowp_f64mat4x2;
+	typedef mat<4, 3, f64, lowp>		lowp_f64mat4x3;
+	typedef mat<4, 4, f64, lowp>		lowp_f64mat4x4;
+
+	typedef mat<2, 2, f64, mediump>		mediump_f64mat2x2;
+	typedef mat<2, 3, f64, mediump>		mediump_f64mat2x3;
+	typedef mat<2, 4, f64, mediump>		mediump_f64mat2x4;
+	typedef mat<3, 2, f64, mediump>		mediump_f64mat3x2;
+	typedef mat<3, 3, f64, mediump>		mediump_f64mat3x3;
+	typedef mat<3, 4, f64, mediump>		mediump_f64mat3x4;
+	typedef mat<4, 2, f64, mediump>		mediump_f64mat4x2;
+	typedef mat<4, 3, f64, mediump>		mediump_f64mat4x3;
+	typedef mat<4, 4, f64, mediump>		mediump_f64mat4x4;
+
+	typedef mat<2, 2, f64, highp>		highp_f64mat2x2;
+	typedef mat<2, 3, f64, highp>		highp_f64mat2x3;
+	typedef mat<2, 4, f64, highp>		highp_f64mat2x4;
+	typedef mat<3, 2, f64, highp>		highp_f64mat3x2;
+	typedef mat<3, 3, f64, highp>		highp_f64mat3x3;
+	typedef mat<3, 4, f64, highp>		highp_f64mat3x4;
+	typedef mat<4, 2, f64, highp>		highp_f64mat4x2;
+	typedef mat<4, 3, f64, highp>		highp_f64mat4x3;
+	typedef mat<4, 4, f64, highp>		highp_f64mat4x4;
+
+	typedef mat<2, 2, f64, defaultp>	f64mat2x2;
+	typedef mat<2, 3, f64, defaultp>	f64mat2x3;
+	typedef mat<2, 4, f64, defaultp>	f64mat2x4;
+	typedef mat<3, 2, f64, defaultp>	f64mat3x2;
+	typedef mat<3, 3, f64, defaultp>	f64mat3x3;
+	typedef mat<3, 4, f64, defaultp>	f64mat3x4;
+	typedef mat<4, 2, f64, defaultp>	f64mat4x2;
+	typedef mat<4, 3, f64, defaultp>	f64mat4x3;
+	typedef mat<4, 4, f64, defaultp>	f64mat4x4;
+
+	// Signed integer matrix MxN
+
+	typedef mat<2, 2, int, lowp>		lowp_imat2x2;
+	typedef mat<2, 3, int, lowp>		lowp_imat2x3;
+	typedef mat<2, 4, int, lowp>		lowp_imat2x4;
+	typedef mat<3, 2, int, lowp>		lowp_imat3x2;
+	typedef mat<3, 3, int, lowp>		lowp_imat3x3;
+	typedef mat<3, 4, int, lowp>		lowp_imat3x4;
+	typedef mat<4, 2, int, lowp>		lowp_imat4x2;
+	typedef mat<4, 3, int, lowp>		lowp_imat4x3;
+	typedef mat<4, 4, int, lowp>		lowp_imat4x4;
+
+	typedef mat<2, 2, int, mediump>		mediump_imat2x2;
+	typedef mat<2, 3, int, mediump>		mediump_imat2x3;
+	typedef mat<2, 4, int, mediump>		mediump_imat2x4;
+	typedef mat<3, 2, int, mediump>		mediump_imat3x2;
+	typedef mat<3, 3, int, mediump>		mediump_imat3x3;
+	typedef mat<3, 4, int, mediump>		mediump_imat3x4;
+	typedef mat<4, 2, int, mediump>		mediump_imat4x2;
+	typedef mat<4, 3, int, mediump>		mediump_imat4x3;
+	typedef mat<4, 4, int, mediump>		mediump_imat4x4;
+
+	typedef mat<2, 2, int, highp>		highp_imat2x2;
+	typedef mat<2, 3, int, highp>		highp_imat2x3;
+	typedef mat<2, 4, int, highp>		highp_imat2x4;
+	typedef mat<3, 2, int, highp>		highp_imat3x2;
+	typedef mat<3, 3, int, highp>		highp_imat3x3;
+	typedef mat<3, 4, int, highp>		highp_imat3x4;
+	typedef mat<4, 2, int, highp>		highp_imat4x2;
+	typedef mat<4, 3, int, highp>		highp_imat4x3;
+	typedef mat<4, 4, int, highp>		highp_imat4x4;
+
+	typedef mat<2, 2, int, defaultp>	imat2x2;
+	typedef mat<2, 3, int, defaultp>	imat2x3;
+	typedef mat<2, 4, int, defaultp>	imat2x4;
+	typedef mat<3, 2, int, defaultp>	imat3x2;
+	typedef mat<3, 3, int, defaultp>	imat3x3;
+	typedef mat<3, 4, int, defaultp>	imat3x4;
+	typedef mat<4, 2, int, defaultp>	imat4x2;
+	typedef mat<4, 3, int, defaultp>	imat4x3;
+	typedef mat<4, 4, int, defaultp>	imat4x4;
+
+
+	typedef mat<2, 2, int8, lowp>		lowp_i8mat2x2;
+	typedef mat<2, 3, int8, lowp>		lowp_i8mat2x3;
+	typedef mat<2, 4, int8, lowp>		lowp_i8mat2x4;
+	typedef mat<3, 2, int8, lowp>		lowp_i8mat3x2;
+	typedef mat<3, 3, int8, lowp>		lowp_i8mat3x3;
+	typedef mat<3, 4, int8, lowp>		lowp_i8mat3x4;
+	typedef mat<4, 2, int8, lowp>		lowp_i8mat4x2;
+	typedef mat<4, 3, int8, lowp>		lowp_i8mat4x3;
+	typedef mat<4, 4, int8, lowp>		lowp_i8mat4x4;
+
+	typedef mat<2, 2, int8, mediump>	mediump_i8mat2x2;
+	typedef mat<2, 3, int8, mediump>	mediump_i8mat2x3;
+	typedef mat<2, 4, int8, mediump>	mediump_i8mat2x4;
+	typedef mat<3, 2, int8, mediump>	mediump_i8mat3x2;
+	typedef mat<3, 3, int8, mediump>	mediump_i8mat3x3;
+	typedef mat<3, 4, int8, mediump>	mediump_i8mat3x4;
+	typedef mat<4, 2, int8, mediump>	mediump_i8mat4x2;
+	typedef mat<4, 3, int8, mediump>	mediump_i8mat4x3;
+	typedef mat<4, 4, int8, mediump>	mediump_i8mat4x4;
+
+	typedef mat<2, 2, int8, highp>		highp_i8mat2x2;
+	typedef mat<2, 3, int8, highp>		highp_i8mat2x3;
+	typedef mat<2, 4, int8, highp>		highp_i8mat2x4;
+	typedef mat<3, 2, int8, highp>		highp_i8mat3x2;
+	typedef mat<3, 3, int8, highp>		highp_i8mat3x3;
+	typedef mat<3, 4, int8, highp>		highp_i8mat3x4;
+	typedef mat<4, 2, int8, highp>		highp_i8mat4x2;
+	typedef mat<4, 3, int8, highp>		highp_i8mat4x3;
+	typedef mat<4, 4, int8, highp>		highp_i8mat4x4;
+
+	typedef mat<2, 2, int8, defaultp>	i8mat2x2;
+	typedef mat<2, 3, int8, defaultp>	i8mat2x3;
+	typedef mat<2, 4, int8, defaultp>	i8mat2x4;
+	typedef mat<3, 2, int8, defaultp>	i8mat3x2;
+	typedef mat<3, 3, int8, defaultp>	i8mat3x3;
+	typedef mat<3, 4, int8, defaultp>	i8mat3x4;
+	typedef mat<4, 2, int8, defaultp>	i8mat4x2;
+	typedef mat<4, 3, int8, defaultp>	i8mat4x3;
+	typedef mat<4, 4, int8, defaultp>	i8mat4x4;
+
+
+	typedef mat<2, 2, int16, lowp>		lowp_i16mat2x2;
+	typedef mat<2, 3, int16, lowp>		lowp_i16mat2x3;
+	typedef mat<2, 4, int16, lowp>		lowp_i16mat2x4;
+	typedef mat<3, 2, int16, lowp>		lowp_i16mat3x2;
+	typedef mat<3, 3, int16, lowp>		lowp_i16mat3x3;
+	typedef mat<3, 4, int16, lowp>		lowp_i16mat3x4;
+	typedef mat<4, 2, int16, lowp>		lowp_i16mat4x2;
+	typedef mat<4, 3, int16, lowp>		lowp_i16mat4x3;
+	typedef mat<4, 4, int16, lowp>		lowp_i16mat4x4;
+
+	typedef mat<2, 2, int16, mediump>	mediump_i16mat2x2;
+	typedef mat<2, 3, int16, mediump>	mediump_i16mat2x3;
+	typedef mat<2, 4, int16, mediump>	mediump_i16mat2x4;
+	typedef mat<3, 2, int16, mediump>	mediump_i16mat3x2;
+	typedef mat<3, 3, int16, mediump>	mediump_i16mat3x3;
+	typedef mat<3, 4, int16, mediump>	mediump_i16mat3x4;
+	typedef mat<4, 2, int16, mediump>	mediump_i16mat4x2;
+	typedef mat<4, 3, int16, mediump>	mediump_i16mat4x3;
+	typedef mat<4, 4, int16, mediump>	mediump_i16mat4x4;
+
+	typedef mat<2, 2, int16, highp>		highp_i16mat2x2;
+	typedef mat<2, 3, int16, highp>		highp_i16mat2x3;
+	typedef mat<2, 4, int16, highp>		highp_i16mat2x4;
+	typedef mat<3, 2, int16, highp>		highp_i16mat3x2;
+	typedef mat<3, 3, int16, highp>		highp_i16mat3x3;
+	typedef mat<3, 4, int16, highp>		highp_i16mat3x4;
+	typedef mat<4, 2, int16, highp>		highp_i16mat4x2;
+	typedef mat<4, 3, int16, highp>		highp_i16mat4x3;
+	typedef mat<4, 4, int16, highp>		highp_i16mat4x4;
+
+	typedef mat<2, 2, int16, defaultp>	i16mat2x2;
+	typedef mat<2, 3, int16, defaultp>	i16mat2x3;
+	typedef mat<2, 4, int16, defaultp>	i16mat2x4;
+	typedef mat<3, 2, int16, defaultp>	i16mat3x2;
+	typedef mat<3, 3, int16, defaultp>	i16mat3x3;
+	typedef mat<3, 4, int16, defaultp>	i16mat3x4;
+	typedef mat<4, 2, int16, defaultp>	i16mat4x2;
+	typedef mat<4, 3, int16, defaultp>	i16mat4x3;
+	typedef mat<4, 4, int16, defaultp>	i16mat4x4;
+
+
+	typedef mat<2, 2, int32, lowp>		lowp_i32mat2x2;
+	typedef mat<2, 3, int32, lowp>		lowp_i32mat2x3;
+	typedef mat<2, 4, int32, lowp>		lowp_i32mat2x4;
+	typedef mat<3, 2, int32, lowp>		lowp_i32mat3x2;
+	typedef mat<3, 3, int32, lowp>		lowp_i32mat3x3;
+	typedef mat<3, 4, int32, lowp>		lowp_i32mat3x4;
+	typedef mat<4, 2, int32, lowp>		lowp_i32mat4x2;
+	typedef mat<4, 3, int32, lowp>		lowp_i32mat4x3;
+	typedef mat<4, 4, int32, lowp>		lowp_i32mat4x4;
+
+	typedef mat<2, 2, int32, mediump>	mediump_i32mat2x2;
+	typedef mat<2, 3, int32, mediump>	mediump_i32mat2x3;
+	typedef mat<2, 4, int32, mediump>	mediump_i32mat2x4;
+	typedef mat<3, 2, int32, mediump>	mediump_i32mat3x2;
+	typedef mat<3, 3, int32, mediump>	mediump_i32mat3x3;
+	typedef mat<3, 4, int32, mediump>	mediump_i32mat3x4;
+	typedef mat<4, 2, int32, mediump>	mediump_i32mat4x2;
+	typedef mat<4, 3, int32, mediump>	mediump_i32mat4x3;
+	typedef mat<4, 4, int32, mediump>	mediump_i32mat4x4;
+
+	typedef mat<2, 2, int32, highp>		highp_i32mat2x2;
+	typedef mat<2, 3, int32, highp>		highp_i32mat2x3;
+	typedef mat<2, 4, int32, highp>		highp_i32mat2x4;
+	typedef mat<3, 2, int32, highp>		highp_i32mat3x2;
+	typedef mat<3, 3, int32, highp>		highp_i32mat3x3;
+	typedef mat<3, 4, int32, highp>		highp_i32mat3x4;
+	typedef mat<4, 2, int32, highp>		highp_i32mat4x2;
+	typedef mat<4, 3, int32, highp>		highp_i32mat4x3;
+	typedef mat<4, 4, int32, highp>		highp_i32mat4x4;
+
+	typedef mat<2, 2, int32, defaultp>	i32mat2x2;
+	typedef mat<2, 3, int32, defaultp>	i32mat2x3;
+	typedef mat<2, 4, int32, defaultp>	i32mat2x4;
+	typedef mat<3, 2, int32, defaultp>	i32mat3x2;
+	typedef mat<3, 3, int32, defaultp>	i32mat3x3;
+	typedef mat<3, 4, int32, defaultp>	i32mat3x4;
+	typedef mat<4, 2, int32, defaultp>	i32mat4x2;
+	typedef mat<4, 3, int32, defaultp>	i32mat4x3;
+	typedef mat<4, 4, int32, defaultp>	i32mat4x4;
+
+
+	typedef mat<2, 2, int64, lowp>		lowp_i64mat2x2;
+	typedef mat<2, 3, int64, lowp>		lowp_i64mat2x3;
+	typedef mat<2, 4, int64, lowp>		lowp_i64mat2x4;
+	typedef mat<3, 2, int64, lowp>		lowp_i64mat3x2;
+	typedef mat<3, 3, int64, lowp>		lowp_i64mat3x3;
+	typedef mat<3, 4, int64, lowp>		lowp_i64mat3x4;
+	typedef mat<4, 2, int64, lowp>		lowp_i64mat4x2;
+	typedef mat<4, 3, int64, lowp>		lowp_i64mat4x3;
+	typedef mat<4, 4, int64, lowp>		lowp_i64mat4x4;
+
+	typedef mat<2, 2, int64, mediump>	mediump_i64mat2x2;
+	typedef mat<2, 3, int64, mediump>	mediump_i64mat2x3;
+	typedef mat<2, 4, int64, mediump>	mediump_i64mat2x4;
+	typedef mat<3, 2, int64, mediump>	mediump_i64mat3x2;
+	typedef mat<3, 3, int64, mediump>	mediump_i64mat3x3;
+	typedef mat<3, 4, int64, mediump>	mediump_i64mat3x4;
+	typedef mat<4, 2, int64, mediump>	mediump_i64mat4x2;
+	typedef mat<4, 3, int64, mediump>	mediump_i64mat4x3;
+	typedef mat<4, 4, int64, mediump>	mediump_i64mat4x4;
+
+	typedef mat<2, 2, int64, highp>		highp_i64mat2x2;
+	typedef mat<2, 3, int64, highp>		highp_i64mat2x3;
+	typedef mat<2, 4, int64, highp>		highp_i64mat2x4;
+	typedef mat<3, 2, int64, highp>		highp_i64mat3x2;
+	typedef mat<3, 3, int64, highp>		highp_i64mat3x3;
+	typedef mat<3, 4, int64, highp>		highp_i64mat3x4;
+	typedef mat<4, 2, int64, highp>		highp_i64mat4x2;
+	typedef mat<4, 3, int64, highp>		highp_i64mat4x3;
+	typedef mat<4, 4, int64, highp>		highp_i64mat4x4;
+
+	typedef mat<2, 2, int64, defaultp>	i64mat2x2;
+	typedef mat<2, 3, int64, defaultp>	i64mat2x3;
+	typedef mat<2, 4, int64, defaultp>	i64mat2x4;
+	typedef mat<3, 2, int64, defaultp>	i64mat3x2;
+	typedef mat<3, 3, int64, defaultp>	i64mat3x3;
+	typedef mat<3, 4, int64, defaultp>	i64mat3x4;
+	typedef mat<4, 2, int64, defaultp>	i64mat4x2;
+	typedef mat<4, 3, int64, defaultp>	i64mat4x3;
+	typedef mat<4, 4, int64, defaultp>	i64mat4x4;
+
+
+	// Unsigned integer matrix MxN
+
+	typedef mat<2, 2, uint, lowp>		lowp_umat2x2;
+	typedef mat<2, 3, uint, lowp>		lowp_umat2x3;
+	typedef mat<2, 4, uint, lowp>		lowp_umat2x4;
+	typedef mat<3, 2, uint, lowp>		lowp_umat3x2;
+	typedef mat<3, 3, uint, lowp>		lowp_umat3x3;
+	typedef mat<3, 4, uint, lowp>		lowp_umat3x4;
+	typedef mat<4, 2, uint, lowp>		lowp_umat4x2;
+	typedef mat<4, 3, uint, lowp>		lowp_umat4x3;
+	typedef mat<4, 4, uint, lowp>		lowp_umat4x4;
+
+	typedef mat<2, 2, uint, mediump>	mediump_umat2x2;
+	typedef mat<2, 3, uint, mediump>	mediump_umat2x3;
+	typedef mat<2, 4, uint, mediump>	mediump_umat2x4;
+	typedef mat<3, 2, uint, mediump>	mediump_umat3x2;
+	typedef mat<3, 3, uint, mediump>	mediump_umat3x3;
+	typedef mat<3, 4, uint, mediump>	mediump_umat3x4;
+	typedef mat<4, 2, uint, mediump>	mediump_umat4x2;
+	typedef mat<4, 3, uint, mediump>	mediump_umat4x3;
+	typedef mat<4, 4, uint, mediump>	mediump_umat4x4;
+
+	typedef mat<2, 2, uint, highp>		highp_umat2x2;
+	typedef mat<2, 3, uint, highp>		highp_umat2x3;
+	typedef mat<2, 4, uint, highp>		highp_umat2x4;
+	typedef mat<3, 2, uint, highp>		highp_umat3x2;
+	typedef mat<3, 3, uint, highp>		highp_umat3x3;
+	typedef mat<3, 4, uint, highp>		highp_umat3x4;
+	typedef mat<4, 2, uint, highp>		highp_umat4x2;
+	typedef mat<4, 3, uint, highp>		highp_umat4x3;
+	typedef mat<4, 4, uint, highp>		highp_umat4x4;
+
+	typedef mat<2, 2, uint, defaultp>	umat2x2;
+	typedef mat<2, 3, uint, defaultp>	umat2x3;
+	typedef mat<2, 4, uint, defaultp>	umat2x4;
+	typedef mat<3, 2, uint, defaultp>	umat3x2;
+	typedef mat<3, 3, uint, defaultp>	umat3x3;
+	typedef mat<3, 4, uint, defaultp>	umat3x4;
+	typedef mat<4, 2, uint, defaultp>	umat4x2;
+	typedef mat<4, 3, uint, defaultp>	umat4x3;
+	typedef mat<4, 4, uint, defaultp>	umat4x4;
+
+
+	typedef mat<2, 2, uint8, lowp>		lowp_u8mat2x2;
+	typedef mat<2, 3, uint8, lowp>		lowp_u8mat2x3;
+	typedef mat<2, 4, uint8, lowp>		lowp_u8mat2x4;
+	typedef mat<3, 2, uint8, lowp>		lowp_u8mat3x2;
+	typedef mat<3, 3, uint8, lowp>		lowp_u8mat3x3;
+	typedef mat<3, 4, uint8, lowp>		lowp_u8mat3x4;
+	typedef mat<4, 2, uint8, lowp>		lowp_u8mat4x2;
+	typedef mat<4, 3, uint8, lowp>		lowp_u8mat4x3;
+	typedef mat<4, 4, uint8, lowp>		lowp_u8mat4x4;
+
+	typedef mat<2, 2, uint8, mediump>	mediump_u8mat2x2;
+	typedef mat<2, 3, uint8, mediump>	mediump_u8mat2x3;
+	typedef mat<2, 4, uint8, mediump>	mediump_u8mat2x4;
+	typedef mat<3, 2, uint8, mediump>	mediump_u8mat3x2;
+	typedef mat<3, 3, uint8, mediump>	mediump_u8mat3x3;
+	typedef mat<3, 4, uint8, mediump>	mediump_u8mat3x4;
+	typedef mat<4, 2, uint8, mediump>	mediump_u8mat4x2;
+	typedef mat<4, 3, uint8, mediump>	mediump_u8mat4x3;
+	typedef mat<4, 4, uint8, mediump>	mediump_u8mat4x4;
+
+	typedef mat<2, 2, uint8, highp>		highp_u8mat2x2;
+	typedef mat<2, 3, uint8, highp>		highp_u8mat2x3;
+	typedef mat<2, 4, uint8, highp>		highp_u8mat2x4;
+	typedef mat<3, 2, uint8, highp>		highp_u8mat3x2;
+	typedef mat<3, 3, uint8, highp>		highp_u8mat3x3;
+	typedef mat<3, 4, uint8, highp>		highp_u8mat3x4;
+	typedef mat<4, 2, uint8, highp>		highp_u8mat4x2;
+	typedef mat<4, 3, uint8, highp>		highp_u8mat4x3;
+	typedef mat<4, 4, uint8, highp>		highp_u8mat4x4;
+
+	typedef mat<2, 2, uint8, defaultp>	u8mat2x2;
+	typedef mat<2, 3, uint8, defaultp>	u8mat2x3;
+	typedef mat<2, 4, uint8, defaultp>	u8mat2x4;
+	typedef mat<3, 2, uint8, defaultp>	u8mat3x2;
+	typedef mat<3, 3, uint8, defaultp>	u8mat3x3;
+	typedef mat<3, 4, uint8, defaultp>	u8mat3x4;
+	typedef mat<4, 2, uint8, defaultp>	u8mat4x2;
+	typedef mat<4, 3, uint8, defaultp>	u8mat4x3;
+	typedef mat<4, 4, uint8, defaultp>	u8mat4x4;
+
+
+	typedef mat<2, 2, uint16, lowp>		lowp_u16mat2x2;
+	typedef mat<2, 3, uint16, lowp>		lowp_u16mat2x3;
+	typedef mat<2, 4, uint16, lowp>		lowp_u16mat2x4;
+	typedef mat<3, 2, uint16, lowp>		lowp_u16mat3x2;
+	typedef mat<3, 3, uint16, lowp>		lowp_u16mat3x3;
+	typedef mat<3, 4, uint16, lowp>		lowp_u16mat3x4;
+	typedef mat<4, 2, uint16, lowp>		lowp_u16mat4x2;
+	typedef mat<4, 3, uint16, lowp>		lowp_u16mat4x3;
+	typedef mat<4, 4, uint16, lowp>		lowp_u16mat4x4;
+
+	typedef mat<2, 2, uint16, mediump>	mediump_u16mat2x2;
+	typedef mat<2, 3, uint16, mediump>	mediump_u16mat2x3;
+	typedef mat<2, 4, uint16, mediump>	mediump_u16mat2x4;
+	typedef mat<3, 2, uint16, mediump>	mediump_u16mat3x2;
+	typedef mat<3, 3, uint16, mediump>	mediump_u16mat3x3;
+	typedef mat<3, 4, uint16, mediump>	mediump_u16mat3x4;
+	typedef mat<4, 2, uint16, mediump>	mediump_u16mat4x2;
+	typedef mat<4, 3, uint16, mediump>	mediump_u16mat4x3;
+	typedef mat<4, 4, uint16, mediump>	mediump_u16mat4x4;
+
+	typedef mat<2, 2, uint16, highp>	highp_u16mat2x2;
+	typedef mat<2, 3, uint16, highp>	highp_u16mat2x3;
+	typedef mat<2, 4, uint16, highp>	highp_u16mat2x4;
+	typedef mat<3, 2, uint16, highp>	highp_u16mat3x2;
+	typedef mat<3, 3, uint16, highp>	highp_u16mat3x3;
+	typedef mat<3, 4, uint16, highp>	highp_u16mat3x4;
+	typedef mat<4, 2, uint16, highp>	highp_u16mat4x2;
+	typedef mat<4, 3, uint16, highp>	highp_u16mat4x3;
+	typedef mat<4, 4, uint16, highp>	highp_u16mat4x4;
+
+	typedef mat<2, 2, uint16, defaultp>	u16mat2x2;
+	typedef mat<2, 3, uint16, defaultp>	u16mat2x3;
+	typedef mat<2, 4, uint16, defaultp>	u16mat2x4;
+	typedef mat<3, 2, uint16, defaultp>	u16mat3x2;
+	typedef mat<3, 3, uint16, defaultp>	u16mat3x3;
+	typedef mat<3, 4, uint16, defaultp>	u16mat3x4;
+	typedef mat<4, 2, uint16, defaultp>	u16mat4x2;
+	typedef mat<4, 3, uint16, defaultp>	u16mat4x3;
+	typedef mat<4, 4, uint16, defaultp>	u16mat4x4;
+
+
+	typedef mat<2, 2, uint32, lowp>		lowp_u32mat2x2;
+	typedef mat<2, 3, uint32, lowp>		lowp_u32mat2x3;
+	typedef mat<2, 4, uint32, lowp>		lowp_u32mat2x4;
+	typedef mat<3, 2, uint32, lowp>		lowp_u32mat3x2;
+	typedef mat<3, 3, uint32, lowp>		lowp_u32mat3x3;
+	typedef mat<3, 4, uint32, lowp>		lowp_u32mat3x4;
+	typedef mat<4, 2, uint32, lowp>		lowp_u32mat4x2;
+	typedef mat<4, 3, uint32, lowp>		lowp_u32mat4x3;
+	typedef mat<4, 4, uint32, lowp>		lowp_u32mat4x4;
+
+	typedef mat<2, 2, uint32, mediump>	mediump_u32mat2x2;
+	typedef mat<2, 3, uint32, mediump>	mediump_u32mat2x3;
+	typedef mat<2, 4, uint32, mediump>	mediump_u32mat2x4;
+	typedef mat<3, 2, uint32, mediump>	mediump_u32mat3x2;
+	typedef mat<3, 3, uint32, mediump>	mediump_u32mat3x3;
+	typedef mat<3, 4, uint32, mediump>	mediump_u32mat3x4;
+	typedef mat<4, 2, uint32, mediump>	mediump_u32mat4x2;
+	typedef mat<4, 3, uint32, mediump>	mediump_u32mat4x3;
+	typedef mat<4, 4, uint32, mediump>	mediump_u32mat4x4;
+
+	typedef mat<2, 2, uint32, highp>	highp_u32mat2x2;
+	typedef mat<2, 3, uint32, highp>	highp_u32mat2x3;
+	typedef mat<2, 4, uint32, highp>	highp_u32mat2x4;
+	typedef mat<3, 2, uint32, highp>	highp_u32mat3x2;
+	typedef mat<3, 3, uint32, highp>	highp_u32mat3x3;
+	typedef mat<3, 4, uint32, highp>	highp_u32mat3x4;
+	typedef mat<4, 2, uint32, highp>	highp_u32mat4x2;
+	typedef mat<4, 3, uint32, highp>	highp_u32mat4x3;
+	typedef mat<4, 4, uint32, highp>	highp_u32mat4x4;
+
+	typedef mat<2, 2, uint32, defaultp>	u32mat2x2;
+	typedef mat<2, 3, uint32, defaultp>	u32mat2x3;
+	typedef mat<2, 4, uint32, defaultp>	u32mat2x4;
+	typedef mat<3, 2, uint32, defaultp>	u32mat3x2;
+	typedef mat<3, 3, uint32, defaultp>	u32mat3x3;
+	typedef mat<3, 4, uint32, defaultp>	u32mat3x4;
+	typedef mat<4, 2, uint32, defaultp>	u32mat4x2;
+	typedef mat<4, 3, uint32, defaultp>	u32mat4x3;
+	typedef mat<4, 4, uint32, defaultp>	u32mat4x4;
+
+
+	typedef mat<2, 2, uint64, lowp>		lowp_u64mat2x2;
+	typedef mat<2, 3, uint64, lowp>		lowp_u64mat2x3;
+	typedef mat<2, 4, uint64, lowp>		lowp_u64mat2x4;
+	typedef mat<3, 2, uint64, lowp>		lowp_u64mat3x2;
+	typedef mat<3, 3, uint64, lowp>		lowp_u64mat3x3;
+	typedef mat<3, 4, uint64, lowp>		lowp_u64mat3x4;
+	typedef mat<4, 2, uint64, lowp>		lowp_u64mat4x2;
+	typedef mat<4, 3, uint64, lowp>		lowp_u64mat4x3;
+	typedef mat<4, 4, uint64, lowp>		lowp_u64mat4x4;
+
+	typedef mat<2, 2, uint64, mediump>	mediump_u64mat2x2;
+	typedef mat<2, 3, uint64, mediump>	mediump_u64mat2x3;
+	typedef mat<2, 4, uint64, mediump>	mediump_u64mat2x4;
+	typedef mat<3, 2, uint64, mediump>	mediump_u64mat3x2;
+	typedef mat<3, 3, uint64, mediump>	mediump_u64mat3x3;
+	typedef mat<3, 4, uint64, mediump>	mediump_u64mat3x4;
+	typedef mat<4, 2, uint64, mediump>	mediump_u64mat4x2;
+	typedef mat<4, 3, uint64, mediump>	mediump_u64mat4x3;
+	typedef mat<4, 4, uint64, mediump>	mediump_u64mat4x4;
+
+	typedef mat<2, 2, uint64, highp>	highp_u64mat2x2;
+	typedef mat<2, 3, uint64, highp>	highp_u64mat2x3;
+	typedef mat<2, 4, uint64, highp>	highp_u64mat2x4;
+	typedef mat<3, 2, uint64, highp>	highp_u64mat3x2;
+	typedef mat<3, 3, uint64, highp>	highp_u64mat3x3;
+	typedef mat<3, 4, uint64, highp>	highp_u64mat3x4;
+	typedef mat<4, 2, uint64, highp>	highp_u64mat4x2;
+	typedef mat<4, 3, uint64, highp>	highp_u64mat4x3;
+	typedef mat<4, 4, uint64, highp>	highp_u64mat4x4;
+
+	typedef mat<2, 2, uint64, defaultp>	u64mat2x2;
+	typedef mat<2, 3, uint64, defaultp>	u64mat2x3;
+	typedef mat<2, 4, uint64, defaultp>	u64mat2x4;
+	typedef mat<3, 2, uint64, defaultp>	u64mat3x2;
+	typedef mat<3, 3, uint64, defaultp>	u64mat3x3;
+	typedef mat<3, 4, uint64, defaultp>	u64mat3x4;
+	typedef mat<4, 2, uint64, defaultp>	u64mat4x2;
+	typedef mat<4, 3, uint64, defaultp>	u64mat4x3;
+	typedef mat<4, 4, uint64, defaultp>	u64mat4x4;
+
+	// Quaternion
+
+	typedef qua<float, lowp>			lowp_quat;
+	typedef qua<float, mediump>			mediump_quat;
+	typedef qua<float, highp>			highp_quat;
+	typedef qua<float, defaultp>		quat;
+
+	typedef qua<float, lowp>			lowp_fquat;
+	typedef qua<float, mediump>			mediump_fquat;
+	typedef qua<float, highp>			highp_fquat;
+	typedef qua<float, defaultp>		fquat;
+
+	typedef qua<f32, lowp>				lowp_f32quat;
+	typedef qua<f32, mediump>			mediump_f32quat;
+	typedef qua<f32, highp>				highp_f32quat;
+	typedef qua<f32, defaultp>			f32quat;
+
+	typedef qua<double, lowp>			lowp_dquat;
+	typedef qua<double, mediump>		mediump_dquat;
+	typedef qua<double, highp>			highp_dquat;
+	typedef qua<double, defaultp>		dquat;
+
+	typedef qua<f64, lowp>				lowp_f64quat;
+	typedef qua<f64, mediump>			mediump_f64quat;
+	typedef qua<f64, highp>				highp_f64quat;
+	typedef qua<f64, defaultp>			f64quat;
+}//namespace glm
+
+

cuda_extensions/third_party/glm/glm/geometric.hpp

@@ -0,0 +1,116 @@
+/// @ref core
+/// @file glm/geometric.hpp
+///
+/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+///
+/// @defgroup core_func_geometric Geometric functions
+/// @ingroup core
+///
+/// These operate on vectors as vectors, not component-wise.
+///
+/// Include <glm/geometric.hpp> to use these core features.
+
+#pragma once
+
+#include "detail/type_vec3.hpp"
+
+namespace glm
+{
+	/// @addtogroup core_func_geometric
+	/// @{
+
+	/// Returns the length of x, i.e., sqrt(x * x).
+	///
+	/// @tparam L An integer between 1 and 4 included that qualify the dimension of the vector.
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/length.xml">GLSL length man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL T length(vec<L, T, Q> const& x);
+
+	/// Returns the distance between p0 and p1, i.e., length(p0 - p1).
+	///
+	/// @tparam L An integer between 1 and 4 included that qualify the dimension of the vector.
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/distance.xml">GLSL distance man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL T distance(vec<L, T, Q> const& p0, vec<L, T, Q> const& p1);
+
+	/// Returns the dot product of x and y, i.e., result = x * y.
+	///
+	/// @tparam L An integer between 1 and 4 included that qualify the dimension of the vector.
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/dot.xml">GLSL dot man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR T dot(vec<L, T, Q> const& x, vec<L, T, Q> const& y);
+
+	/// Returns the cross product of x and y.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/cross.xml">GLSL cross man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL GLM_CONSTEXPR vec<3, T, Q> cross(vec<3, T, Q> const& x, vec<3, T, Q> const& y);
+
+	/// Returns a vector in the same direction as x but with length of 1.
+	/// According to issue 10 GLSL 1.10 specification, if length(x) == 0 then result is undefined and generate an error.
+	///
+	/// @tparam L An integer between 1 and 4 included that qualify the dimension of the vector.
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/normalize.xml">GLSL normalize man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> normalize(vec<L, T, Q> const& x);
+
+	/// If dot(Nref, I) < 0.0, return N, otherwise, return -N.
+	///
+	/// @tparam L An integer between 1 and 4 included that qualify the dimension of the vector.
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/faceforward.xml">GLSL faceforward man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> faceforward(
+		vec<L, T, Q> const& N,
+		vec<L, T, Q> const& I,
+		vec<L, T, Q> const& Nref);
+
+	/// For the incident vector I and surface orientation N,
+	/// returns the reflection direction : result = I - 2.0 * dot(N, I) * N.
+	///
+	/// @tparam L An integer between 1 and 4 included that qualify the dimension of the vector.
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/reflect.xml">GLSL reflect man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> reflect(
+		vec<L, T, Q> const& I,
+		vec<L, T, Q> const& N);
+
+	/// For the incident vector I and surface normal N,
+	/// and the ratio of indices of refraction eta,
+	/// return the refraction vector.
+	///
+	/// @tparam L An integer between 1 and 4 included that qualify the dimension of the vector.
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/refract.xml">GLSL refract man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.5 Geometric Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> refract(
+		vec<L, T, Q> const& I,
+		vec<L, T, Q> const& N,
+		T eta);
+
+	/// @}
+}//namespace glm
+
+#include "detail/func_geometric.inl"

cuda_extensions/third_party/glm/glm/glm.hpp

@@ -0,0 +1,137 @@
+/// @ref core
+/// @file glm/glm.hpp
+///
+/// @mainpage OpenGL Mathematics (GLM)
+/// - Website: <a href="https://glm.g-truc.net">glm.g-truc.net</a>
+/// - <a href="modules.html">GLM API documentation</a>
+/// - <a href="https://github.com/g-truc/glm/blob/master/manual.md">GLM Manual</a>
+///
+/// @defgroup core Core features
+///
+/// @brief Features that implement in C++ the GLSL specification as closely as possible.
+///
+/// The GLM core consists of C++ types that mirror GLSL types and
+/// C++ functions that mirror the GLSL functions.
+///
+/// The best documentation for GLM Core is the current GLSL specification,
+/// <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.clean.pdf">version 4.2
+/// (pdf file)</a>.
+///
+/// GLM core functionalities require <glm/glm.hpp> to be included to be used.
+///
+///
+/// @defgroup core_vector Vector types
+///
+/// Vector types of two to four components with an exhaustive set of operators.
+///
+/// @ingroup core
+///
+///
+/// @defgroup core_vector_precision Vector types with precision qualifiers
+///
+/// @brief Vector types with precision qualifiers which may result in various precision in term of ULPs
+///
+/// GLSL allows defining qualifiers for particular variables.
+/// With OpenGL's GLSL, these qualifiers have no effect; they are there for compatibility,
+/// with OpenGL ES's GLSL, these qualifiers do have an effect.
+///
+/// C++ has no language equivalent to qualifier qualifiers. So GLM provides the next-best thing:
+/// a number of typedefs that use a particular qualifier.
+///
+/// None of these types make any guarantees about the actual qualifier used.
+///
+/// @ingroup core
+///
+///
+/// @defgroup core_matrix Matrix types
+///
+/// Matrix types of with C columns and R rows where C and R are values between 2 to 4 included.
+/// These types have exhaustive sets of operators.
+///
+/// @ingroup core
+///
+///
+/// @defgroup core_matrix_precision Matrix types with precision qualifiers
+///
+/// @brief Matrix types with precision qualifiers which may result in various precision in term of ULPs
+///
+/// GLSL allows defining qualifiers for particular variables.
+/// With OpenGL's GLSL, these qualifiers have no effect; they are there for compatibility,
+/// with OpenGL ES's GLSL, these qualifiers do have an effect.
+///
+/// C++ has no language equivalent to qualifier qualifiers. So GLM provides the next-best thing:
+/// a number of typedefs that use a particular qualifier.
+///
+/// None of these types make any guarantees about the actual qualifier used.
+///
+/// @ingroup core
+///
+///
+/// @defgroup ext Stable extensions
+///
+/// @brief Additional features not specified by GLSL specification.
+///
+/// EXT extensions are fully tested and documented.
+///
+/// Even if it's highly unrecommended, it's possible to include all the extensions at once by
+/// including <glm/ext.hpp>. Otherwise, each extension needs to be included  a specific file.
+///
+///
+/// @defgroup gtc Recommended extensions
+///
+/// @brief Additional features not specified by GLSL specification.
+///
+/// GTC extensions aim to be stable with tests and documentation.
+///
+/// Even if it's highly unrecommended, it's possible to include all the extensions at once by
+/// including <glm/ext.hpp>. Otherwise, each extension needs to be included  a specific file.
+///
+///
+/// @defgroup gtx Experimental extensions
+///
+/// @brief Experimental features not specified by GLSL specification.
+///
+/// Experimental extensions are useful functions and types, but the development of
+/// their API and functionality is not necessarily stable. They can change
+/// substantially between versions. Backwards compatibility is not much of an issue
+/// for them.
+///
+/// Even if it's highly unrecommended, it's possible to include all the extensions
+/// at once by including <glm/ext.hpp>. Otherwise, each extension needs to be
+/// included  a specific file.
+///
+
+#include "detail/_fixes.hpp"
+
+#include "detail/setup.hpp"
+
+#pragma once
+
+#include <cmath>
+#include <climits>
+#include <cfloat>
+#include <limits>
+#include <cassert>
+#include "fwd.hpp"
+
+#include "vec2.hpp"
+#include "vec3.hpp"
+#include "vec4.hpp"
+#include "mat2x2.hpp"
+#include "mat2x3.hpp"
+#include "mat2x4.hpp"
+#include "mat3x2.hpp"
+#include "mat3x3.hpp"
+#include "mat3x4.hpp"
+#include "mat4x2.hpp"
+#include "mat4x3.hpp"
+#include "mat4x4.hpp"
+
+#include "trigonometric.hpp"
+#include "exponential.hpp"
+#include "common.hpp"
+#include "packing.hpp"
+#include "geometric.hpp"
+#include "matrix.hpp"
+#include "vector_relational.hpp"
+#include "integer.hpp"

cuda_extensions/third_party/glm/glm/gtc/bitfield.hpp

@@ -0,0 +1,266 @@
+/// @ref gtc_bitfield
+/// @file glm/gtc/bitfield.hpp
+///
+/// @see core (dependence)
+/// @see gtc_bitfield (dependence)
+///
+/// @defgroup gtc_bitfield GLM_GTC_bitfield
+/// @ingroup gtc
+///
+/// Include <glm/gtc/bitfield.hpp> to use the features of this extension.
+///
+/// Allow to perform bit operations on integer values
+
+#include "../detail/setup.hpp"
+
+#pragma once
+
+// Dependencies
+#include "../ext/scalar_int_sized.hpp"
+#include "../ext/scalar_uint_sized.hpp"
+#include "../detail/qualifier.hpp"
+#include "../detail/_vectorize.hpp"
+#include "type_precision.hpp"
+#include <limits>
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_bitfield extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_bitfield
+	/// @{
+
+	/// Build a mask of 'count' bits
+	///
+	/// @see gtc_bitfield
+	template<typename genIUType>
+	GLM_FUNC_DECL genIUType mask(genIUType Bits);
+
+	/// Build a mask of 'count' bits
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed and unsigned integer scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see gtc_bitfield
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> mask(vec<L, T, Q> const& v);
+
+	/// Rotate all bits to the right. All the bits dropped in the right side are inserted back on the left side.
+	///
+	/// @see gtc_bitfield
+	template<typename genIUType>
+	GLM_FUNC_DECL genIUType bitfieldRotateRight(genIUType In, int Shift);
+
+	/// Rotate all bits to the right. All the bits dropped in the right side are inserted back on the left side.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed and unsigned integer scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see gtc_bitfield
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> bitfieldRotateRight(vec<L, T, Q> const& In, int Shift);
+
+	/// Rotate all bits to the left. All the bits dropped in the left side are inserted back on the right side.
+	///
+	/// @see gtc_bitfield
+	template<typename genIUType>
+	GLM_FUNC_DECL genIUType bitfieldRotateLeft(genIUType In, int Shift);
+
+	/// Rotate all bits to the left. All the bits dropped in the left side are inserted back on the right side.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed and unsigned integer scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see gtc_bitfield
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> bitfieldRotateLeft(vec<L, T, Q> const& In, int Shift);
+
+	/// Set to 1 a range of bits.
+	///
+	/// @see gtc_bitfield
+	template<typename genIUType>
+	GLM_FUNC_DECL genIUType bitfieldFillOne(genIUType Value, int FirstBit, int BitCount);
+
+	/// Set to 1 a range of bits.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed and unsigned integer scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see gtc_bitfield
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> bitfieldFillOne(vec<L, T, Q> const& Value, int FirstBit, int BitCount);
+
+	/// Set to 0 a range of bits.
+	///
+	/// @see gtc_bitfield
+	template<typename genIUType>
+	GLM_FUNC_DECL genIUType bitfieldFillZero(genIUType Value, int FirstBit, int BitCount);
+
+	/// Set to 0 a range of bits.
+	///
+	/// @tparam L Integer between 1 and 4 included that qualify the dimension of the vector
+	/// @tparam T Signed and unsigned integer scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see gtc_bitfield
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> bitfieldFillZero(vec<L, T, Q> const& Value, int FirstBit, int BitCount);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of x followed by the first bit of y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL int16 bitfieldInterleave(int8 x, int8 y);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of x followed by the first bit of y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint16 bitfieldInterleave(uint8 x, uint8 y);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of v.x followed by the first bit of v.y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint16 bitfieldInterleave(u8vec2 const& v);
+
+	/// Deinterleaves the bits of x.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL glm::u8vec2 bitfieldDeinterleave(glm::uint16 x);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of x followed by the first bit of y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL int32 bitfieldInterleave(int16 x, int16 y);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of x followed by the first bit of y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint32 bitfieldInterleave(uint16 x, uint16 y);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of v.x followed by the first bit of v.y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint32 bitfieldInterleave(u16vec2 const& v);
+
+	/// Deinterleaves the bits of x.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL glm::u16vec2 bitfieldDeinterleave(glm::uint32 x);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of x followed by the first bit of y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL int64 bitfieldInterleave(int32 x, int32 y);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of x followed by the first bit of y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint64 bitfieldInterleave(uint32 x, uint32 y);
+
+	/// Interleaves the bits of x and y.
+	/// The first bit is the first bit of v.x followed by the first bit of v.y.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint64 bitfieldInterleave(u32vec2 const& v);
+
+	/// Deinterleaves the bits of x.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL glm::u32vec2 bitfieldDeinterleave(glm::uint64 x);
+
+	/// Interleaves the bits of x, y and z.
+	/// The first bit is the first bit of x followed by the first bit of y and the first bit of z.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL int32 bitfieldInterleave(int8 x, int8 y, int8 z);
+
+	/// Interleaves the bits of x, y and z.
+	/// The first bit is the first bit of x followed by the first bit of y and the first bit of z.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint32 bitfieldInterleave(uint8 x, uint8 y, uint8 z);
+
+	/// Interleaves the bits of x, y and z.
+	/// The first bit is the first bit of x followed by the first bit of y and the first bit of z.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL int64 bitfieldInterleave(int16 x, int16 y, int16 z);
+
+	/// Interleaves the bits of x, y and z.
+	/// The first bit is the first bit of x followed by the first bit of y and the first bit of z.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint64 bitfieldInterleave(uint16 x, uint16 y, uint16 z);
+
+	/// Interleaves the bits of x, y and z.
+	/// The first bit is the first bit of x followed by the first bit of y and the first bit of z.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL int64 bitfieldInterleave(int32 x, int32 y, int32 z);
+
+	/// Interleaves the bits of x, y and z.
+	/// The first bit is the first bit of x followed by the first bit of y and the first bit of z.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint64 bitfieldInterleave(uint32 x, uint32 y, uint32 z);
+
+	/// Interleaves the bits of x, y, z and w.
+	/// The first bit is the first bit of x followed by the first bit of y, the first bit of z and finally the first bit of w.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL int32 bitfieldInterleave(int8 x, int8 y, int8 z, int8 w);
+
+	/// Interleaves the bits of x, y, z and w.
+	/// The first bit is the first bit of x followed by the first bit of y, the first bit of z and finally the first bit of w.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint32 bitfieldInterleave(uint8 x, uint8 y, uint8 z, uint8 w);
+
+	/// Interleaves the bits of x, y, z and w.
+	/// The first bit is the first bit of x followed by the first bit of y, the first bit of z and finally the first bit of w.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL int64 bitfieldInterleave(int16 x, int16 y, int16 z, int16 w);
+
+	/// Interleaves the bits of x, y, z and w.
+	/// The first bit is the first bit of x followed by the first bit of y, the first bit of z and finally the first bit of w.
+	/// The other bits are interleaved following the previous sequence.
+	///
+	/// @see gtc_bitfield
+	GLM_FUNC_DECL uint64 bitfieldInterleave(uint16 x, uint16 y, uint16 z, uint16 w);
+
+	/// @}
+} //namespace glm
+
+#include "bitfield.inl"

cuda_extensions/third_party/glm/glm/gtc/bitfield.inl

@@ -0,0 +1,626 @@
+/// @ref gtc_bitfield
+
+#include "../simd/integer.h"
+
+namespace glm{
+namespace detail
+{
+	template<typename PARAM, typename RET>
+	GLM_FUNC_DECL RET bitfieldInterleave(PARAM x, PARAM y);
+
+	template<typename PARAM, typename RET>
+	GLM_FUNC_DECL RET bitfieldInterleave(PARAM x, PARAM y, PARAM z);
+
+	template<typename PARAM, typename RET>
+	GLM_FUNC_DECL RET bitfieldInterleave(PARAM x, PARAM y, PARAM z, PARAM w);
+
+	template<>
+	GLM_FUNC_QUALIFIER glm::uint16 bitfieldInterleave(glm::uint8 x, glm::uint8 y)
+	{
+		glm::uint16 REG1(x);
+		glm::uint16 REG2(y);
+
+		REG1 = ((REG1 <<  4) | REG1) & static_cast<glm::uint16>(0x0F0F);
+		REG2 = ((REG2 <<  4) | REG2) & static_cast<glm::uint16>(0x0F0F);
+
+		REG1 = ((REG1 <<  2) | REG1) & static_cast<glm::uint16>(0x3333);
+		REG2 = ((REG2 <<  2) | REG2) & static_cast<glm::uint16>(0x3333);
+
+		REG1 = ((REG1 <<  1) | REG1) & static_cast<glm::uint16>(0x5555);
+		REG2 = ((REG2 <<  1) | REG2) & static_cast<glm::uint16>(0x5555);
+
+		return REG1 | static_cast<glm::uint16>(REG2 << 1);
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER glm::uint32 bitfieldInterleave(glm::uint16 x, glm::uint16 y)
+	{
+		glm::uint32 REG1(x);
+		glm::uint32 REG2(y);
+
+		REG1 = ((REG1 <<  8) | REG1) & static_cast<glm::uint32>(0x00FF00FF);
+		REG2 = ((REG2 <<  8) | REG2) & static_cast<glm::uint32>(0x00FF00FF);
+
+		REG1 = ((REG1 <<  4) | REG1) & static_cast<glm::uint32>(0x0F0F0F0F);
+		REG2 = ((REG2 <<  4) | REG2) & static_cast<glm::uint32>(0x0F0F0F0F);
+
+		REG1 = ((REG1 <<  2) | REG1) & static_cast<glm::uint32>(0x33333333);
+		REG2 = ((REG2 <<  2) | REG2) & static_cast<glm::uint32>(0x33333333);
+
+		REG1 = ((REG1 <<  1) | REG1) & static_cast<glm::uint32>(0x55555555);
+		REG2 = ((REG2 <<  1) | REG2) & static_cast<glm::uint32>(0x55555555);
+
+		return REG1 | (REG2 << 1);
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER glm::uint64 bitfieldInterleave(glm::uint32 x, glm::uint32 y)
+	{
+		glm::uint64 REG1(x);
+		glm::uint64 REG2(y);
+
+		REG1 = ((REG1 << 16) | REG1) & static_cast<glm::uint64>(0x0000FFFF0000FFFFull);
+		REG2 = ((REG2 << 16) | REG2) & static_cast<glm::uint64>(0x0000FFFF0000FFFFull);
+
+		REG1 = ((REG1 <<  8) | REG1) & static_cast<glm::uint64>(0x00FF00FF00FF00FFull);
+		REG2 = ((REG2 <<  8) | REG2) & static_cast<glm::uint64>(0x00FF00FF00FF00FFull);
+
+		REG1 = ((REG1 <<  4) | REG1) & static_cast<glm::uint64>(0x0F0F0F0F0F0F0F0Full);
+		REG2 = ((REG2 <<  4) | REG2) & static_cast<glm::uint64>(0x0F0F0F0F0F0F0F0Full);
+
+		REG1 = ((REG1 <<  2) | REG1) & static_cast<glm::uint64>(0x3333333333333333ull);
+		REG2 = ((REG2 <<  2) | REG2) & static_cast<glm::uint64>(0x3333333333333333ull);
+
+		REG1 = ((REG1 <<  1) | REG1) & static_cast<glm::uint64>(0x5555555555555555ull);
+		REG2 = ((REG2 <<  1) | REG2) & static_cast<glm::uint64>(0x5555555555555555ull);
+
+		return REG1 | (REG2 << 1);
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER glm::uint32 bitfieldInterleave(glm::uint8 x, glm::uint8 y, glm::uint8 z)
+	{
+		glm::uint32 REG1(x);
+		glm::uint32 REG2(y);
+		glm::uint32 REG3(z);
+
+		REG1 = ((REG1 << 16) | REG1) & static_cast<glm::uint32>(0xFF0000FFu);
+		REG2 = ((REG2 << 16) | REG2) & static_cast<glm::uint32>(0xFF0000FFu);
+		REG3 = ((REG3 << 16) | REG3) & static_cast<glm::uint32>(0xFF0000FFu);
+
+		REG1 = ((REG1 <<  8) | REG1) & static_cast<glm::uint32>(0x0F00F00Fu);
+		REG2 = ((REG2 <<  8) | REG2) & static_cast<glm::uint32>(0x0F00F00Fu);
+		REG3 = ((REG3 <<  8) | REG3) & static_cast<glm::uint32>(0x0F00F00Fu);
+
+		REG1 = ((REG1 <<  4) | REG1) & static_cast<glm::uint32>(0xC30C30C3u);
+		REG2 = ((REG2 <<  4) | REG2) & static_cast<glm::uint32>(0xC30C30C3u);
+		REG3 = ((REG3 <<  4) | REG3) & static_cast<glm::uint32>(0xC30C30C3u);
+
+		REG1 = ((REG1 <<  2) | REG1) & static_cast<glm::uint32>(0x49249249u);
+		REG2 = ((REG2 <<  2) | REG2) & static_cast<glm::uint32>(0x49249249u);
+		REG3 = ((REG3 <<  2) | REG3) & static_cast<glm::uint32>(0x49249249u);
+
+		return REG1 | (REG2 << 1) | (REG3 << 2);
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER glm::uint64 bitfieldInterleave(glm::uint16 x, glm::uint16 y, glm::uint16 z)
+	{
+		glm::uint64 REG1(x);
+		glm::uint64 REG2(y);
+		glm::uint64 REG3(z);
+
+		REG1 = ((REG1 << 32) | REG1) & static_cast<glm::uint64>(0xFFFF00000000FFFFull);
+		REG2 = ((REG2 << 32) | REG2) & static_cast<glm::uint64>(0xFFFF00000000FFFFull);
+		REG3 = ((REG3 << 32) | REG3) & static_cast<glm::uint64>(0xFFFF00000000FFFFull);
+
+		REG1 = ((REG1 << 16) | REG1) & static_cast<glm::uint64>(0x00FF0000FF0000FFull);
+		REG2 = ((REG2 << 16) | REG2) & static_cast<glm::uint64>(0x00FF0000FF0000FFull);
+		REG3 = ((REG3 << 16) | REG3) & static_cast<glm::uint64>(0x00FF0000FF0000FFull);
+
+		REG1 = ((REG1 <<  8) | REG1) & static_cast<glm::uint64>(0xF00F00F00F00F00Full);
+		REG2 = ((REG2 <<  8) | REG2) & static_cast<glm::uint64>(0xF00F00F00F00F00Full);
+		REG3 = ((REG3 <<  8) | REG3) & static_cast<glm::uint64>(0xF00F00F00F00F00Full);
+
+		REG1 = ((REG1 <<  4) | REG1) & static_cast<glm::uint64>(0x30C30C30C30C30C3ull);
+		REG2 = ((REG2 <<  4) | REG2) & static_cast<glm::uint64>(0x30C30C30C30C30C3ull);
+		REG3 = ((REG3 <<  4) | REG3) & static_cast<glm::uint64>(0x30C30C30C30C30C3ull);
+
+		REG1 = ((REG1 <<  2) | REG1) & static_cast<glm::uint64>(0x9249249249249249ull);
+		REG2 = ((REG2 <<  2) | REG2) & static_cast<glm::uint64>(0x9249249249249249ull);
+		REG3 = ((REG3 <<  2) | REG3) & static_cast<glm::uint64>(0x9249249249249249ull);
+
+		return REG1 | (REG2 << 1) | (REG3 << 2);
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER glm::uint64 bitfieldInterleave(glm::uint32 x, glm::uint32 y, glm::uint32 z)
+	{
+		glm::uint64 REG1(x);
+		glm::uint64 REG2(y);
+		glm::uint64 REG3(z);
+
+		REG1 = ((REG1 << 32) | REG1) & static_cast<glm::uint64>(0xFFFF00000000FFFFull);
+		REG2 = ((REG2 << 32) | REG2) & static_cast<glm::uint64>(0xFFFF00000000FFFFull);
+		REG3 = ((REG3 << 32) | REG3) & static_cast<glm::uint64>(0xFFFF00000000FFFFull);
+
+		REG1 = ((REG1 << 16) | REG1) & static_cast<glm::uint64>(0x00FF0000FF0000FFull);
+		REG2 = ((REG2 << 16) | REG2) & static_cast<glm::uint64>(0x00FF0000FF0000FFull);
+		REG3 = ((REG3 << 16) | REG3) & static_cast<glm::uint64>(0x00FF0000FF0000FFull);
+
+		REG1 = ((REG1 <<  8) | REG1) & static_cast<glm::uint64>(0xF00F00F00F00F00Full);
+		REG2 = ((REG2 <<  8) | REG2) & static_cast<glm::uint64>(0xF00F00F00F00F00Full);
+		REG3 = ((REG3 <<  8) | REG3) & static_cast<glm::uint64>(0xF00F00F00F00F00Full);
+
+		REG1 = ((REG1 <<  4) | REG1) & static_cast<glm::uint64>(0x30C30C30C30C30C3ull);
+		REG2 = ((REG2 <<  4) | REG2) & static_cast<glm::uint64>(0x30C30C30C30C30C3ull);
+		REG3 = ((REG3 <<  4) | REG3) & static_cast<glm::uint64>(0x30C30C30C30C30C3ull);
+
+		REG1 = ((REG1 <<  2) | REG1) & static_cast<glm::uint64>(0x9249249249249249ull);
+		REG2 = ((REG2 <<  2) | REG2) & static_cast<glm::uint64>(0x9249249249249249ull);
+		REG3 = ((REG3 <<  2) | REG3) & static_cast<glm::uint64>(0x9249249249249249ull);
+
+		return REG1 | (REG2 << 1) | (REG3 << 2);
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER glm::uint32 bitfieldInterleave(glm::uint8 x, glm::uint8 y, glm::uint8 z, glm::uint8 w)
+	{
+		glm::uint32 REG1(x);
+		glm::uint32 REG2(y);
+		glm::uint32 REG3(z);
+		glm::uint32 REG4(w);
+
+		REG1 = ((REG1 << 12) | REG1) & static_cast<glm::uint32>(0x000F000Fu);
+		REG2 = ((REG2 << 12) | REG2) & static_cast<glm::uint32>(0x000F000Fu);
+		REG3 = ((REG3 << 12) | REG3) & static_cast<glm::uint32>(0x000F000Fu);
+		REG4 = ((REG4 << 12) | REG4) & static_cast<glm::uint32>(0x000F000Fu);
+
+		REG1 = ((REG1 <<  6) | REG1) & static_cast<glm::uint32>(0x03030303u);
+		REG2 = ((REG2 <<  6) | REG2) & static_cast<glm::uint32>(0x03030303u);
+		REG3 = ((REG3 <<  6) | REG3) & static_cast<glm::uint32>(0x03030303u);
+		REG4 = ((REG4 <<  6) | REG4) & static_cast<glm::uint32>(0x03030303u);
+
+		REG1 = ((REG1 <<  3) | REG1) & static_cast<glm::uint32>(0x11111111u);
+		REG2 = ((REG2 <<  3) | REG2) & static_cast<glm::uint32>(0x11111111u);
+		REG3 = ((REG3 <<  3) | REG3) & static_cast<glm::uint32>(0x11111111u);
+		REG4 = ((REG4 <<  3) | REG4) & static_cast<glm::uint32>(0x11111111u);
+
+		return REG1 | (REG2 << 1) | (REG3 << 2) | (REG4 << 3);
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER glm::uint64 bitfieldInterleave(glm::uint16 x, glm::uint16 y, glm::uint16 z, glm::uint16 w)
+	{
+		glm::uint64 REG1(x);
+		glm::uint64 REG2(y);
+		glm::uint64 REG3(z);
+		glm::uint64 REG4(w);
+
+		REG1 = ((REG1 << 24) | REG1) & static_cast<glm::uint64>(0x000000FF000000FFull);
+		REG2 = ((REG2 << 24) | REG2) & static_cast<glm::uint64>(0x000000FF000000FFull);
+		REG3 = ((REG3 << 24) | REG3) & static_cast<glm::uint64>(0x000000FF000000FFull);
+		REG4 = ((REG4 << 24) | REG4) & static_cast<glm::uint64>(0x000000FF000000FFull);
+
+		REG1 = ((REG1 << 12) | REG1) & static_cast<glm::uint64>(0x000F000F000F000Full);
+		REG2 = ((REG2 << 12) | REG2) & static_cast<glm::uint64>(0x000F000F000F000Full);
+		REG3 = ((REG3 << 12) | REG3) & static_cast<glm::uint64>(0x000F000F000F000Full);
+		REG4 = ((REG4 << 12) | REG4) & static_cast<glm::uint64>(0x000F000F000F000Full);
+
+		REG1 = ((REG1 <<  6) | REG1) & static_cast<glm::uint64>(0x0303030303030303ull);
+		REG2 = ((REG2 <<  6) | REG2) & static_cast<glm::uint64>(0x0303030303030303ull);
+		REG3 = ((REG3 <<  6) | REG3) & static_cast<glm::uint64>(0x0303030303030303ull);
+		REG4 = ((REG4 <<  6) | REG4) & static_cast<glm::uint64>(0x0303030303030303ull);
+
+		REG1 = ((REG1 <<  3) | REG1) & static_cast<glm::uint64>(0x1111111111111111ull);
+		REG2 = ((REG2 <<  3) | REG2) & static_cast<glm::uint64>(0x1111111111111111ull);
+		REG3 = ((REG3 <<  3) | REG3) & static_cast<glm::uint64>(0x1111111111111111ull);
+		REG4 = ((REG4 <<  3) | REG4) & static_cast<glm::uint64>(0x1111111111111111ull);
+
+		return REG1 | (REG2 << 1) | (REG3 << 2) | (REG4 << 3);
+	}
+}//namespace detail
+
+	template<typename genIUType>
+	GLM_FUNC_QUALIFIER genIUType mask(genIUType Bits)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<genIUType>::is_integer, "'mask' accepts only integer values");
+
+		return Bits >= sizeof(genIUType) * 8 ? ~static_cast<genIUType>(0) : (static_cast<genIUType>(1) << Bits) - static_cast<genIUType>(1);
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> mask(vec<L, T, Q> const& v)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'mask' accepts only integer values");
+
+		return detail::functor1<vec, L, T, T, Q>::call(mask, v);
+	}
+
+	template<typename genIType>
+	GLM_FUNC_QUALIFIER genIType bitfieldRotateRight(genIType In, int Shift)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<genIType>::is_integer, "'bitfieldRotateRight' accepts only integer values");
+
+		int const BitSize = static_cast<genIType>(sizeof(genIType) * 8);
+		return (In << static_cast<genIType>(Shift)) | (In >> static_cast<genIType>(BitSize - Shift));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> bitfieldRotateRight(vec<L, T, Q> const& In, int Shift)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'bitfieldRotateRight' accepts only integer values");
+
+		int const BitSize = static_cast<int>(sizeof(T) * 8);
+		return (In << static_cast<T>(Shift)) | (In >> static_cast<T>(BitSize - Shift));
+	}
+
+	template<typename genIType>
+	GLM_FUNC_QUALIFIER genIType bitfieldRotateLeft(genIType In, int Shift)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<genIType>::is_integer, "'bitfieldRotateLeft' accepts only integer values");
+
+		int const BitSize = static_cast<genIType>(sizeof(genIType) * 8);
+		return (In >> static_cast<genIType>(Shift)) | (In << static_cast<genIType>(BitSize - Shift));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> bitfieldRotateLeft(vec<L, T, Q> const& In, int Shift)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<T>::is_integer, "'bitfieldRotateLeft' accepts only integer values");
+
+		int const BitSize = static_cast<int>(sizeof(T) * 8);
+		return (In >> static_cast<T>(Shift)) | (In << static_cast<T>(BitSize - Shift));
+	}
+
+	template<typename genIUType>
+	GLM_FUNC_QUALIFIER genIUType bitfieldFillOne(genIUType Value, int FirstBit, int BitCount)
+	{
+		return Value | static_cast<genIUType>(mask(BitCount) << FirstBit);
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> bitfieldFillOne(vec<L, T, Q> const& Value, int FirstBit, int BitCount)
+	{
+		return Value | static_cast<T>(mask(BitCount) << FirstBit);
+	}
+
+	template<typename genIUType>
+	GLM_FUNC_QUALIFIER genIUType bitfieldFillZero(genIUType Value, int FirstBit, int BitCount)
+	{
+		return Value & static_cast<genIUType>(~(mask(BitCount) << FirstBit));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> bitfieldFillZero(vec<L, T, Q> const& Value, int FirstBit, int BitCount)
+	{
+		return Value & static_cast<T>(~(mask(BitCount) << FirstBit));
+	}
+
+	GLM_FUNC_QUALIFIER int16 bitfieldInterleave(int8 x, int8 y)
+	{
+		union sign8
+		{
+			int8 i;
+			uint8 u;
+		} sign_x, sign_y;
+
+		union sign16
+		{
+			int16 i;
+			uint16 u;
+		} result;
+
+		sign_x.i = x;
+		sign_y.i = y;
+		result.u = bitfieldInterleave(sign_x.u, sign_y.u);
+
+		return result.i;
+	}
+
+	GLM_FUNC_QUALIFIER uint16 bitfieldInterleave(uint8 x, uint8 y)
+	{
+		return detail::bitfieldInterleave<uint8, uint16>(x, y);
+	}
+
+	GLM_FUNC_QUALIFIER uint16 bitfieldInterleave(u8vec2 const& v)
+	{
+		return detail::bitfieldInterleave<uint8, uint16>(v.x, v.y);
+	}
+
+	GLM_FUNC_QUALIFIER u8vec2 bitfieldDeinterleave(glm::uint16 x)
+	{
+		uint16 REG1(x);
+		uint16 REG2(x >>= 1);
+
+		REG1 = REG1 & static_cast<uint16>(0x5555);
+		REG2 = REG2 & static_cast<uint16>(0x5555);
+
+		REG1 = ((REG1 >> 1) | REG1) & static_cast<uint16>(0x3333);
+		REG2 = ((REG2 >> 1) | REG2) & static_cast<uint16>(0x3333);
+
+		REG1 = ((REG1 >> 2) | REG1) & static_cast<uint16>(0x0F0F);
+		REG2 = ((REG2 >> 2) | REG2) & static_cast<uint16>(0x0F0F);
+
+		REG1 = ((REG1 >> 4) | REG1) & static_cast<uint16>(0x00FF);
+		REG2 = ((REG2 >> 4) | REG2) & static_cast<uint16>(0x00FF);
+
+		REG1 = ((REG1 >> 8) | REG1) & static_cast<uint16>(0xFFFF);
+		REG2 = ((REG2 >> 8) | REG2) & static_cast<uint16>(0xFFFF);
+
+		return glm::u8vec2(REG1, REG2);
+	}
+
+	GLM_FUNC_QUALIFIER int32 bitfieldInterleave(int16 x, int16 y)
+	{
+		union sign16
+		{
+			int16 i;
+			uint16 u;
+		} sign_x, sign_y;
+
+		union sign32
+		{
+			int32 i;
+			uint32 u;
+		} result;
+
+		sign_x.i = x;
+		sign_y.i = y;
+		result.u = bitfieldInterleave(sign_x.u, sign_y.u);
+
+		return result.i;
+	}
+
+	GLM_FUNC_QUALIFIER uint32 bitfieldInterleave(uint16 x, uint16 y)
+	{
+		return detail::bitfieldInterleave<uint16, uint32>(x, y);
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint32 bitfieldInterleave(u16vec2 const& v)
+	{
+		return detail::bitfieldInterleave<uint16, uint32>(v.x, v.y);
+	}
+
+	GLM_FUNC_QUALIFIER glm::u16vec2 bitfieldDeinterleave(glm::uint32 x)
+	{
+		glm::uint32 REG1(x);
+		glm::uint32 REG2(x >>= 1);
+
+		REG1 = REG1 & static_cast<glm::uint32>(0x55555555);
+		REG2 = REG2 & static_cast<glm::uint32>(0x55555555);
+
+		REG1 = ((REG1 >> 1) | REG1) & static_cast<glm::uint32>(0x33333333);
+		REG2 = ((REG2 >> 1) | REG2) & static_cast<glm::uint32>(0x33333333);
+
+		REG1 = ((REG1 >> 2) | REG1) & static_cast<glm::uint32>(0x0F0F0F0F);
+		REG2 = ((REG2 >> 2) | REG2) & static_cast<glm::uint32>(0x0F0F0F0F);
+
+		REG1 = ((REG1 >> 4) | REG1) & static_cast<glm::uint32>(0x00FF00FF);
+		REG2 = ((REG2 >> 4) | REG2) & static_cast<glm::uint32>(0x00FF00FF);
+
+		REG1 = ((REG1 >> 8) | REG1) & static_cast<glm::uint32>(0x0000FFFF);
+		REG2 = ((REG2 >> 8) | REG2) & static_cast<glm::uint32>(0x0000FFFF);
+
+		return glm::u16vec2(REG1, REG2);
+	}
+
+	GLM_FUNC_QUALIFIER int64 bitfieldInterleave(int32 x, int32 y)
+	{
+		union sign32
+		{
+			int32 i;
+			uint32 u;
+		} sign_x, sign_y;
+
+		union sign64
+		{
+			int64 i;
+			uint64 u;
+		} result;
+
+		sign_x.i = x;
+		sign_y.i = y;
+		result.u = bitfieldInterleave(sign_x.u, sign_y.u);
+
+		return result.i;
+	}
+
+	GLM_FUNC_QUALIFIER uint64 bitfieldInterleave(uint32 x, uint32 y)
+	{
+		return detail::bitfieldInterleave<uint32, uint64>(x, y);
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint64 bitfieldInterleave(u32vec2 const& v)
+	{
+		return detail::bitfieldInterleave<uint32, uint64>(v.x, v.y);
+	}
+
+	GLM_FUNC_QUALIFIER glm::u32vec2 bitfieldDeinterleave(glm::uint64 x)
+	{
+		glm::uint64 REG1(x);
+		glm::uint64 REG2(x >>= 1);
+
+		REG1 = REG1 & static_cast<glm::uint64>(0x5555555555555555ull);
+		REG2 = REG2 & static_cast<glm::uint64>(0x5555555555555555ull);
+
+		REG1 = ((REG1 >> 1) | REG1) & static_cast<glm::uint64>(0x3333333333333333ull);
+		REG2 = ((REG2 >> 1) | REG2) & static_cast<glm::uint64>(0x3333333333333333ull);
+
+		REG1 = ((REG1 >> 2) | REG1) & static_cast<glm::uint64>(0x0F0F0F0F0F0F0F0Full);
+		REG2 = ((REG2 >> 2) | REG2) & static_cast<glm::uint64>(0x0F0F0F0F0F0F0F0Full);
+
+		REG1 = ((REG1 >> 4) | REG1) & static_cast<glm::uint64>(0x00FF00FF00FF00FFull);
+		REG2 = ((REG2 >> 4) | REG2) & static_cast<glm::uint64>(0x00FF00FF00FF00FFull);
+
+		REG1 = ((REG1 >> 8) | REG1) & static_cast<glm::uint64>(0x0000FFFF0000FFFFull);
+		REG2 = ((REG2 >> 8) | REG2) & static_cast<glm::uint64>(0x0000FFFF0000FFFFull);
+
+		REG1 = ((REG1 >> 16) | REG1) & static_cast<glm::uint64>(0x00000000FFFFFFFFull);
+		REG2 = ((REG2 >> 16) | REG2) & static_cast<glm::uint64>(0x00000000FFFFFFFFull);
+
+		return glm::u32vec2(REG1, REG2);
+	}
+
+	GLM_FUNC_QUALIFIER int32 bitfieldInterleave(int8 x, int8 y, int8 z)
+	{
+		union sign8
+		{
+			int8 i;
+			uint8 u;
+		} sign_x, sign_y, sign_z;
+
+		union sign32
+		{
+			int32 i;
+			uint32 u;
+		} result;
+
+		sign_x.i = x;
+		sign_y.i = y;
+		sign_z.i = z;
+		result.u = bitfieldInterleave(sign_x.u, sign_y.u, sign_z.u);
+
+		return result.i;
+	}
+
+	GLM_FUNC_QUALIFIER uint32 bitfieldInterleave(uint8 x, uint8 y, uint8 z)
+	{
+		return detail::bitfieldInterleave<uint8, uint32>(x, y, z);
+	}
+
+	GLM_FUNC_QUALIFIER uint32 bitfieldInterleave(u8vec3 const& v)
+	{
+		return detail::bitfieldInterleave<uint8, uint32>(v.x, v.y, v.z);
+	}
+
+	GLM_FUNC_QUALIFIER int64 bitfieldInterleave(int16 x, int16 y, int16 z)
+	{
+		union sign16
+		{
+			int16 i;
+			uint16 u;
+		} sign_x, sign_y, sign_z;
+
+		union sign64
+		{
+			int64 i;
+			uint64 u;
+		} result;
+
+		sign_x.i = x;
+		sign_y.i = y;
+		sign_z.i = z;
+		result.u = bitfieldInterleave(sign_x.u, sign_y.u, sign_z.u);
+
+		return result.i;
+	}
+
+	GLM_FUNC_QUALIFIER uint64 bitfieldInterleave(uint16 x, uint16 y, uint16 z)
+	{
+		return detail::bitfieldInterleave<uint32, uint64>(x, y, z);
+	}
+
+	GLM_FUNC_QUALIFIER uint64 bitfieldInterleave(u16vec3 const& v)
+	{
+		return detail::bitfieldInterleave<uint32, uint64>(v.x, v.y, v.z);
+	}
+
+	GLM_FUNC_QUALIFIER int64 bitfieldInterleave(int32 x, int32 y, int32 z)
+	{
+		union sign16
+		{
+			int32 i;
+			uint32 u;
+		} sign_x, sign_y, sign_z;
+
+		union sign64
+		{
+			int64 i;
+			uint64 u;
+		} result;
+
+		sign_x.i = x;
+		sign_y.i = y;
+		sign_z.i = z;
+		result.u = bitfieldInterleave(sign_x.u, sign_y.u, sign_z.u);
+
+		return result.i;
+	}
+
+	GLM_FUNC_QUALIFIER uint64 bitfieldInterleave(uint32 x, uint32 y, uint32 z)
+	{
+		return detail::bitfieldInterleave<uint32, uint64>(x, y, z);
+	}
+
+	GLM_FUNC_QUALIFIER uint64 bitfieldInterleave(u32vec3 const& v)
+	{
+		return detail::bitfieldInterleave<uint32, uint64>(v.x, v.y, v.z);
+	}
+
+	GLM_FUNC_QUALIFIER int32 bitfieldInterleave(int8 x, int8 y, int8 z, int8 w)
+	{
+		union sign8
+		{
+			int8 i;
+			uint8 u;
+		} sign_x, sign_y, sign_z, sign_w;
+
+		union sign32
+		{
+			int32 i;
+			uint32 u;
+		} result;
+
+		sign_x.i = x;
+		sign_y.i = y;
+		sign_z.i = z;
+		sign_w.i = w;
+		result.u = bitfieldInterleave(sign_x.u, sign_y.u, sign_z.u, sign_w.u);
+
+		return result.i;
+	}
+
+	GLM_FUNC_QUALIFIER uint32 bitfieldInterleave(uint8 x, uint8 y, uint8 z, uint8 w)
+	{
+		return detail::bitfieldInterleave<uint8, uint32>(x, y, z, w);
+	}
+
+	GLM_FUNC_QUALIFIER uint32 bitfieldInterleave(u8vec4 const& v)
+	{
+		return detail::bitfieldInterleave<uint8, uint32>(v.x, v.y, v.z, v.w);
+	}
+
+	GLM_FUNC_QUALIFIER int64 bitfieldInterleave(int16 x, int16 y, int16 z, int16 w)
+	{
+		union sign16
+		{
+			int16 i;
+			uint16 u;
+		} sign_x, sign_y, sign_z, sign_w;
+
+		union sign64
+		{
+			int64 i;
+			uint64 u;
+		} result;
+
+		sign_x.i = x;
+		sign_y.i = y;
+		sign_z.i = z;
+		sign_w.i = w;
+		result.u = bitfieldInterleave(sign_x.u, sign_y.u, sign_z.u, sign_w.u);
+
+		return result.i;
+	}
+
+	GLM_FUNC_QUALIFIER uint64 bitfieldInterleave(uint16 x, uint16 y, uint16 z, uint16 w)
+	{
+		return detail::bitfieldInterleave<uint16, uint64>(x, y, z, w);
+	}
+
+	GLM_FUNC_QUALIFIER uint64 bitfieldInterleave(u16vec4 const& v)
+	{
+		return detail::bitfieldInterleave<uint16, uint64>(v.x, v.y, v.z, v.w);
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/gtc/color_space.hpp

@@ -0,0 +1,56 @@
+/// @ref gtc_color_space
+/// @file glm/gtc/color_space.hpp
+///
+/// @see core (dependence)
+/// @see gtc_color_space (dependence)
+///
+/// @defgroup gtc_color_space GLM_GTC_color_space
+/// @ingroup gtc
+///
+/// Include <glm/gtc/color_space.hpp> to use the features of this extension.
+///
+/// Allow to perform bit operations on integer values
+
+#pragma once
+
+// Dependencies
+#include "../detail/setup.hpp"
+#include "../detail/qualifier.hpp"
+#include "../exponential.hpp"
+#include "../vec3.hpp"
+#include "../vec4.hpp"
+#include <limits>
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_color_space extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_color_space
+	/// @{
+
+	/// Convert a linear color to sRGB color using a standard gamma correction.
+	/// IEC 61966-2-1:1999 / Rec. 709 specification https://www.w3.org/Graphics/Color/srgb
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> convertLinearToSRGB(vec<L, T, Q> const& ColorLinear);
+
+	/// Convert a linear color to sRGB color using a custom gamma correction.
+	/// IEC 61966-2-1:1999 / Rec. 709 specification https://www.w3.org/Graphics/Color/srgb
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> convertLinearToSRGB(vec<L, T, Q> const& ColorLinear, T Gamma);
+
+	/// Convert a sRGB color to linear color using a standard gamma correction.
+	/// IEC 61966-2-1:1999 / Rec. 709 specification https://www.w3.org/Graphics/Color/srgb
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> convertSRGBToLinear(vec<L, T, Q> const& ColorSRGB);
+
+	/// Convert a sRGB color to linear color using a custom gamma correction.
+	// IEC 61966-2-1:1999 / Rec. 709 specification https://www.w3.org/Graphics/Color/srgb
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> convertSRGBToLinear(vec<L, T, Q> const& ColorSRGB, T Gamma);
+
+	/// @}
+} //namespace glm
+
+#include "color_space.inl"

cuda_extensions/third_party/glm/glm/gtc/color_space.inl

@@ -0,0 +1,84 @@
+/// @ref gtc_color_space
+
+namespace glm{
+namespace detail
+{
+	template<length_t L, typename T, qualifier Q>
+	struct compute_rgbToSrgb
+	{
+		GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& ColorRGB, T GammaCorrection)
+		{
+			vec<L, T, Q> const ClampedColor(clamp(ColorRGB, static_cast<T>(0), static_cast<T>(1)));
+
+			return mix(
+				pow(ClampedColor, vec<L, T, Q>(GammaCorrection)) * static_cast<T>(1.055) - static_cast<T>(0.055),
+				ClampedColor * static_cast<T>(12.92),
+				lessThan(ClampedColor, vec<L, T, Q>(static_cast<T>(0.0031308))));
+		}
+	};
+
+	template<typename T, qualifier Q>
+	struct compute_rgbToSrgb<4, T, Q>
+	{
+		GLM_FUNC_QUALIFIER static vec<4, T, Q> call(vec<4, T, Q> const& ColorRGB, T GammaCorrection)
+		{
+			return vec<4, T, Q>(compute_rgbToSrgb<3, T, Q>::call(vec<3, T, Q>(ColorRGB), GammaCorrection), ColorRGB.w);
+		}
+	};
+
+	template<length_t L, typename T, qualifier Q>
+	struct compute_srgbToRgb
+	{
+		GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& ColorSRGB, T Gamma)
+		{
+			return mix(
+				pow((ColorSRGB + static_cast<T>(0.055)) * static_cast<T>(0.94786729857819905213270142180095), vec<L, T, Q>(Gamma)),
+				ColorSRGB * static_cast<T>(0.07739938080495356037151702786378),
+				lessThanEqual(ColorSRGB, vec<L, T, Q>(static_cast<T>(0.04045))));
+		}
+	};
+
+	template<typename T, qualifier Q>
+	struct compute_srgbToRgb<4, T, Q>
+	{
+		GLM_FUNC_QUALIFIER static vec<4, T, Q> call(vec<4, T, Q> const& ColorSRGB, T Gamma)
+		{
+			return vec<4, T, Q>(compute_srgbToRgb<3, T, Q>::call(vec<3, T, Q>(ColorSRGB), Gamma), ColorSRGB.w);
+		}
+	};
+}//namespace detail
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> convertLinearToSRGB(vec<L, T, Q> const& ColorLinear)
+	{
+		return detail::compute_rgbToSrgb<L, T, Q>::call(ColorLinear, static_cast<T>(0.41666));
+	}
+
+	// Based on Ian Taylor http://chilliant.blogspot.fr/2012/08/srgb-approximations-for-hlsl.html
+	template<>
+	GLM_FUNC_QUALIFIER vec<3, float, lowp> convertLinearToSRGB(vec<3, float, lowp> const& ColorLinear)
+	{
+		vec<3, float, lowp> S1 = sqrt(ColorLinear);
+		vec<3, float, lowp> S2 = sqrt(S1);
+		vec<3, float, lowp> S3 = sqrt(S2);
+		return 0.662002687f * S1 + 0.684122060f * S2 - 0.323583601f * S3 - 0.0225411470f * ColorLinear;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> convertLinearToSRGB(vec<L, T, Q> const& ColorLinear, T Gamma)
+	{
+		return detail::compute_rgbToSrgb<L, T, Q>::call(ColorLinear, static_cast<T>(1) / Gamma);
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> convertSRGBToLinear(vec<L, T, Q> const& ColorSRGB)
+	{
+		return detail::compute_srgbToRgb<L, T, Q>::call(ColorSRGB, static_cast<T>(2.4));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, T, Q> convertSRGBToLinear(vec<L, T, Q> const& ColorSRGB, T Gamma)
+	{
+		return detail::compute_srgbToRgb<L, T, Q>::call(ColorSRGB, Gamma);
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/gtc/constants.hpp

@@ -0,0 +1,165 @@
+/// @ref gtc_constants
+/// @file glm/gtc/constants.hpp
+///
+/// @see core (dependence)
+///
+/// @defgroup gtc_constants GLM_GTC_constants
+/// @ingroup gtc
+///
+/// Include <glm/gtc/constants.hpp> to use the features of this extension.
+///
+/// Provide a list of constants and precomputed useful values.
+
+#pragma once
+
+// Dependencies
+#include "../ext/scalar_constants.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_constants extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_constants
+	/// @{
+
+	/// Return 0.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType zero();
+
+	/// Return 1.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType one();
+
+	/// Return pi * 2.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType two_pi();
+
+	/// Return square root of pi.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType root_pi();
+
+	/// Return pi / 2.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType half_pi();
+
+	/// Return pi / 2 * 3.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType three_over_two_pi();
+
+	/// Return pi / 4.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType quarter_pi();
+
+	/// Return 1 / pi.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType one_over_pi();
+
+	/// Return 1 / (pi * 2).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType one_over_two_pi();
+
+	/// Return 2 / pi.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType two_over_pi();
+
+	/// Return 4 / pi.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType four_over_pi();
+
+	/// Return 2 / sqrt(pi).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType two_over_root_pi();
+
+	/// Return 1 / sqrt(2).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType one_over_root_two();
+
+	/// Return sqrt(pi / 2).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType root_half_pi();
+
+	/// Return sqrt(2 * pi).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType root_two_pi();
+
+	/// Return sqrt(ln(4)).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType root_ln_four();
+
+	/// Return e constant.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType e();
+
+	/// Return Euler's constant.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType euler();
+
+	/// Return sqrt(2).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType root_two();
+
+	/// Return sqrt(3).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType root_three();
+
+	/// Return sqrt(5).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType root_five();
+
+	/// Return ln(2).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType ln_two();
+
+	/// Return ln(10).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType ln_ten();
+
+	/// Return ln(ln(2)).
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType ln_ln_two();
+
+	/// Return 1 / 3.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType third();
+
+	/// Return 2 / 3.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType two_thirds();
+
+	/// Return the golden ratio constant.
+	/// @see gtc_constants
+	template<typename genType>
+	GLM_FUNC_DECL GLM_CONSTEXPR genType golden_ratio();
+
+	/// @}
+} //namespace glm
+
+#include "constants.inl"

cuda_extensions/third_party/glm/glm/gtc/constants.inl

@@ -0,0 +1,167 @@
+/// @ref gtc_constants
+
+namespace glm
+{
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType zero()
+	{
+		return genType(0);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType one()
+	{
+		return genType(1);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType two_pi()
+	{
+		return genType(6.28318530717958647692528676655900576);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType root_pi()
+	{
+		return genType(1.772453850905516027);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType half_pi()
+	{
+		return genType(1.57079632679489661923132169163975144);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType three_over_two_pi()
+	{
+		return genType(4.71238898038468985769396507491925432);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType quarter_pi()
+	{
+		return genType(0.785398163397448309615660845819875721);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType one_over_pi()
+	{
+		return genType(0.318309886183790671537767526745028724);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType one_over_two_pi()
+	{
+		return genType(0.159154943091895335768883763372514362);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType two_over_pi()
+	{
+		return genType(0.636619772367581343075535053490057448);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType four_over_pi()
+	{
+		return genType(1.273239544735162686151070106980114898);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType two_over_root_pi()
+	{
+		return genType(1.12837916709551257389615890312154517);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType one_over_root_two()
+	{
+		return genType(0.707106781186547524400844362104849039);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType root_half_pi()
+	{
+		return genType(1.253314137315500251);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType root_two_pi()
+	{
+		return genType(2.506628274631000502);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType root_ln_four()
+	{
+		return genType(1.17741002251547469);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType e()
+	{
+		return genType(2.71828182845904523536);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType euler()
+	{
+		return genType(0.577215664901532860606);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType root_two()
+	{
+		return genType(1.41421356237309504880168872420969808);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType root_three()
+	{
+		return genType(1.73205080756887729352744634150587236);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType root_five()
+	{
+		return genType(2.23606797749978969640917366873127623);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType ln_two()
+	{
+		return genType(0.693147180559945309417232121458176568);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType ln_ten()
+	{
+		return genType(2.30258509299404568401799145468436421);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType ln_ln_two()
+	{
+		return genType(-0.3665129205816643);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType third()
+	{
+		return genType(0.3333333333333333333333333333333333333333);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType two_thirds()
+	{
+		return genType(0.666666666666666666666666666666666666667);
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER GLM_CONSTEXPR genType golden_ratio()
+	{
+		return genType(1.61803398874989484820458683436563811);
+	}
+
+} //namespace glm

cuda_extensions/third_party/glm/glm/gtc/epsilon.hpp

@@ -0,0 +1,60 @@
+/// @ref gtc_epsilon
+/// @file glm/gtc/epsilon.hpp
+///
+/// @see core (dependence)
+/// @see gtc_quaternion (dependence)
+///
+/// @defgroup gtc_epsilon GLM_GTC_epsilon
+/// @ingroup gtc
+///
+/// Include <glm/gtc/epsilon.hpp> to use the features of this extension.
+///
+/// Comparison functions for a user defined epsilon values.
+
+#pragma once
+
+// Dependencies
+#include "../detail/setup.hpp"
+#include "../detail/qualifier.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_epsilon extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_epsilon
+	/// @{
+
+	/// Returns the component-wise comparison of |x - y| < epsilon.
+	/// True if this expression is satisfied.
+	///
+	/// @see gtc_epsilon
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, bool, Q> epsilonEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, T const& epsilon);
+
+	/// Returns the component-wise comparison of |x - y| < epsilon.
+	/// True if this expression is satisfied.
+	///
+	/// @see gtc_epsilon
+	template<typename genType>
+	GLM_FUNC_DECL bool epsilonEqual(genType const& x, genType const& y, genType const& epsilon);
+
+	/// Returns the component-wise comparison of |x - y| < epsilon.
+	/// True if this expression is not satisfied.
+	///
+	/// @see gtc_epsilon
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, bool, Q> epsilonNotEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, T const& epsilon);
+
+	/// Returns the component-wise comparison of |x - y| >= epsilon.
+	/// True if this expression is not satisfied.
+	///
+	/// @see gtc_epsilon
+	template<typename genType>
+	GLM_FUNC_DECL bool epsilonNotEqual(genType const& x, genType const& y, genType const& epsilon);
+
+	/// @}
+}//namespace glm
+
+#include "epsilon.inl"

cuda_extensions/third_party/glm/glm/gtc/epsilon.inl

@@ -0,0 +1,80 @@
+/// @ref gtc_epsilon
+
+// Dependency:
+#include "../vector_relational.hpp"
+#include "../common.hpp"
+
+namespace glm
+{
+	template<>
+	GLM_FUNC_QUALIFIER bool epsilonEqual
+	(
+		float const& x,
+		float const& y,
+		float const& epsilon
+	)
+	{
+		return abs(x - y) < epsilon;
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER bool epsilonEqual
+	(
+		double const& x,
+		double const& y,
+		double const& epsilon
+	)
+	{
+		return abs(x - y) < epsilon;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, bool, Q> epsilonEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, T const& epsilon)
+	{
+		return lessThan(abs(x - y), vec<L, T, Q>(epsilon));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, bool, Q> epsilonEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, T, Q> const& epsilon)
+	{
+		return lessThan(abs(x - y), vec<L, T, Q>(epsilon));
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER bool epsilonNotEqual(float const& x, float const& y, float const& epsilon)
+	{
+		return abs(x - y) >= epsilon;
+	}
+
+	template<>
+	GLM_FUNC_QUALIFIER bool epsilonNotEqual(double const& x, double const& y, double const& epsilon)
+	{
+		return abs(x - y) >= epsilon;
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, bool, Q> epsilonNotEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, T const& epsilon)
+	{
+		return greaterThanEqual(abs(x - y), vec<L, T, Q>(epsilon));
+	}
+
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, bool, Q> epsilonNotEqual(vec<L, T, Q> const& x, vec<L, T, Q> const& y, vec<L, T, Q> const& epsilon)
+	{
+		return greaterThanEqual(abs(x - y), vec<L, T, Q>(epsilon));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<4, bool, Q> epsilonEqual(qua<T, Q> const& x, qua<T, Q> const& y, T const& epsilon)
+	{
+		vec<4, T, Q> v(x.x - y.x, x.y - y.y, x.z - y.z, x.w - y.w);
+		return lessThan(abs(v), vec<4, T, Q>(epsilon));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<4, bool, Q> epsilonNotEqual(qua<T, Q> const& x, qua<T, Q> const& y, T const& epsilon)
+	{
+		vec<4, T, Q> v(x.x - y.x, x.y - y.y, x.z - y.z, x.w - y.w);
+		return greaterThanEqual(abs(v), vec<4, T, Q>(epsilon));
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/gtc/integer.hpp

@@ -0,0 +1,43 @@
+/// @ref gtc_integer
+/// @file glm/gtc/integer.hpp
+///
+/// @see core (dependence)
+/// @see gtc_integer (dependence)
+///
+/// @defgroup gtc_integer GLM_GTC_integer
+/// @ingroup gtc
+///
+/// Include <glm/gtc/integer.hpp> to use the features of this extension.
+///
+/// @brief Allow to perform bit operations on integer values
+
+#pragma once
+
+// Dependencies
+#include "../detail/setup.hpp"
+#include "../detail/qualifier.hpp"
+#include "../common.hpp"
+#include "../integer.hpp"
+#include "../exponential.hpp"
+#include "../ext/scalar_common.hpp"
+#include "../ext/vector_common.hpp"
+#include <limits>
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_integer extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_integer
+	/// @{
+
+	/// Returns the log2 of x for integer values. Useful to compute mipmap count from the texture size.
+	/// @see gtc_integer
+	template<typename genIUType>
+	GLM_FUNC_DECL genIUType log2(genIUType x);
+
+	/// @}
+} //namespace glm
+
+#include "integer.inl"

cuda_extensions/third_party/glm/glm/gtc/integer.inl

@@ -0,0 +1,33 @@
+/// @ref gtc_integer
+
+namespace glm{
+namespace detail
+{
+	template<length_t L, typename T, qualifier Q, bool Aligned>
+	struct compute_log2<L, T, Q, false, Aligned>
+	{
+		GLM_FUNC_QUALIFIER static vec<L, T, Q> call(vec<L, T, Q> const& v)
+		{
+			//Equivalent to return findMSB(vec); but save one function call in ASM with VC
+			//return findMSB(vec);
+			return vec<L, T, Q>(detail::compute_findMSB_vec<L, T, Q, sizeof(T) * 8>::call(v));
+		}
+	};
+
+#	if GLM_HAS_BITSCAN_WINDOWS
+		template<qualifier Q, bool Aligned>
+		struct compute_log2<4, int, Q, false, Aligned>
+		{
+			GLM_FUNC_QUALIFIER static vec<4, int, Q> call(vec<4, int, Q> const& v)
+			{
+				vec<4, int, Q> Result;
+				_BitScanReverse(reinterpret_cast<unsigned long*>(&Result.x), v.x);
+				_BitScanReverse(reinterpret_cast<unsigned long*>(&Result.y), v.y);
+				_BitScanReverse(reinterpret_cast<unsigned long*>(&Result.z), v.z);
+				_BitScanReverse(reinterpret_cast<unsigned long*>(&Result.w), v.w);
+				return Result;
+			}
+		};
+#	endif//GLM_HAS_BITSCAN_WINDOWS
+}//namespace detail
+}//namespace glm

cuda_extensions/third_party/glm/glm/gtc/matrix_access.hpp

@@ -0,0 +1,60 @@
+/// @ref gtc_matrix_access
+/// @file glm/gtc/matrix_access.hpp
+///
+/// @see core (dependence)
+///
+/// @defgroup gtc_matrix_access GLM_GTC_matrix_access
+/// @ingroup gtc
+///
+/// Include <glm/gtc/matrix_access.hpp> to use the features of this extension.
+///
+/// Defines functions to access rows or columns of a matrix easily.
+
+#pragma once
+
+// Dependency:
+#include "../detail/setup.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_matrix_access extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_matrix_access
+	/// @{
+
+	/// Get a specific row of a matrix.
+	/// @see gtc_matrix_access
+	template<typename genType>
+	GLM_FUNC_DECL typename genType::row_type row(
+		genType const& m,
+		length_t index);
+
+	/// Set a specific row to a matrix.
+	/// @see gtc_matrix_access
+	template<typename genType>
+	GLM_FUNC_DECL genType row(
+		genType const& m,
+		length_t index,
+		typename genType::row_type const& x);
+
+	/// Get a specific column of a matrix.
+	/// @see gtc_matrix_access
+	template<typename genType>
+	GLM_FUNC_DECL typename genType::col_type column(
+		genType const& m,
+		length_t index);
+
+	/// Set a specific column to a matrix.
+	/// @see gtc_matrix_access
+	template<typename genType>
+	GLM_FUNC_DECL genType column(
+		genType const& m,
+		length_t index,
+		typename genType::col_type const& x);
+
+	/// @}
+}//namespace glm
+
+#include "matrix_access.inl"

cuda_extensions/third_party/glm/glm/gtc/matrix_access.inl

@@ -0,0 +1,62 @@
+/// @ref gtc_matrix_access
+
+namespace glm
+{
+	template<typename genType>
+	GLM_FUNC_QUALIFIER genType row
+	(
+		genType const& m,
+		length_t index,
+		typename genType::row_type const& x
+	)
+	{
+		assert(index >= 0 && index < m[0].length());
+
+		genType Result = m;
+		for(length_t i = 0; i < m.length(); ++i)
+			Result[i][index] = x[i];
+		return Result;
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER typename genType::row_type row
+	(
+		genType const& m,
+		length_t index
+	)
+	{
+		assert(index >= 0 && index < m[0].length());
+
+		typename genType::row_type Result(0);
+		for(length_t i = 0; i < m.length(); ++i)
+			Result[i] = m[i][index];
+		return Result;
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER genType column
+	(
+		genType const& m,
+		length_t index,
+		typename genType::col_type const& x
+	)
+	{
+		assert(index >= 0 && index < m.length());
+
+		genType Result = m;
+		Result[index] = x;
+		return Result;
+	}
+
+	template<typename genType>
+	GLM_FUNC_QUALIFIER typename genType::col_type column
+	(
+		genType const& m,
+		length_t index
+	)
+	{
+		assert(index >= 0 && index < m.length());
+
+		return m[index];
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/gtc/matrix_integer.hpp

@@ -0,0 +1,433 @@
+/// @ref gtc_matrix_integer
+/// @file glm/gtc/matrix_integer.hpp
+///
+/// @see core (dependence)
+///
+/// @defgroup gtc_matrix_integer GLM_GTC_matrix_integer
+/// @ingroup gtc
+///
+/// Include <glm/gtc/matrix_integer.hpp> to use the features of this extension.
+///
+/// Defines a number of matrices with integer types.
+
+#pragma once
+
+// Dependency:
+#include "../mat2x2.hpp"
+#include "../mat2x3.hpp"
+#include "../mat2x4.hpp"
+#include "../mat3x2.hpp"
+#include "../mat3x3.hpp"
+#include "../mat3x4.hpp"
+#include "../mat4x2.hpp"
+#include "../mat4x3.hpp"
+#include "../mat4x4.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_matrix_integer extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_matrix_integer
+	/// @{
+
+	/// High-qualifier signed integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, int, highp>				highp_imat2;
+
+	/// High-qualifier signed integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, int, highp>				highp_imat3;
+
+	/// High-qualifier signed integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, int, highp>				highp_imat4;
+
+	/// High-qualifier signed integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, int, highp>				highp_imat2x2;
+
+	/// High-qualifier signed integer 2x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 3, int, highp>				highp_imat2x3;
+
+	/// High-qualifier signed integer 2x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 4, int, highp>				highp_imat2x4;
+
+	/// High-qualifier signed integer 3x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 2, int, highp>				highp_imat3x2;
+
+	/// High-qualifier signed integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, int, highp>				highp_imat3x3;
+
+	/// High-qualifier signed integer 3x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 4, int, highp>				highp_imat3x4;
+
+	/// High-qualifier signed integer 4x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 2, int, highp>				highp_imat4x2;
+
+	/// High-qualifier signed integer 4x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 3, int, highp>				highp_imat4x3;
+
+	/// High-qualifier signed integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, int, highp>				highp_imat4x4;
+
+
+	/// Medium-qualifier signed integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, int, mediump>			mediump_imat2;
+
+	/// Medium-qualifier signed integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, int, mediump>			mediump_imat3;
+
+	/// Medium-qualifier signed integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, int, mediump>			mediump_imat4;
+
+
+	/// Medium-qualifier signed integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, int, mediump>			mediump_imat2x2;
+
+	/// Medium-qualifier signed integer 2x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 3, int, mediump>			mediump_imat2x3;
+
+	/// Medium-qualifier signed integer 2x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 4, int, mediump>			mediump_imat2x4;
+
+	/// Medium-qualifier signed integer 3x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 2, int, mediump>			mediump_imat3x2;
+
+	/// Medium-qualifier signed integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, int, mediump>			mediump_imat3x3;
+
+	/// Medium-qualifier signed integer 3x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 4, int, mediump>			mediump_imat3x4;
+
+	/// Medium-qualifier signed integer 4x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 2, int, mediump>			mediump_imat4x2;
+
+	/// Medium-qualifier signed integer 4x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 3, int, mediump>			mediump_imat4x3;
+
+	/// Medium-qualifier signed integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, int, mediump>			mediump_imat4x4;
+
+
+	/// Low-qualifier signed integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, int, lowp>				lowp_imat2;
+
+	/// Low-qualifier signed integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, int, lowp>				lowp_imat3;
+
+	/// Low-qualifier signed integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, int, lowp>				lowp_imat4;
+
+
+	/// Low-qualifier signed integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, int, lowp>				lowp_imat2x2;
+
+	/// Low-qualifier signed integer 2x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 3, int, lowp>				lowp_imat2x3;
+
+	/// Low-qualifier signed integer 2x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 4, int, lowp>				lowp_imat2x4;
+
+	/// Low-qualifier signed integer 3x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 2, int, lowp>				lowp_imat3x2;
+
+	/// Low-qualifier signed integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, int, lowp>				lowp_imat3x3;
+
+	/// Low-qualifier signed integer 3x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 4, int, lowp>				lowp_imat3x4;
+
+	/// Low-qualifier signed integer 4x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 2, int, lowp>				lowp_imat4x2;
+
+	/// Low-qualifier signed integer 4x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 3, int, lowp>				lowp_imat4x3;
+
+	/// Low-qualifier signed integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, int, lowp>				lowp_imat4x4;
+
+
+	/// High-qualifier unsigned integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, uint, highp>				highp_umat2;
+
+	/// High-qualifier unsigned integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, uint, highp>				highp_umat3;
+
+	/// High-qualifier unsigned integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, uint, highp>				highp_umat4;
+
+	/// High-qualifier unsigned integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, uint, highp>				highp_umat2x2;
+
+	/// High-qualifier unsigned integer 2x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 3, uint, highp>				highp_umat2x3;
+
+	/// High-qualifier unsigned integer 2x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 4, uint, highp>				highp_umat2x4;
+
+	/// High-qualifier unsigned integer 3x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 2, uint, highp>				highp_umat3x2;
+
+	/// High-qualifier unsigned integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, uint, highp>				highp_umat3x3;
+
+	/// High-qualifier unsigned integer 3x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 4, uint, highp>				highp_umat3x4;
+
+	/// High-qualifier unsigned integer 4x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 2, uint, highp>				highp_umat4x2;
+
+	/// High-qualifier unsigned integer 4x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 3, uint, highp>				highp_umat4x3;
+
+	/// High-qualifier unsigned integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, uint, highp>				highp_umat4x4;
+
+
+	/// Medium-qualifier unsigned integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, uint, mediump>			mediump_umat2;
+
+	/// Medium-qualifier unsigned integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, uint, mediump>			mediump_umat3;
+
+	/// Medium-qualifier unsigned integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, uint, mediump>			mediump_umat4;
+
+
+	/// Medium-qualifier unsigned integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, uint, mediump>			mediump_umat2x2;
+
+	/// Medium-qualifier unsigned integer 2x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 3, uint, mediump>			mediump_umat2x3;
+
+	/// Medium-qualifier unsigned integer 2x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 4, uint, mediump>			mediump_umat2x4;
+
+	/// Medium-qualifier unsigned integer 3x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 2, uint, mediump>			mediump_umat3x2;
+
+	/// Medium-qualifier unsigned integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, uint, mediump>			mediump_umat3x3;
+
+	/// Medium-qualifier unsigned integer 3x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 4, uint, mediump>			mediump_umat3x4;
+
+	/// Medium-qualifier unsigned integer 4x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 2, uint, mediump>			mediump_umat4x2;
+
+	/// Medium-qualifier unsigned integer 4x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 3, uint, mediump>			mediump_umat4x3;
+
+	/// Medium-qualifier unsigned integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, uint, mediump>			mediump_umat4x4;
+
+
+	/// Low-qualifier unsigned integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, uint, lowp>				lowp_umat2;
+
+	/// Low-qualifier unsigned integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, uint, lowp>				lowp_umat3;
+
+	/// Low-qualifier unsigned integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, uint, lowp>				lowp_umat4;
+
+
+	/// Low-qualifier unsigned integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, uint, lowp>				lowp_umat2x2;
+
+	/// Low-qualifier unsigned integer 2x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 3, uint, lowp>				lowp_umat2x3;
+
+	/// Low-qualifier unsigned integer 2x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 4, uint, lowp>				lowp_umat2x4;
+
+	/// Low-qualifier unsigned integer 3x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 2, uint, lowp>				lowp_umat3x2;
+
+	/// Low-qualifier unsigned integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, uint, lowp>				lowp_umat3x3;
+
+	/// Low-qualifier unsigned integer 3x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 4, uint, lowp>				lowp_umat3x4;
+
+	/// Low-qualifier unsigned integer 4x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 2, uint, lowp>				lowp_umat4x2;
+
+	/// Low-qualifier unsigned integer 4x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 3, uint, lowp>				lowp_umat4x3;
+
+	/// Low-qualifier unsigned integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, uint, lowp>				lowp_umat4x4;
+
+
+
+	/// Signed integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, int, defaultp>				imat2;
+
+	/// Signed integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, int, defaultp>				imat3;
+
+	/// Signed integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, int, defaultp>				imat4;
+
+	/// Signed integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, int, defaultp>				imat2x2;
+
+	/// Signed integer 2x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 3, int, defaultp>				imat2x3;
+
+	/// Signed integer 2x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 4, int, defaultp>				imat2x4;
+
+	/// Signed integer 3x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 2, int, defaultp>				imat3x2;
+
+	/// Signed integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, int, defaultp>				imat3x3;
+
+	/// Signed integer 3x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 4, int, defaultp>				imat3x4;
+
+	/// Signed integer 4x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 2, int, defaultp>				imat4x2;
+
+	/// Signed integer 4x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 3, int, defaultp>				imat4x3;
+
+	/// Signed integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, int, defaultp>				imat4x4;
+
+
+
+	/// Unsigned integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, uint, defaultp>				umat2;
+
+	/// Unsigned integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, uint, defaultp>				umat3;
+
+	/// Unsigned integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, uint, defaultp>				umat4;
+
+	/// Unsigned integer 2x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 2, uint, defaultp>				umat2x2;
+
+	/// Unsigned integer 2x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 3, uint, defaultp>				umat2x3;
+
+	/// Unsigned integer 2x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<2, 4, uint, defaultp>				umat2x4;
+
+	/// Unsigned integer 3x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 2, uint, defaultp>				umat3x2;
+
+	/// Unsigned integer 3x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 3, uint, defaultp>				umat3x3;
+
+	/// Unsigned integer 3x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<3, 4, uint, defaultp>				umat3x4;
+
+	/// Unsigned integer 4x2 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 2, uint, defaultp>				umat4x2;
+
+	/// Unsigned integer 4x3 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 3, uint, defaultp>				umat4x3;
+
+	/// Unsigned integer 4x4 matrix.
+	/// @see gtc_matrix_integer
+	typedef mat<4, 4, uint, defaultp>				umat4x4;
+
+	/// @}
+}//namespace glm

cuda_extensions/third_party/glm/glm/gtc/matrix_inverse.hpp

@@ -0,0 +1,50 @@
+/// @ref gtc_matrix_inverse
+/// @file glm/gtc/matrix_inverse.hpp
+///
+/// @see core (dependence)
+///
+/// @defgroup gtc_matrix_inverse GLM_GTC_matrix_inverse
+/// @ingroup gtc
+///
+/// Include <glm/gtc/matrix_inverse.hpp> to use the features of this extension.
+///
+/// Defines additional matrix inverting functions.
+
+#pragma once
+
+// Dependencies
+#include "../detail/setup.hpp"
+#include "../matrix.hpp"
+#include "../mat2x2.hpp"
+#include "../mat3x3.hpp"
+#include "../mat4x4.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_matrix_inverse extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_matrix_inverse
+	/// @{
+
+	/// Fast matrix inverse for affine matrix.
+	///
+	/// @param m Input matrix to invert.
+	/// @tparam genType Squared floating-point matrix: half, float or double. Inverse of matrix based of half-qualifier floating point value is highly inaccurate.
+	/// @see gtc_matrix_inverse
+	template<typename genType>
+	GLM_FUNC_DECL genType affineInverse(genType const& m);
+
+	/// Compute the inverse transpose of a matrix.
+	///
+	/// @param m Input matrix to invert transpose.
+	/// @tparam genType Squared floating-point matrix: half, float or double. Inverse of matrix based of half-qualifier floating point value is highly inaccurate.
+	/// @see gtc_matrix_inverse
+	template<typename genType>
+	GLM_FUNC_DECL genType inverseTranspose(genType const& m);
+
+	/// @}
+}//namespace glm
+
+#include "matrix_inverse.inl"

cuda_extensions/third_party/glm/glm/gtc/matrix_inverse.inl

@@ -0,0 +1,118 @@
+/// @ref gtc_matrix_inverse
+
+namespace glm
+{
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER mat<3, 3, T, Q> affineInverse(mat<3, 3, T, Q> const& m)
+	{
+		mat<2, 2, T, Q> const Inv(inverse(mat<2, 2, T, Q>(m)));
+
+		return mat<3, 3, T, Q>(
+			vec<3, T, Q>(Inv[0], static_cast<T>(0)),
+			vec<3, T, Q>(Inv[1], static_cast<T>(0)),
+			vec<3, T, Q>(-Inv * vec<2, T, Q>(m[2]), static_cast<T>(1)));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER mat<4, 4, T, Q> affineInverse(mat<4, 4, T, Q> const& m)
+	{
+		mat<3, 3, T, Q> const Inv(inverse(mat<3, 3, T, Q>(m)));
+
+		return mat<4, 4, T, Q>(
+			vec<4, T, Q>(Inv[0], static_cast<T>(0)),
+			vec<4, T, Q>(Inv[1], static_cast<T>(0)),
+			vec<4, T, Q>(Inv[2], static_cast<T>(0)),
+			vec<4, T, Q>(-Inv * vec<3, T, Q>(m[3]), static_cast<T>(1)));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER mat<2, 2, T, Q> inverseTranspose(mat<2, 2, T, Q> const& m)
+	{
+		T Determinant = m[0][0] * m[1][1] - m[1][0] * m[0][1];
+
+		mat<2, 2, T, Q> Inverse(
+			+ m[1][1] / Determinant,
+			- m[0][1] / Determinant,
+			- m[1][0] / Determinant,
+			+ m[0][0] / Determinant);
+
+		return Inverse;
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER mat<3, 3, T, Q> inverseTranspose(mat<3, 3, T, Q> const& m)
+	{
+		T Determinant =
+			+ m[0][0] * (m[1][1] * m[2][2] - m[1][2] * m[2][1])
+			- m[0][1] * (m[1][0] * m[2][2] - m[1][2] * m[2][0])
+			+ m[0][2] * (m[1][0] * m[2][1] - m[1][1] * m[2][0]);
+
+		mat<3, 3, T, Q> Inverse;
+		Inverse[0][0] = + (m[1][1] * m[2][2] - m[2][1] * m[1][2]);
+		Inverse[0][1] = - (m[1][0] * m[2][2] - m[2][0] * m[1][2]);
+		Inverse[0][2] = + (m[1][0] * m[2][1] - m[2][0] * m[1][1]);
+		Inverse[1][0] = - (m[0][1] * m[2][2] - m[2][1] * m[0][2]);
+		Inverse[1][1] = + (m[0][0] * m[2][2] - m[2][0] * m[0][2]);
+		Inverse[1][2] = - (m[0][0] * m[2][1] - m[2][0] * m[0][1]);
+		Inverse[2][0] = + (m[0][1] * m[1][2] - m[1][1] * m[0][2]);
+		Inverse[2][1] = - (m[0][0] * m[1][2] - m[1][0] * m[0][2]);
+		Inverse[2][2] = + (m[0][0] * m[1][1] - m[1][0] * m[0][1]);
+		Inverse /= Determinant;
+
+		return Inverse;
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER mat<4, 4, T, Q> inverseTranspose(mat<4, 4, T, Q> const& m)
+	{
+		T SubFactor00 = m[2][2] * m[3][3] - m[3][2] * m[2][3];
+		T SubFactor01 = m[2][1] * m[3][3] - m[3][1] * m[2][3];
+		T SubFactor02 = m[2][1] * m[3][2] - m[3][1] * m[2][2];
+		T SubFactor03 = m[2][0] * m[3][3] - m[3][0] * m[2][3];
+		T SubFactor04 = m[2][0] * m[3][2] - m[3][0] * m[2][2];
+		T SubFactor05 = m[2][0] * m[3][1] - m[3][0] * m[2][1];
+		T SubFactor06 = m[1][2] * m[3][3] - m[3][2] * m[1][3];
+		T SubFactor07 = m[1][1] * m[3][3] - m[3][1] * m[1][3];
+		T SubFactor08 = m[1][1] * m[3][2] - m[3][1] * m[1][2];
+		T SubFactor09 = m[1][0] * m[3][3] - m[3][0] * m[1][3];
+		T SubFactor10 = m[1][0] * m[3][2] - m[3][0] * m[1][2];
+		T SubFactor11 = m[1][0] * m[3][1] - m[3][0] * m[1][1];
+		T SubFactor12 = m[1][2] * m[2][3] - m[2][2] * m[1][3];
+		T SubFactor13 = m[1][1] * m[2][3] - m[2][1] * m[1][3];
+		T SubFactor14 = m[1][1] * m[2][2] - m[2][1] * m[1][2];
+		T SubFactor15 = m[1][0] * m[2][3] - m[2][0] * m[1][3];
+		T SubFactor16 = m[1][0] * m[2][2] - m[2][0] * m[1][2];
+		T SubFactor17 = m[1][0] * m[2][1] - m[2][0] * m[1][1];
+
+		mat<4, 4, T, Q> Inverse;
+		Inverse[0][0] = + (m[1][1] * SubFactor00 - m[1][2] * SubFactor01 + m[1][3] * SubFactor02);
+		Inverse[0][1] = - (m[1][0] * SubFactor00 - m[1][2] * SubFactor03 + m[1][3] * SubFactor04);
+		Inverse[0][2] = + (m[1][0] * SubFactor01 - m[1][1] * SubFactor03 + m[1][3] * SubFactor05);
+		Inverse[0][3] = - (m[1][0] * SubFactor02 - m[1][1] * SubFactor04 + m[1][2] * SubFactor05);
+
+		Inverse[1][0] = - (m[0][1] * SubFactor00 - m[0][2] * SubFactor01 + m[0][3] * SubFactor02);
+		Inverse[1][1] = + (m[0][0] * SubFactor00 - m[0][2] * SubFactor03 + m[0][3] * SubFactor04);
+		Inverse[1][2] = - (m[0][0] * SubFactor01 - m[0][1] * SubFactor03 + m[0][3] * SubFactor05);
+		Inverse[1][3] = + (m[0][0] * SubFactor02 - m[0][1] * SubFactor04 + m[0][2] * SubFactor05);
+
+		Inverse[2][0] = + (m[0][1] * SubFactor06 - m[0][2] * SubFactor07 + m[0][3] * SubFactor08);
+		Inverse[2][1] = - (m[0][0] * SubFactor06 - m[0][2] * SubFactor09 + m[0][3] * SubFactor10);
+		Inverse[2][2] = + (m[0][0] * SubFactor07 - m[0][1] * SubFactor09 + m[0][3] * SubFactor11);
+		Inverse[2][3] = - (m[0][0] * SubFactor08 - m[0][1] * SubFactor10 + m[0][2] * SubFactor11);
+
+		Inverse[3][0] = - (m[0][1] * SubFactor12 - m[0][2] * SubFactor13 + m[0][3] * SubFactor14);
+		Inverse[3][1] = + (m[0][0] * SubFactor12 - m[0][2] * SubFactor15 + m[0][3] * SubFactor16);
+		Inverse[3][2] = - (m[0][0] * SubFactor13 - m[0][1] * SubFactor15 + m[0][3] * SubFactor17);
+		Inverse[3][3] = + (m[0][0] * SubFactor14 - m[0][1] * SubFactor16 + m[0][2] * SubFactor17);
+
+		T Determinant =
+			+ m[0][0] * Inverse[0][0]
+			+ m[0][1] * Inverse[0][1]
+			+ m[0][2] * Inverse[0][2]
+			+ m[0][3] * Inverse[0][3];
+
+		Inverse /= Determinant;
+
+		return Inverse;
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/gtc/matrix_transform.hpp

@@ -0,0 +1,36 @@
+/// @ref gtc_matrix_transform
+/// @file glm/gtc/matrix_transform.hpp
+///
+/// @see core (dependence)
+/// @see gtx_transform
+/// @see gtx_transform2
+///
+/// @defgroup gtc_matrix_transform GLM_GTC_matrix_transform
+/// @ingroup gtc
+///
+/// Include <glm/gtc/matrix_transform.hpp> to use the features of this extension.
+///
+/// Defines functions that generate common transformation matrices.
+///
+/// The matrices generated by this extension use standard OpenGL fixed-function
+/// conventions. For example, the lookAt function generates a transform from world
+/// space into the specific eye space that the projective matrix functions
+/// (perspective, ortho, etc) are designed to expect. The OpenGL compatibility
+/// specifications defines the particular layout of this eye space.
+
+#pragma once
+
+// Dependencies
+#include "../mat4x4.hpp"
+#include "../vec2.hpp"
+#include "../vec3.hpp"
+#include "../vec4.hpp"
+#include "../ext/matrix_projection.hpp"
+#include "../ext/matrix_clip_space.hpp"
+#include "../ext/matrix_transform.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_matrix_transform extension included")
+#endif
+
+#include "matrix_transform.inl"

cuda_extensions/third_party/glm/glm/gtc/matrix_transform.inl

@@ -0,0 +1,3 @@
+#include "../geometric.hpp"
+#include "../trigonometric.hpp"
+#include "../matrix.hpp"

cuda_extensions/third_party/glm/glm/gtc/noise.hpp

@@ -0,0 +1,61 @@
+/// @ref gtc_noise
+/// @file glm/gtc/noise.hpp
+///
+/// @see core (dependence)
+///
+/// @defgroup gtc_noise GLM_GTC_noise
+/// @ingroup gtc
+///
+/// Include <glm/gtc/noise.hpp> to use the features of this extension.
+///
+/// Defines 2D, 3D and 4D procedural noise functions
+/// Based on the work of Stefan Gustavson and Ashima Arts on "webgl-noise":
+/// https://github.com/ashima/webgl-noise
+/// Following Stefan Gustavson's paper "Simplex noise demystified":
+/// http://www.itn.liu.se/~stegu/simplexnoise/simplexnoise.pdf
+
+#pragma once
+
+// Dependencies
+#include "../detail/setup.hpp"
+#include "../detail/qualifier.hpp"
+#include "../detail/_noise.hpp"
+#include "../geometric.hpp"
+#include "../common.hpp"
+#include "../vector_relational.hpp"
+#include "../vec2.hpp"
+#include "../vec3.hpp"
+#include "../vec4.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_noise extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_noise
+	/// @{
+
+	/// Classic perlin noise.
+	/// @see gtc_noise
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL T perlin(
+		vec<L, T, Q> const& p);
+
+	/// Periodic perlin noise.
+	/// @see gtc_noise
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL T perlin(
+		vec<L, T, Q> const& p,
+		vec<L, T, Q> const& rep);
+
+	/// Simplex noise.
+	/// @see gtc_noise
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL T simplex(
+		vec<L, T, Q> const& p);
+
+	/// @}
+}//namespace glm
+
+#include "noise.inl"

cuda_extensions/third_party/glm/glm/gtc/noise.inl

@@ -0,0 +1,807 @@
+/// @ref gtc_noise
+///
+// Based on the work of Stefan Gustavson and Ashima Arts on "webgl-noise":
+// https://github.com/ashima/webgl-noise
+// Following Stefan Gustavson's paper "Simplex noise demystified":
+// http://www.itn.liu.se/~stegu/simplexnoise/simplexnoise.pdf
+
+namespace glm{
+namespace gtc
+{
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<4, T, Q> grad4(T const& j, vec<4, T, Q> const& ip)
+	{
+		vec<3, T, Q> pXYZ = floor(fract(vec<3, T, Q>(j) * vec<3, T, Q>(ip)) * T(7)) * ip[2] - T(1);
+		T pW = static_cast<T>(1.5) - dot(abs(pXYZ), vec<3, T, Q>(1));
+		vec<4, T, Q> s = vec<4, T, Q>(lessThan(vec<4, T, Q>(pXYZ, pW), vec<4, T, Q>(0.0)));
+		pXYZ = pXYZ + (vec<3, T, Q>(s) * T(2) - T(1)) * s.w;
+		return vec<4, T, Q>(pXYZ, pW);
+	}
+}//namespace gtc
+
+	// Classic Perlin noise
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T perlin(vec<2, T, Q> const& Position)
+	{
+		vec<4, T, Q> Pi = glm::floor(vec<4, T, Q>(Position.x, Position.y, Position.x, Position.y)) + vec<4, T, Q>(0.0, 0.0, 1.0, 1.0);
+		vec<4, T, Q> Pf = glm::fract(vec<4, T, Q>(Position.x, Position.y, Position.x, Position.y)) - vec<4, T, Q>(0.0, 0.0, 1.0, 1.0);
+		Pi = mod(Pi, vec<4, T, Q>(289)); // To avoid truncation effects in permutation
+		vec<4, T, Q> ix(Pi.x, Pi.z, Pi.x, Pi.z);
+		vec<4, T, Q> iy(Pi.y, Pi.y, Pi.w, Pi.w);
+		vec<4, T, Q> fx(Pf.x, Pf.z, Pf.x, Pf.z);
+		vec<4, T, Q> fy(Pf.y, Pf.y, Pf.w, Pf.w);
+
+		vec<4, T, Q> i = detail::permute(detail::permute(ix) + iy);
+
+		vec<4, T, Q> gx = static_cast<T>(2) * glm::fract(i / T(41)) - T(1);
+		vec<4, T, Q> gy = glm::abs(gx) - T(0.5);
+		vec<4, T, Q> tx = glm::floor(gx + T(0.5));
+		gx = gx - tx;
+
+		vec<2, T, Q> g00(gx.x, gy.x);
+		vec<2, T, Q> g10(gx.y, gy.y);
+		vec<2, T, Q> g01(gx.z, gy.z);
+		vec<2, T, Q> g11(gx.w, gy.w);
+
+		vec<4, T, Q> norm = detail::taylorInvSqrt(vec<4, T, Q>(dot(g00, g00), dot(g01, g01), dot(g10, g10), dot(g11, g11)));
+		g00 *= norm.x;
+		g01 *= norm.y;
+		g10 *= norm.z;
+		g11 *= norm.w;
+
+		T n00 = dot(g00, vec<2, T, Q>(fx.x, fy.x));
+		T n10 = dot(g10, vec<2, T, Q>(fx.y, fy.y));
+		T n01 = dot(g01, vec<2, T, Q>(fx.z, fy.z));
+		T n11 = dot(g11, vec<2, T, Q>(fx.w, fy.w));
+
+		vec<2, T, Q> fade_xy = detail::fade(vec<2, T, Q>(Pf.x, Pf.y));
+		vec<2, T, Q> n_x = mix(vec<2, T, Q>(n00, n01), vec<2, T, Q>(n10, n11), fade_xy.x);
+		T n_xy = mix(n_x.x, n_x.y, fade_xy.y);
+		return T(2.3) * n_xy;
+	}
+
+	// Classic Perlin noise
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T perlin(vec<3, T, Q> const& Position)
+	{
+		vec<3, T, Q> Pi0 = floor(Position); // Integer part for indexing
+		vec<3, T, Q> Pi1 = Pi0 + T(1); // Integer part + 1
+		Pi0 = detail::mod289(Pi0);
+		Pi1 = detail::mod289(Pi1);
+		vec<3, T, Q> Pf0 = fract(Position); // Fractional part for interpolation
+		vec<3, T, Q> Pf1 = Pf0 - T(1); // Fractional part - 1.0
+		vec<4, T, Q> ix(Pi0.x, Pi1.x, Pi0.x, Pi1.x);
+		vec<4, T, Q> iy = vec<4, T, Q>(vec<2, T, Q>(Pi0.y), vec<2, T, Q>(Pi1.y));
+		vec<4, T, Q> iz0(Pi0.z);
+		vec<4, T, Q> iz1(Pi1.z);
+
+		vec<4, T, Q> ixy = detail::permute(detail::permute(ix) + iy);
+		vec<4, T, Q> ixy0 = detail::permute(ixy + iz0);
+		vec<4, T, Q> ixy1 = detail::permute(ixy + iz1);
+
+		vec<4, T, Q> gx0 = ixy0 * T(1.0 / 7.0);
+		vec<4, T, Q> gy0 = fract(floor(gx0) * T(1.0 / 7.0)) - T(0.5);
+		gx0 = fract(gx0);
+		vec<4, T, Q> gz0 = vec<4, T, Q>(0.5) - abs(gx0) - abs(gy0);
+		vec<4, T, Q> sz0 = step(gz0, vec<4, T, Q>(0.0));
+		gx0 -= sz0 * (step(T(0), gx0) - T(0.5));
+		gy0 -= sz0 * (step(T(0), gy0) - T(0.5));
+
+		vec<4, T, Q> gx1 = ixy1 * T(1.0 / 7.0);
+		vec<4, T, Q> gy1 = fract(floor(gx1) * T(1.0 / 7.0)) - T(0.5);
+		gx1 = fract(gx1);
+		vec<4, T, Q> gz1 = vec<4, T, Q>(0.5) - abs(gx1) - abs(gy1);
+		vec<4, T, Q> sz1 = step(gz1, vec<4, T, Q>(0.0));
+		gx1 -= sz1 * (step(T(0), gx1) - T(0.5));
+		gy1 -= sz1 * (step(T(0), gy1) - T(0.5));
+
+		vec<3, T, Q> g000(gx0.x, gy0.x, gz0.x);
+		vec<3, T, Q> g100(gx0.y, gy0.y, gz0.y);
+		vec<3, T, Q> g010(gx0.z, gy0.z, gz0.z);
+		vec<3, T, Q> g110(gx0.w, gy0.w, gz0.w);
+		vec<3, T, Q> g001(gx1.x, gy1.x, gz1.x);
+		vec<3, T, Q> g101(gx1.y, gy1.y, gz1.y);
+		vec<3, T, Q> g011(gx1.z, gy1.z, gz1.z);
+		vec<3, T, Q> g111(gx1.w, gy1.w, gz1.w);
+
+		vec<4, T, Q> norm0 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g000, g000), dot(g010, g010), dot(g100, g100), dot(g110, g110)));
+		g000 *= norm0.x;
+		g010 *= norm0.y;
+		g100 *= norm0.z;
+		g110 *= norm0.w;
+		vec<4, T, Q> norm1 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g001, g001), dot(g011, g011), dot(g101, g101), dot(g111, g111)));
+		g001 *= norm1.x;
+		g011 *= norm1.y;
+		g101 *= norm1.z;
+		g111 *= norm1.w;
+
+		T n000 = dot(g000, Pf0);
+		T n100 = dot(g100, vec<3, T, Q>(Pf1.x, Pf0.y, Pf0.z));
+		T n010 = dot(g010, vec<3, T, Q>(Pf0.x, Pf1.y, Pf0.z));
+		T n110 = dot(g110, vec<3, T, Q>(Pf1.x, Pf1.y, Pf0.z));
+		T n001 = dot(g001, vec<3, T, Q>(Pf0.x, Pf0.y, Pf1.z));
+		T n101 = dot(g101, vec<3, T, Q>(Pf1.x, Pf0.y, Pf1.z));
+		T n011 = dot(g011, vec<3, T, Q>(Pf0.x, Pf1.y, Pf1.z));
+		T n111 = dot(g111, Pf1);
+
+		vec<3, T, Q> fade_xyz = detail::fade(Pf0);
+		vec<4, T, Q> n_z = mix(vec<4, T, Q>(n000, n100, n010, n110), vec<4, T, Q>(n001, n101, n011, n111), fade_xyz.z);
+		vec<2, T, Q> n_yz = mix(vec<2, T, Q>(n_z.x, n_z.y), vec<2, T, Q>(n_z.z, n_z.w), fade_xyz.y);
+		T n_xyz = mix(n_yz.x, n_yz.y, fade_xyz.x);
+		return T(2.2) * n_xyz;
+	}
+	/*
+	// Classic Perlin noise
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T perlin(vec<3, T, Q> const& P)
+	{
+		vec<3, T, Q> Pi0 = floor(P); // Integer part for indexing
+		vec<3, T, Q> Pi1 = Pi0 + T(1); // Integer part + 1
+		Pi0 = mod(Pi0, T(289));
+		Pi1 = mod(Pi1, T(289));
+		vec<3, T, Q> Pf0 = fract(P); // Fractional part for interpolation
+		vec<3, T, Q> Pf1 = Pf0 - T(1); // Fractional part - 1.0
+		vec<4, T, Q> ix(Pi0.x, Pi1.x, Pi0.x, Pi1.x);
+		vec<4, T, Q> iy(Pi0.y, Pi0.y, Pi1.y, Pi1.y);
+		vec<4, T, Q> iz0(Pi0.z);
+		vec<4, T, Q> iz1(Pi1.z);
+
+		vec<4, T, Q> ixy = permute(permute(ix) + iy);
+		vec<4, T, Q> ixy0 = permute(ixy + iz0);
+		vec<4, T, Q> ixy1 = permute(ixy + iz1);
+
+		vec<4, T, Q> gx0 = ixy0 / T(7);
+		vec<4, T, Q> gy0 = fract(floor(gx0) / T(7)) - T(0.5);
+		gx0 = fract(gx0);
+		vec<4, T, Q> gz0 = vec<4, T, Q>(0.5) - abs(gx0) - abs(gy0);
+		vec<4, T, Q> sz0 = step(gz0, vec<4, T, Q>(0.0));
+		gx0 -= sz0 * (step(0.0, gx0) - T(0.5));
+		gy0 -= sz0 * (step(0.0, gy0) - T(0.5));
+
+		vec<4, T, Q> gx1 = ixy1 / T(7);
+		vec<4, T, Q> gy1 = fract(floor(gx1) / T(7)) - T(0.5);
+		gx1 = fract(gx1);
+		vec<4, T, Q> gz1 = vec<4, T, Q>(0.5) - abs(gx1) - abs(gy1);
+		vec<4, T, Q> sz1 = step(gz1, vec<4, T, Q>(0.0));
+		gx1 -= sz1 * (step(T(0), gx1) - T(0.5));
+		gy1 -= sz1 * (step(T(0), gy1) - T(0.5));
+
+		vec<3, T, Q> g000(gx0.x, gy0.x, gz0.x);
+		vec<3, T, Q> g100(gx0.y, gy0.y, gz0.y);
+		vec<3, T, Q> g010(gx0.z, gy0.z, gz0.z);
+		vec<3, T, Q> g110(gx0.w, gy0.w, gz0.w);
+		vec<3, T, Q> g001(gx1.x, gy1.x, gz1.x);
+		vec<3, T, Q> g101(gx1.y, gy1.y, gz1.y);
+		vec<3, T, Q> g011(gx1.z, gy1.z, gz1.z);
+		vec<3, T, Q> g111(gx1.w, gy1.w, gz1.w);
+
+		vec<4, T, Q> norm0 = taylorInvSqrt(vec<4, T, Q>(dot(g000, g000), dot(g010, g010), dot(g100, g100), dot(g110, g110)));
+		g000 *= norm0.x;
+		g010 *= norm0.y;
+		g100 *= norm0.z;
+		g110 *= norm0.w;
+		vec<4, T, Q> norm1 = taylorInvSqrt(vec<4, T, Q>(dot(g001, g001), dot(g011, g011), dot(g101, g101), dot(g111, g111)));
+		g001 *= norm1.x;
+		g011 *= norm1.y;
+		g101 *= norm1.z;
+		g111 *= norm1.w;
+
+		T n000 = dot(g000, Pf0);
+		T n100 = dot(g100, vec<3, T, Q>(Pf1.x, Pf0.y, Pf0.z));
+		T n010 = dot(g010, vec<3, T, Q>(Pf0.x, Pf1.y, Pf0.z));
+		T n110 = dot(g110, vec<3, T, Q>(Pf1.x, Pf1.y, Pf0.z));
+		T n001 = dot(g001, vec<3, T, Q>(Pf0.x, Pf0.y, Pf1.z));
+		T n101 = dot(g101, vec<3, T, Q>(Pf1.x, Pf0.y, Pf1.z));
+		T n011 = dot(g011, vec<3, T, Q>(Pf0.x, Pf1.y, Pf1.z));
+		T n111 = dot(g111, Pf1);
+
+		vec<3, T, Q> fade_xyz = fade(Pf0);
+		vec<4, T, Q> n_z = mix(vec<4, T, Q>(n000, n100, n010, n110), vec<4, T, Q>(n001, n101, n011, n111), fade_xyz.z);
+		vec<2, T, Q> n_yz = mix(
+			vec<2, T, Q>(n_z.x, n_z.y),
+			vec<2, T, Q>(n_z.z, n_z.w), fade_xyz.y);
+		T n_xyz = mix(n_yz.x, n_yz.y, fade_xyz.x);
+		return T(2.2) * n_xyz;
+	}
+	*/
+	// Classic Perlin noise
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T perlin(vec<4, T, Q> const& Position)
+	{
+		vec<4, T, Q> Pi0 = floor(Position);	// Integer part for indexing
+		vec<4, T, Q> Pi1 = Pi0 + T(1);		// Integer part + 1
+		Pi0 = mod(Pi0, vec<4, T, Q>(289));
+		Pi1 = mod(Pi1, vec<4, T, Q>(289));
+		vec<4, T, Q> Pf0 = fract(Position);	// Fractional part for interpolation
+		vec<4, T, Q> Pf1 = Pf0 - T(1);		// Fractional part - 1.0
+		vec<4, T, Q> ix(Pi0.x, Pi1.x, Pi0.x, Pi1.x);
+		vec<4, T, Q> iy(Pi0.y, Pi0.y, Pi1.y, Pi1.y);
+		vec<4, T, Q> iz0(Pi0.z);
+		vec<4, T, Q> iz1(Pi1.z);
+		vec<4, T, Q> iw0(Pi0.w);
+		vec<4, T, Q> iw1(Pi1.w);
+
+		vec<4, T, Q> ixy = detail::permute(detail::permute(ix) + iy);
+		vec<4, T, Q> ixy0 = detail::permute(ixy + iz0);
+		vec<4, T, Q> ixy1 = detail::permute(ixy + iz1);
+		vec<4, T, Q> ixy00 = detail::permute(ixy0 + iw0);
+		vec<4, T, Q> ixy01 = detail::permute(ixy0 + iw1);
+		vec<4, T, Q> ixy10 = detail::permute(ixy1 + iw0);
+		vec<4, T, Q> ixy11 = detail::permute(ixy1 + iw1);
+
+		vec<4, T, Q> gx00 = ixy00 / T(7);
+		vec<4, T, Q> gy00 = floor(gx00) / T(7);
+		vec<4, T, Q> gz00 = floor(gy00) / T(6);
+		gx00 = fract(gx00) - T(0.5);
+		gy00 = fract(gy00) - T(0.5);
+		gz00 = fract(gz00) - T(0.5);
+		vec<4, T, Q> gw00 = vec<4, T, Q>(0.75) - abs(gx00) - abs(gy00) - abs(gz00);
+		vec<4, T, Q> sw00 = step(gw00, vec<4, T, Q>(0.0));
+		gx00 -= sw00 * (step(T(0), gx00) - T(0.5));
+		gy00 -= sw00 * (step(T(0), gy00) - T(0.5));
+
+		vec<4, T, Q> gx01 = ixy01 / T(7);
+		vec<4, T, Q> gy01 = floor(gx01) / T(7);
+		vec<4, T, Q> gz01 = floor(gy01) / T(6);
+		gx01 = fract(gx01) - T(0.5);
+		gy01 = fract(gy01) - T(0.5);
+		gz01 = fract(gz01) - T(0.5);
+		vec<4, T, Q> gw01 = vec<4, T, Q>(0.75) - abs(gx01) - abs(gy01) - abs(gz01);
+		vec<4, T, Q> sw01 = step(gw01, vec<4, T, Q>(0.0));
+		gx01 -= sw01 * (step(T(0), gx01) - T(0.5));
+		gy01 -= sw01 * (step(T(0), gy01) - T(0.5));
+
+		vec<4, T, Q> gx10 = ixy10 / T(7);
+		vec<4, T, Q> gy10 = floor(gx10) / T(7);
+		vec<4, T, Q> gz10 = floor(gy10) / T(6);
+		gx10 = fract(gx10) - T(0.5);
+		gy10 = fract(gy10) - T(0.5);
+		gz10 = fract(gz10) - T(0.5);
+		vec<4, T, Q> gw10 = vec<4, T, Q>(0.75) - abs(gx10) - abs(gy10) - abs(gz10);
+		vec<4, T, Q> sw10 = step(gw10, vec<4, T, Q>(0));
+		gx10 -= sw10 * (step(T(0), gx10) - T(0.5));
+		gy10 -= sw10 * (step(T(0), gy10) - T(0.5));
+
+		vec<4, T, Q> gx11 = ixy11 / T(7);
+		vec<4, T, Q> gy11 = floor(gx11) / T(7);
+		vec<4, T, Q> gz11 = floor(gy11) / T(6);
+		gx11 = fract(gx11) - T(0.5);
+		gy11 = fract(gy11) - T(0.5);
+		gz11 = fract(gz11) - T(0.5);
+		vec<4, T, Q> gw11 = vec<4, T, Q>(0.75) - abs(gx11) - abs(gy11) - abs(gz11);
+		vec<4, T, Q> sw11 = step(gw11, vec<4, T, Q>(0.0));
+		gx11 -= sw11 * (step(T(0), gx11) - T(0.5));
+		gy11 -= sw11 * (step(T(0), gy11) - T(0.5));
+
+		vec<4, T, Q> g0000(gx00.x, gy00.x, gz00.x, gw00.x);
+		vec<4, T, Q> g1000(gx00.y, gy00.y, gz00.y, gw00.y);
+		vec<4, T, Q> g0100(gx00.z, gy00.z, gz00.z, gw00.z);
+		vec<4, T, Q> g1100(gx00.w, gy00.w, gz00.w, gw00.w);
+		vec<4, T, Q> g0010(gx10.x, gy10.x, gz10.x, gw10.x);
+		vec<4, T, Q> g1010(gx10.y, gy10.y, gz10.y, gw10.y);
+		vec<4, T, Q> g0110(gx10.z, gy10.z, gz10.z, gw10.z);
+		vec<4, T, Q> g1110(gx10.w, gy10.w, gz10.w, gw10.w);
+		vec<4, T, Q> g0001(gx01.x, gy01.x, gz01.x, gw01.x);
+		vec<4, T, Q> g1001(gx01.y, gy01.y, gz01.y, gw01.y);
+		vec<4, T, Q> g0101(gx01.z, gy01.z, gz01.z, gw01.z);
+		vec<4, T, Q> g1101(gx01.w, gy01.w, gz01.w, gw01.w);
+		vec<4, T, Q> g0011(gx11.x, gy11.x, gz11.x, gw11.x);
+		vec<4, T, Q> g1011(gx11.y, gy11.y, gz11.y, gw11.y);
+		vec<4, T, Q> g0111(gx11.z, gy11.z, gz11.z, gw11.z);
+		vec<4, T, Q> g1111(gx11.w, gy11.w, gz11.w, gw11.w);
+
+		vec<4, T, Q> norm00 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g0000, g0000), dot(g0100, g0100), dot(g1000, g1000), dot(g1100, g1100)));
+		g0000 *= norm00.x;
+		g0100 *= norm00.y;
+		g1000 *= norm00.z;
+		g1100 *= norm00.w;
+
+		vec<4, T, Q> norm01 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g0001, g0001), dot(g0101, g0101), dot(g1001, g1001), dot(g1101, g1101)));
+		g0001 *= norm01.x;
+		g0101 *= norm01.y;
+		g1001 *= norm01.z;
+		g1101 *= norm01.w;
+
+		vec<4, T, Q> norm10 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g0010, g0010), dot(g0110, g0110), dot(g1010, g1010), dot(g1110, g1110)));
+		g0010 *= norm10.x;
+		g0110 *= norm10.y;
+		g1010 *= norm10.z;
+		g1110 *= norm10.w;
+
+		vec<4, T, Q> norm11 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g0011, g0011), dot(g0111, g0111), dot(g1011, g1011), dot(g1111, g1111)));
+		g0011 *= norm11.x;
+		g0111 *= norm11.y;
+		g1011 *= norm11.z;
+		g1111 *= norm11.w;
+
+		T n0000 = dot(g0000, Pf0);
+		T n1000 = dot(g1000, vec<4, T, Q>(Pf1.x, Pf0.y, Pf0.z, Pf0.w));
+		T n0100 = dot(g0100, vec<4, T, Q>(Pf0.x, Pf1.y, Pf0.z, Pf0.w));
+		T n1100 = dot(g1100, vec<4, T, Q>(Pf1.x, Pf1.y, Pf0.z, Pf0.w));
+		T n0010 = dot(g0010, vec<4, T, Q>(Pf0.x, Pf0.y, Pf1.z, Pf0.w));
+		T n1010 = dot(g1010, vec<4, T, Q>(Pf1.x, Pf0.y, Pf1.z, Pf0.w));
+		T n0110 = dot(g0110, vec<4, T, Q>(Pf0.x, Pf1.y, Pf1.z, Pf0.w));
+		T n1110 = dot(g1110, vec<4, T, Q>(Pf1.x, Pf1.y, Pf1.z, Pf0.w));
+		T n0001 = dot(g0001, vec<4, T, Q>(Pf0.x, Pf0.y, Pf0.z, Pf1.w));
+		T n1001 = dot(g1001, vec<4, T, Q>(Pf1.x, Pf0.y, Pf0.z, Pf1.w));
+		T n0101 = dot(g0101, vec<4, T, Q>(Pf0.x, Pf1.y, Pf0.z, Pf1.w));
+		T n1101 = dot(g1101, vec<4, T, Q>(Pf1.x, Pf1.y, Pf0.z, Pf1.w));
+		T n0011 = dot(g0011, vec<4, T, Q>(Pf0.x, Pf0.y, Pf1.z, Pf1.w));
+		T n1011 = dot(g1011, vec<4, T, Q>(Pf1.x, Pf0.y, Pf1.z, Pf1.w));
+		T n0111 = dot(g0111, vec<4, T, Q>(Pf0.x, Pf1.y, Pf1.z, Pf1.w));
+		T n1111 = dot(g1111, Pf1);
+
+		vec<4, T, Q> fade_xyzw = detail::fade(Pf0);
+		vec<4, T, Q> n_0w = mix(vec<4, T, Q>(n0000, n1000, n0100, n1100), vec<4, T, Q>(n0001, n1001, n0101, n1101), fade_xyzw.w);
+		vec<4, T, Q> n_1w = mix(vec<4, T, Q>(n0010, n1010, n0110, n1110), vec<4, T, Q>(n0011, n1011, n0111, n1111), fade_xyzw.w);
+		vec<4, T, Q> n_zw = mix(n_0w, n_1w, fade_xyzw.z);
+		vec<2, T, Q> n_yzw = mix(vec<2, T, Q>(n_zw.x, n_zw.y), vec<2, T, Q>(n_zw.z, n_zw.w), fade_xyzw.y);
+		T n_xyzw = mix(n_yzw.x, n_yzw.y, fade_xyzw.x);
+		return T(2.2) * n_xyzw;
+	}
+
+	// Classic Perlin noise, periodic variant
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T perlin(vec<2, T, Q> const& Position, vec<2, T, Q> const& rep)
+	{
+		vec<4, T, Q> Pi = floor(vec<4, T, Q>(Position.x, Position.y, Position.x, Position.y)) + vec<4, T, Q>(0.0, 0.0, 1.0, 1.0);
+		vec<4, T, Q> Pf = fract(vec<4, T, Q>(Position.x, Position.y, Position.x, Position.y)) - vec<4, T, Q>(0.0, 0.0, 1.0, 1.0);
+		Pi = mod(Pi, vec<4, T, Q>(rep.x, rep.y, rep.x, rep.y)); // To create noise with explicit period
+		Pi = mod(Pi, vec<4, T, Q>(289)); // To avoid truncation effects in permutation
+		vec<4, T, Q> ix(Pi.x, Pi.z, Pi.x, Pi.z);
+		vec<4, T, Q> iy(Pi.y, Pi.y, Pi.w, Pi.w);
+		vec<4, T, Q> fx(Pf.x, Pf.z, Pf.x, Pf.z);
+		vec<4, T, Q> fy(Pf.y, Pf.y, Pf.w, Pf.w);
+
+		vec<4, T, Q> i = detail::permute(detail::permute(ix) + iy);
+
+		vec<4, T, Q> gx = static_cast<T>(2) * fract(i / T(41)) - T(1);
+		vec<4, T, Q> gy = abs(gx) - T(0.5);
+		vec<4, T, Q> tx = floor(gx + T(0.5));
+		gx = gx - tx;
+
+		vec<2, T, Q> g00(gx.x, gy.x);
+		vec<2, T, Q> g10(gx.y, gy.y);
+		vec<2, T, Q> g01(gx.z, gy.z);
+		vec<2, T, Q> g11(gx.w, gy.w);
+
+		vec<4, T, Q> norm = detail::taylorInvSqrt(vec<4, T, Q>(dot(g00, g00), dot(g01, g01), dot(g10, g10), dot(g11, g11)));
+		g00 *= norm.x;
+		g01 *= norm.y;
+		g10 *= norm.z;
+		g11 *= norm.w;
+
+		T n00 = dot(g00, vec<2, T, Q>(fx.x, fy.x));
+		T n10 = dot(g10, vec<2, T, Q>(fx.y, fy.y));
+		T n01 = dot(g01, vec<2, T, Q>(fx.z, fy.z));
+		T n11 = dot(g11, vec<2, T, Q>(fx.w, fy.w));
+
+		vec<2, T, Q> fade_xy = detail::fade(vec<2, T, Q>(Pf.x, Pf.y));
+		vec<2, T, Q> n_x = mix(vec<2, T, Q>(n00, n01), vec<2, T, Q>(n10, n11), fade_xy.x);
+		T n_xy = mix(n_x.x, n_x.y, fade_xy.y);
+		return T(2.3) * n_xy;
+	}
+
+	// Classic Perlin noise, periodic variant
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T perlin(vec<3, T, Q> const& Position, vec<3, T, Q> const& rep)
+	{
+		vec<3, T, Q> Pi0 = mod(floor(Position), rep); // Integer part, modulo period
+		vec<3, T, Q> Pi1 = mod(Pi0 + vec<3, T, Q>(T(1)), rep); // Integer part + 1, mod period
+		Pi0 = mod(Pi0, vec<3, T, Q>(289));
+		Pi1 = mod(Pi1, vec<3, T, Q>(289));
+		vec<3, T, Q> Pf0 = fract(Position); // Fractional part for interpolation
+		vec<3, T, Q> Pf1 = Pf0 - vec<3, T, Q>(T(1)); // Fractional part - 1.0
+		vec<4, T, Q> ix = vec<4, T, Q>(Pi0.x, Pi1.x, Pi0.x, Pi1.x);
+		vec<4, T, Q> iy = vec<4, T, Q>(Pi0.y, Pi0.y, Pi1.y, Pi1.y);
+		vec<4, T, Q> iz0(Pi0.z);
+		vec<4, T, Q> iz1(Pi1.z);
+
+		vec<4, T, Q> ixy = detail::permute(detail::permute(ix) + iy);
+		vec<4, T, Q> ixy0 = detail::permute(ixy + iz0);
+		vec<4, T, Q> ixy1 = detail::permute(ixy + iz1);
+
+		vec<4, T, Q> gx0 = ixy0 / T(7);
+		vec<4, T, Q> gy0 = fract(floor(gx0) / T(7)) - T(0.5);
+		gx0 = fract(gx0);
+		vec<4, T, Q> gz0 = vec<4, T, Q>(0.5) - abs(gx0) - abs(gy0);
+		vec<4, T, Q> sz0 = step(gz0, vec<4, T, Q>(0));
+		gx0 -= sz0 * (step(T(0), gx0) - T(0.5));
+		gy0 -= sz0 * (step(T(0), gy0) - T(0.5));
+
+		vec<4, T, Q> gx1 = ixy1 / T(7);
+		vec<4, T, Q> gy1 = fract(floor(gx1) / T(7)) - T(0.5);
+		gx1 = fract(gx1);
+		vec<4, T, Q> gz1 = vec<4, T, Q>(0.5) - abs(gx1) - abs(gy1);
+		vec<4, T, Q> sz1 = step(gz1, vec<4, T, Q>(T(0)));
+		gx1 -= sz1 * (step(T(0), gx1) - T(0.5));
+		gy1 -= sz1 * (step(T(0), gy1) - T(0.5));
+
+		vec<3, T, Q> g000 = vec<3, T, Q>(gx0.x, gy0.x, gz0.x);
+		vec<3, T, Q> g100 = vec<3, T, Q>(gx0.y, gy0.y, gz0.y);
+		vec<3, T, Q> g010 = vec<3, T, Q>(gx0.z, gy0.z, gz0.z);
+		vec<3, T, Q> g110 = vec<3, T, Q>(gx0.w, gy0.w, gz0.w);
+		vec<3, T, Q> g001 = vec<3, T, Q>(gx1.x, gy1.x, gz1.x);
+		vec<3, T, Q> g101 = vec<3, T, Q>(gx1.y, gy1.y, gz1.y);
+		vec<3, T, Q> g011 = vec<3, T, Q>(gx1.z, gy1.z, gz1.z);
+		vec<3, T, Q> g111 = vec<3, T, Q>(gx1.w, gy1.w, gz1.w);
+
+		vec<4, T, Q> norm0 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g000, g000), dot(g010, g010), dot(g100, g100), dot(g110, g110)));
+		g000 *= norm0.x;
+		g010 *= norm0.y;
+		g100 *= norm0.z;
+		g110 *= norm0.w;
+		vec<4, T, Q> norm1 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g001, g001), dot(g011, g011), dot(g101, g101), dot(g111, g111)));
+		g001 *= norm1.x;
+		g011 *= norm1.y;
+		g101 *= norm1.z;
+		g111 *= norm1.w;
+
+		T n000 = dot(g000, Pf0);
+		T n100 = dot(g100, vec<3, T, Q>(Pf1.x, Pf0.y, Pf0.z));
+		T n010 = dot(g010, vec<3, T, Q>(Pf0.x, Pf1.y, Pf0.z));
+		T n110 = dot(g110, vec<3, T, Q>(Pf1.x, Pf1.y, Pf0.z));
+		T n001 = dot(g001, vec<3, T, Q>(Pf0.x, Pf0.y, Pf1.z));
+		T n101 = dot(g101, vec<3, T, Q>(Pf1.x, Pf0.y, Pf1.z));
+		T n011 = dot(g011, vec<3, T, Q>(Pf0.x, Pf1.y, Pf1.z));
+		T n111 = dot(g111, Pf1);
+
+		vec<3, T, Q> fade_xyz = detail::fade(Pf0);
+		vec<4, T, Q> n_z = mix(vec<4, T, Q>(n000, n100, n010, n110), vec<4, T, Q>(n001, n101, n011, n111), fade_xyz.z);
+		vec<2, T, Q> n_yz = mix(vec<2, T, Q>(n_z.x, n_z.y), vec<2, T, Q>(n_z.z, n_z.w), fade_xyz.y);
+		T n_xyz = mix(n_yz.x, n_yz.y, fade_xyz.x);
+		return T(2.2) * n_xyz;
+	}
+
+	// Classic Perlin noise, periodic version
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T perlin(vec<4, T, Q> const& Position, vec<4, T, Q> const& rep)
+	{
+		vec<4, T, Q> Pi0 = mod(floor(Position), rep); // Integer part modulo rep
+		vec<4, T, Q> Pi1 = mod(Pi0 + T(1), rep); // Integer part + 1 mod rep
+		vec<4, T, Q> Pf0 = fract(Position); // Fractional part for interpolation
+		vec<4, T, Q> Pf1 = Pf0 - T(1); // Fractional part - 1.0
+		vec<4, T, Q> ix = vec<4, T, Q>(Pi0.x, Pi1.x, Pi0.x, Pi1.x);
+		vec<4, T, Q> iy = vec<4, T, Q>(Pi0.y, Pi0.y, Pi1.y, Pi1.y);
+		vec<4, T, Q> iz0(Pi0.z);
+		vec<4, T, Q> iz1(Pi1.z);
+		vec<4, T, Q> iw0(Pi0.w);
+		vec<4, T, Q> iw1(Pi1.w);
+
+		vec<4, T, Q> ixy = detail::permute(detail::permute(ix) + iy);
+		vec<4, T, Q> ixy0 = detail::permute(ixy + iz0);
+		vec<4, T, Q> ixy1 = detail::permute(ixy + iz1);
+		vec<4, T, Q> ixy00 = detail::permute(ixy0 + iw0);
+		vec<4, T, Q> ixy01 = detail::permute(ixy0 + iw1);
+		vec<4, T, Q> ixy10 = detail::permute(ixy1 + iw0);
+		vec<4, T, Q> ixy11 = detail::permute(ixy1 + iw1);
+
+		vec<4, T, Q> gx00 = ixy00 / T(7);
+		vec<4, T, Q> gy00 = floor(gx00) / T(7);
+		vec<4, T, Q> gz00 = floor(gy00) / T(6);
+		gx00 = fract(gx00) - T(0.5);
+		gy00 = fract(gy00) - T(0.5);
+		gz00 = fract(gz00) - T(0.5);
+		vec<4, T, Q> gw00 = vec<4, T, Q>(0.75) - abs(gx00) - abs(gy00) - abs(gz00);
+		vec<4, T, Q> sw00 = step(gw00, vec<4, T, Q>(0));
+		gx00 -= sw00 * (step(T(0), gx00) - T(0.5));
+		gy00 -= sw00 * (step(T(0), gy00) - T(0.5));
+
+		vec<4, T, Q> gx01 = ixy01 / T(7);
+		vec<4, T, Q> gy01 = floor(gx01) / T(7);
+		vec<4, T, Q> gz01 = floor(gy01) / T(6);
+		gx01 = fract(gx01) - T(0.5);
+		gy01 = fract(gy01) - T(0.5);
+		gz01 = fract(gz01) - T(0.5);
+		vec<4, T, Q> gw01 = vec<4, T, Q>(0.75) - abs(gx01) - abs(gy01) - abs(gz01);
+		vec<4, T, Q> sw01 = step(gw01, vec<4, T, Q>(0.0));
+		gx01 -= sw01 * (step(T(0), gx01) - T(0.5));
+		gy01 -= sw01 * (step(T(0), gy01) - T(0.5));
+
+		vec<4, T, Q> gx10 = ixy10 / T(7);
+		vec<4, T, Q> gy10 = floor(gx10) / T(7);
+		vec<4, T, Q> gz10 = floor(gy10) / T(6);
+		gx10 = fract(gx10) - T(0.5);
+		gy10 = fract(gy10) - T(0.5);
+		gz10 = fract(gz10) - T(0.5);
+		vec<4, T, Q> gw10 = vec<4, T, Q>(0.75) - abs(gx10) - abs(gy10) - abs(gz10);
+		vec<4, T, Q> sw10 = step(gw10, vec<4, T, Q>(0.0));
+		gx10 -= sw10 * (step(T(0), gx10) - T(0.5));
+		gy10 -= sw10 * (step(T(0), gy10) - T(0.5));
+
+		vec<4, T, Q> gx11 = ixy11 / T(7);
+		vec<4, T, Q> gy11 = floor(gx11) / T(7);
+		vec<4, T, Q> gz11 = floor(gy11) / T(6);
+		gx11 = fract(gx11) - T(0.5);
+		gy11 = fract(gy11) - T(0.5);
+		gz11 = fract(gz11) - T(0.5);
+		vec<4, T, Q> gw11 = vec<4, T, Q>(0.75) - abs(gx11) - abs(gy11) - abs(gz11);
+		vec<4, T, Q> sw11 = step(gw11, vec<4, T, Q>(T(0)));
+		gx11 -= sw11 * (step(T(0), gx11) - T(0.5));
+		gy11 -= sw11 * (step(T(0), gy11) - T(0.5));
+
+		vec<4, T, Q> g0000(gx00.x, gy00.x, gz00.x, gw00.x);
+		vec<4, T, Q> g1000(gx00.y, gy00.y, gz00.y, gw00.y);
+		vec<4, T, Q> g0100(gx00.z, gy00.z, gz00.z, gw00.z);
+		vec<4, T, Q> g1100(gx00.w, gy00.w, gz00.w, gw00.w);
+		vec<4, T, Q> g0010(gx10.x, gy10.x, gz10.x, gw10.x);
+		vec<4, T, Q> g1010(gx10.y, gy10.y, gz10.y, gw10.y);
+		vec<4, T, Q> g0110(gx10.z, gy10.z, gz10.z, gw10.z);
+		vec<4, T, Q> g1110(gx10.w, gy10.w, gz10.w, gw10.w);
+		vec<4, T, Q> g0001(gx01.x, gy01.x, gz01.x, gw01.x);
+		vec<4, T, Q> g1001(gx01.y, gy01.y, gz01.y, gw01.y);
+		vec<4, T, Q> g0101(gx01.z, gy01.z, gz01.z, gw01.z);
+		vec<4, T, Q> g1101(gx01.w, gy01.w, gz01.w, gw01.w);
+		vec<4, T, Q> g0011(gx11.x, gy11.x, gz11.x, gw11.x);
+		vec<4, T, Q> g1011(gx11.y, gy11.y, gz11.y, gw11.y);
+		vec<4, T, Q> g0111(gx11.z, gy11.z, gz11.z, gw11.z);
+		vec<4, T, Q> g1111(gx11.w, gy11.w, gz11.w, gw11.w);
+
+		vec<4, T, Q> norm00 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g0000, g0000), dot(g0100, g0100), dot(g1000, g1000), dot(g1100, g1100)));
+		g0000 *= norm00.x;
+		g0100 *= norm00.y;
+		g1000 *= norm00.z;
+		g1100 *= norm00.w;
+
+		vec<4, T, Q> norm01 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g0001, g0001), dot(g0101, g0101), dot(g1001, g1001), dot(g1101, g1101)));
+		g0001 *= norm01.x;
+		g0101 *= norm01.y;
+		g1001 *= norm01.z;
+		g1101 *= norm01.w;
+
+		vec<4, T, Q> norm10 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g0010, g0010), dot(g0110, g0110), dot(g1010, g1010), dot(g1110, g1110)));
+		g0010 *= norm10.x;
+		g0110 *= norm10.y;
+		g1010 *= norm10.z;
+		g1110 *= norm10.w;
+
+		vec<4, T, Q> norm11 = detail::taylorInvSqrt(vec<4, T, Q>(dot(g0011, g0011), dot(g0111, g0111), dot(g1011, g1011), dot(g1111, g1111)));
+		g0011 *= norm11.x;
+		g0111 *= norm11.y;
+		g1011 *= norm11.z;
+		g1111 *= norm11.w;
+
+		T n0000 = dot(g0000, Pf0);
+		T n1000 = dot(g1000, vec<4, T, Q>(Pf1.x, Pf0.y, Pf0.z, Pf0.w));
+		T n0100 = dot(g0100, vec<4, T, Q>(Pf0.x, Pf1.y, Pf0.z, Pf0.w));
+		T n1100 = dot(g1100, vec<4, T, Q>(Pf1.x, Pf1.y, Pf0.z, Pf0.w));
+		T n0010 = dot(g0010, vec<4, T, Q>(Pf0.x, Pf0.y, Pf1.z, Pf0.w));
+		T n1010 = dot(g1010, vec<4, T, Q>(Pf1.x, Pf0.y, Pf1.z, Pf0.w));
+		T n0110 = dot(g0110, vec<4, T, Q>(Pf0.x, Pf1.y, Pf1.z, Pf0.w));
+		T n1110 = dot(g1110, vec<4, T, Q>(Pf1.x, Pf1.y, Pf1.z, Pf0.w));
+		T n0001 = dot(g0001, vec<4, T, Q>(Pf0.x, Pf0.y, Pf0.z, Pf1.w));
+		T n1001 = dot(g1001, vec<4, T, Q>(Pf1.x, Pf0.y, Pf0.z, Pf1.w));
+		T n0101 = dot(g0101, vec<4, T, Q>(Pf0.x, Pf1.y, Pf0.z, Pf1.w));
+		T n1101 = dot(g1101, vec<4, T, Q>(Pf1.x, Pf1.y, Pf0.z, Pf1.w));
+		T n0011 = dot(g0011, vec<4, T, Q>(Pf0.x, Pf0.y, Pf1.z, Pf1.w));
+		T n1011 = dot(g1011, vec<4, T, Q>(Pf1.x, Pf0.y, Pf1.z, Pf1.w));
+		T n0111 = dot(g0111, vec<4, T, Q>(Pf0.x, Pf1.y, Pf1.z, Pf1.w));
+		T n1111 = dot(g1111, Pf1);
+
+		vec<4, T, Q> fade_xyzw = detail::fade(Pf0);
+		vec<4, T, Q> n_0w = mix(vec<4, T, Q>(n0000, n1000, n0100, n1100), vec<4, T, Q>(n0001, n1001, n0101, n1101), fade_xyzw.w);
+		vec<4, T, Q> n_1w = mix(vec<4, T, Q>(n0010, n1010, n0110, n1110), vec<4, T, Q>(n0011, n1011, n0111, n1111), fade_xyzw.w);
+		vec<4, T, Q> n_zw = mix(n_0w, n_1w, fade_xyzw.z);
+		vec<2, T, Q> n_yzw = mix(vec<2, T, Q>(n_zw.x, n_zw.y), vec<2, T, Q>(n_zw.z, n_zw.w), fade_xyzw.y);
+		T n_xyzw = mix(n_yzw.x, n_yzw.y, fade_xyzw.x);
+		return T(2.2) * n_xyzw;
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T simplex(glm::vec<2, T, Q> const& v)
+	{
+		vec<4, T, Q> const C = vec<4, T, Q>(
+			T( 0.211324865405187),  // (3.0 -  sqrt(3.0)) / 6.0
+			T( 0.366025403784439),  //  0.5 * (sqrt(3.0)  - 1.0)
+			T(-0.577350269189626),	// -1.0 + 2.0 * C.x
+			T( 0.024390243902439)); //  1.0 / 41.0
+
+		// First corner
+		vec<2, T, Q> i  = floor(v + dot(v, vec<2, T, Q>(C[1])));
+		vec<2, T, Q> x0 = v -   i + dot(i, vec<2, T, Q>(C[0]));
+
+		// Other corners
+		//i1.x = step( x0.y, x0.x ); // x0.x > x0.y ? 1.0 : 0.0
+		//i1.y = 1.0 - i1.x;
+		vec<2, T, Q> i1 = (x0.x > x0.y) ? vec<2, T, Q>(1, 0) : vec<2, T, Q>(0, 1);
+		// x0 = x0 - 0.0 + 0.0 * C.xx ;
+		// x1 = x0 - i1 + 1.0 * C.xx ;
+		// x2 = x0 - 1.0 + 2.0 * C.xx ;
+		vec<4, T, Q> x12 = vec<4, T, Q>(x0.x, x0.y, x0.x, x0.y) + vec<4, T, Q>(C.x, C.x, C.z, C.z);
+		x12 = vec<4, T, Q>(vec<2, T, Q>(x12) - i1, x12.z, x12.w);
+
+		// Permutations
+		i = mod(i, vec<2, T, Q>(289)); // Avoid truncation effects in permutation
+		vec<3, T, Q> p = detail::permute(
+			detail::permute(i.y + vec<3, T, Q>(T(0), i1.y, T(1)))
+			+ i.x + vec<3, T, Q>(T(0), i1.x, T(1)));
+
+		vec<3, T, Q> m = max(vec<3, T, Q>(0.5) - vec<3, T, Q>(
+			dot(x0, x0),
+			dot(vec<2, T, Q>(x12.x, x12.y), vec<2, T, Q>(x12.x, x12.y)),
+			dot(vec<2, T, Q>(x12.z, x12.w), vec<2, T, Q>(x12.z, x12.w))), vec<3, T, Q>(0));
+		m = m * m ;
+		m = m * m ;
+
+		// Gradients: 41 points uniformly over a line, mapped onto a diamond.
+		// The ring size 17*17 = 289 is close to a multiple of 41 (41*7 = 287)
+
+		vec<3, T, Q> x = static_cast<T>(2) * fract(p * C.w) - T(1);
+		vec<3, T, Q> h = abs(x) - T(0.5);
+		vec<3, T, Q> ox = floor(x + T(0.5));
+		vec<3, T, Q> a0 = x - ox;
+
+		// Normalise gradients implicitly by scaling m
+		// Inlined for speed: m *= taylorInvSqrt( a0*a0 + h*h );
+		m *= static_cast<T>(1.79284291400159) - T(0.85373472095314) * (a0 * a0 + h * h);
+
+		// Compute final noise value at P
+		vec<3, T, Q> g;
+		g.x  = a0.x  * x0.x  + h.x  * x0.y;
+		//g.yz = a0.yz * x12.xz + h.yz * x12.yw;
+		g.y = a0.y * x12.x + h.y * x12.y;
+		g.z = a0.z * x12.z + h.z * x12.w;
+		return T(130) * dot(m, g);
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T simplex(vec<3, T, Q> const& v)
+	{
+		vec<2, T, Q> const C(1.0 / 6.0, 1.0 / 3.0);
+		vec<4, T, Q> const D(0.0, 0.5, 1.0, 2.0);
+
+		// First corner
+		vec<3, T, Q> i(floor(v + dot(v, vec<3, T, Q>(C.y))));
+		vec<3, T, Q> x0(v - i + dot(i, vec<3, T, Q>(C.x)));
+
+		// Other corners
+		vec<3, T, Q> g(step(vec<3, T, Q>(x0.y, x0.z, x0.x), x0));
+		vec<3, T, Q> l(T(1) - g);
+		vec<3, T, Q> i1(min(g, vec<3, T, Q>(l.z, l.x, l.y)));
+		vec<3, T, Q> i2(max(g, vec<3, T, Q>(l.z, l.x, l.y)));
+
+		//   x0 = x0 - 0.0 + 0.0 * C.xxx;
+		//   x1 = x0 - i1  + 1.0 * C.xxx;
+		//   x2 = x0 - i2  + 2.0 * C.xxx;
+		//   x3 = x0 - 1.0 + 3.0 * C.xxx;
+		vec<3, T, Q> x1(x0 - i1 + C.x);
+		vec<3, T, Q> x2(x0 - i2 + C.y); // 2.0*C.x = 1/3 = C.y
+		vec<3, T, Q> x3(x0 - D.y);      // -1.0+3.0*C.x = -0.5 = -D.y
+
+		// Permutations
+		i = detail::mod289(i);
+		vec<4, T, Q> p(detail::permute(detail::permute(detail::permute(
+			i.z + vec<4, T, Q>(T(0), i1.z, i2.z, T(1))) +
+			i.y + vec<4, T, Q>(T(0), i1.y, i2.y, T(1))) +
+			i.x + vec<4, T, Q>(T(0), i1.x, i2.x, T(1))));
+
+		// Gradients: 7x7 points over a square, mapped onto an octahedron.
+		// The ring size 17*17 = 289 is close to a multiple of 49 (49*6 = 294)
+		T n_ = static_cast<T>(0.142857142857); // 1.0/7.0
+		vec<3, T, Q> ns(n_ * vec<3, T, Q>(D.w, D.y, D.z) - vec<3, T, Q>(D.x, D.z, D.x));
+
+		vec<4, T, Q> j(p - T(49) * floor(p * ns.z * ns.z));  //  mod(p,7*7)
+
+		vec<4, T, Q> x_(floor(j * ns.z));
+		vec<4, T, Q> y_(floor(j - T(7) * x_));    // mod(j,N)
+
+		vec<4, T, Q> x(x_ * ns.x + ns.y);
+		vec<4, T, Q> y(y_ * ns.x + ns.y);
+		vec<4, T, Q> h(T(1) - abs(x) - abs(y));
+
+		vec<4, T, Q> b0(x.x, x.y, y.x, y.y);
+		vec<4, T, Q> b1(x.z, x.w, y.z, y.w);
+
+		// vec4 s0 = vec4(lessThan(b0,0.0))*2.0 - 1.0;
+		// vec4 s1 = vec4(lessThan(b1,0.0))*2.0 - 1.0;
+		vec<4, T, Q> s0(floor(b0) * T(2) + T(1));
+		vec<4, T, Q> s1(floor(b1) * T(2) + T(1));
+		vec<4, T, Q> sh(-step(h, vec<4, T, Q>(0.0)));
+
+		vec<4, T, Q> a0 = vec<4, T, Q>(b0.x, b0.z, b0.y, b0.w) + vec<4, T, Q>(s0.x, s0.z, s0.y, s0.w) * vec<4, T, Q>(sh.x, sh.x, sh.y, sh.y);
+		vec<4, T, Q> a1 = vec<4, T, Q>(b1.x, b1.z, b1.y, b1.w) + vec<4, T, Q>(s1.x, s1.z, s1.y, s1.w) * vec<4, T, Q>(sh.z, sh.z, sh.w, sh.w);
+
+		vec<3, T, Q> p0(a0.x, a0.y, h.x);
+		vec<3, T, Q> p1(a0.z, a0.w, h.y);
+		vec<3, T, Q> p2(a1.x, a1.y, h.z);
+		vec<3, T, Q> p3(a1.z, a1.w, h.w);
+
+		// Normalise gradients
+		vec<4, T, Q> norm = detail::taylorInvSqrt(vec<4, T, Q>(dot(p0, p0), dot(p1, p1), dot(p2, p2), dot(p3, p3)));
+		p0 *= norm.x;
+		p1 *= norm.y;
+		p2 *= norm.z;
+		p3 *= norm.w;
+
+		// Mix final noise value
+		vec<4, T, Q> m = max(T(0.6) - vec<4, T, Q>(dot(x0, x0), dot(x1, x1), dot(x2, x2), dot(x3, x3)), vec<4, T, Q>(0));
+		m = m * m;
+		return T(42) * dot(m * m, vec<4, T, Q>(dot(p0, x0), dot(p1, x1), dot(p2, x2), dot(p3, x3)));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T simplex(vec<4, T, Q> const& v)
+	{
+		vec<4, T, Q> const C(
+			0.138196601125011,  // (5 - sqrt(5))/20  G4
+			0.276393202250021,  // 2 * G4
+			0.414589803375032,  // 3 * G4
+			-0.447213595499958); // -1 + 4 * G4
+
+		// (sqrt(5) - 1)/4 = F4, used once below
+		T const F4 = static_cast<T>(0.309016994374947451);
+
+		// First corner
+		vec<4, T, Q> i  = floor(v + dot(v, vec<4, T, Q>(F4)));
+		vec<4, T, Q> x0 = v -   i + dot(i, vec<4, T, Q>(C.x));
+
+		// Other corners
+
+		// Rank sorting originally contributed by Bill Licea-Kane, AMD (formerly ATI)
+		vec<4, T, Q> i0;
+		vec<3, T, Q> isX = step(vec<3, T, Q>(x0.y, x0.z, x0.w), vec<3, T, Q>(x0.x));
+		vec<3, T, Q> isYZ = step(vec<3, T, Q>(x0.z, x0.w, x0.w), vec<3, T, Q>(x0.y, x0.y, x0.z));
+		//  i0.x = dot(isX, vec3(1.0));
+		//i0.x = isX.x + isX.y + isX.z;
+		//i0.yzw = static_cast<T>(1) - isX;
+		i0 = vec<4, T, Q>(isX.x + isX.y + isX.z, T(1) - isX);
+		//  i0.y += dot(isYZ.xy, vec2(1.0));
+		i0.y += isYZ.x + isYZ.y;
+		//i0.zw += 1.0 - vec<2, T, Q>(isYZ.x, isYZ.y);
+		i0.z += static_cast<T>(1) - isYZ.x;
+		i0.w += static_cast<T>(1) - isYZ.y;
+		i0.z += isYZ.z;
+		i0.w += static_cast<T>(1) - isYZ.z;
+
+		// i0 now contains the unique values 0,1,2,3 in each channel
+		vec<4, T, Q> i3 = clamp(i0, T(0), T(1));
+		vec<4, T, Q> i2 = clamp(i0 - T(1), T(0), T(1));
+		vec<4, T, Q> i1 = clamp(i0 - T(2), T(0), T(1));
+
+		//  x0 = x0 - 0.0 + 0.0 * C.xxxx
+		//  x1 = x0 - i1  + 0.0 * C.xxxx
+		//  x2 = x0 - i2  + 0.0 * C.xxxx
+		//  x3 = x0 - i3  + 0.0 * C.xxxx
+		//  x4 = x0 - 1.0 + 4.0 * C.xxxx
+		vec<4, T, Q> x1 = x0 - i1 + C.x;
+		vec<4, T, Q> x2 = x0 - i2 + C.y;
+		vec<4, T, Q> x3 = x0 - i3 + C.z;
+		vec<4, T, Q> x4 = x0 + C.w;
+
+		// Permutations
+		i = mod(i, vec<4, T, Q>(289));
+		T j0 = detail::permute(detail::permute(detail::permute(detail::permute(i.w) + i.z) + i.y) + i.x);
+		vec<4, T, Q> j1 = detail::permute(detail::permute(detail::permute(detail::permute(
+			i.w + vec<4, T, Q>(i1.w, i2.w, i3.w, T(1))) +
+			i.z + vec<4, T, Q>(i1.z, i2.z, i3.z, T(1))) +
+			i.y + vec<4, T, Q>(i1.y, i2.y, i3.y, T(1))) +
+			i.x + vec<4, T, Q>(i1.x, i2.x, i3.x, T(1)));
+
+		// Gradients: 7x7x6 points over a cube, mapped onto a 4-cross polytope
+		// 7*7*6 = 294, which is close to the ring size 17*17 = 289.
+		vec<4, T, Q> ip = vec<4, T, Q>(T(1) / T(294), T(1) / T(49), T(1) / T(7), T(0));
+
+		vec<4, T, Q> p0 = gtc::grad4(j0,   ip);
+		vec<4, T, Q> p1 = gtc::grad4(j1.x, ip);
+		vec<4, T, Q> p2 = gtc::grad4(j1.y, ip);
+		vec<4, T, Q> p3 = gtc::grad4(j1.z, ip);
+		vec<4, T, Q> p4 = gtc::grad4(j1.w, ip);
+
+		// Normalise gradients
+		vec<4, T, Q> norm = detail::taylorInvSqrt(vec<4, T, Q>(dot(p0, p0), dot(p1, p1), dot(p2, p2), dot(p3, p3)));
+		p0 *= norm.x;
+		p1 *= norm.y;
+		p2 *= norm.z;
+		p3 *= norm.w;
+		p4 *= detail::taylorInvSqrt(dot(p4, p4));
+
+		// Mix contributions from the five corners
+		vec<3, T, Q> m0 = max(T(0.6) - vec<3, T, Q>(dot(x0, x0), dot(x1, x1), dot(x2, x2)), vec<3, T, Q>(0));
+		vec<2, T, Q> m1 = max(T(0.6) - vec<2, T, Q>(dot(x3, x3), dot(x4, x4)             ), vec<2, T, Q>(0));
+		m0 = m0 * m0;
+		m1 = m1 * m1;
+		return T(49) *
+			(dot(m0 * m0, vec<3, T, Q>(dot(p0, x0), dot(p1, x1), dot(p2, x2))) +
+			dot(m1 * m1, vec<2, T, Q>(dot(p3, x3), dot(p4, x4))));
+	}
+}//namespace glm

cuda_extensions/third_party/glm/glm/gtc/packing.hpp

@@ -0,0 +1,728 @@
+/// @ref gtc_packing
+/// @file glm/gtc/packing.hpp
+///
+/// @see core (dependence)
+///
+/// @defgroup gtc_packing GLM_GTC_packing
+/// @ingroup gtc
+///
+/// Include <glm/gtc/packing.hpp> to use the features of this extension.
+///
+/// This extension provides a set of function to convert vertors to packed
+/// formats.
+
+#pragma once
+
+// Dependency:
+#include "type_precision.hpp"
+#include "../ext/vector_packing.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_packing extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_packing
+	/// @{
+
+	/// First, converts the normalized floating-point value v into a 8-bit integer value.
+	/// Then, the results are packed into the returned 8-bit unsigned integer.
+	///
+	/// The conversion for component c of v to fixed point is done as follows:
+	/// packUnorm1x8:	round(clamp(c, 0, +1) * 255.0)
+	///
+	/// @see gtc_packing
+	/// @see uint16 packUnorm2x8(vec2 const& v)
+	/// @see uint32 packUnorm4x8(vec4 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packUnorm4x8.xml">GLSL packUnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint8 packUnorm1x8(float v);
+
+	/// Convert a single 8-bit integer to a normalized floating-point value.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackUnorm4x8: f / 255.0
+	///
+	/// @see gtc_packing
+	/// @see vec2 unpackUnorm2x8(uint16 p)
+	/// @see vec4 unpackUnorm4x8(uint32 p)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackUnorm4x8.xml">GLSL unpackUnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL float unpackUnorm1x8(uint8 p);
+
+	/// First, converts each component of the normalized floating-point value v into 8-bit integer values.
+	/// Then, the results are packed into the returned 16-bit unsigned integer.
+	///
+	/// The conversion for component c of v to fixed point is done as follows:
+	/// packUnorm2x8:	round(clamp(c, 0, +1) * 255.0)
+	///
+	/// The first component of the vector will be written to the least significant bits of the output;
+	/// the last component will be written to the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint8 packUnorm1x8(float const& v)
+	/// @see uint32 packUnorm4x8(vec4 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packUnorm4x8.xml">GLSL packUnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint16 packUnorm2x8(vec2 const& v);
+
+	/// First, unpacks a single 16-bit unsigned integer p into a pair of 8-bit unsigned integers.
+	/// Then, each component is converted to a normalized floating-point value to generate the returned two-component vector.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackUnorm4x8: f / 255.0
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see float unpackUnorm1x8(uint8 v)
+	/// @see vec4 unpackUnorm4x8(uint32 p)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackUnorm4x8.xml">GLSL unpackUnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL vec2 unpackUnorm2x8(uint16 p);
+
+	/// First, converts the normalized floating-point value v into 8-bit integer value.
+	/// Then, the results are packed into the returned 8-bit unsigned integer.
+	///
+	/// The conversion to fixed point is done as follows:
+	/// packSnorm1x8:	round(clamp(s, -1, +1) * 127.0)
+	///
+	/// @see gtc_packing
+	/// @see uint16 packSnorm2x8(vec2 const& v)
+	/// @see uint32 packSnorm4x8(vec4 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packSnorm4x8.xml">GLSL packSnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint8 packSnorm1x8(float s);
+
+	/// First, unpacks a single 8-bit unsigned integer p into a single 8-bit signed integers.
+	/// Then, the value is converted to a normalized floating-point value to generate the returned scalar.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackSnorm1x8: clamp(f / 127.0, -1, +1)
+	///
+	/// @see gtc_packing
+	/// @see vec2 unpackSnorm2x8(uint16 p)
+	/// @see vec4 unpackSnorm4x8(uint32 p)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackSnorm4x8.xml">GLSL unpackSnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL float unpackSnorm1x8(uint8 p);
+
+	/// First, converts each component of the normalized floating-point value v into 8-bit integer values.
+	/// Then, the results are packed into the returned 16-bit unsigned integer.
+	///
+	/// The conversion for component c of v to fixed point is done as follows:
+	/// packSnorm2x8:	round(clamp(c, -1, +1) * 127.0)
+	///
+	/// The first component of the vector will be written to the least significant bits of the output;
+	/// the last component will be written to the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint8 packSnorm1x8(float const& v)
+	/// @see uint32 packSnorm4x8(vec4 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packSnorm4x8.xml">GLSL packSnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint16 packSnorm2x8(vec2 const& v);
+
+	/// First, unpacks a single 16-bit unsigned integer p into a pair of 8-bit signed integers.
+	/// Then, each component is converted to a normalized floating-point value to generate the returned two-component vector.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackSnorm2x8: clamp(f / 127.0, -1, +1)
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see float unpackSnorm1x8(uint8 p)
+	/// @see vec4 unpackSnorm4x8(uint32 p)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackSnorm4x8.xml">GLSL unpackSnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL vec2 unpackSnorm2x8(uint16 p);
+
+	/// First, converts the normalized floating-point value v into a 16-bit integer value.
+	/// Then, the results are packed into the returned 16-bit unsigned integer.
+	///
+	/// The conversion for component c of v to fixed point is done as follows:
+	/// packUnorm1x16:	round(clamp(c, 0, +1) * 65535.0)
+	///
+	/// @see gtc_packing
+	/// @see uint16 packSnorm1x16(float const& v)
+	/// @see uint64 packSnorm4x16(vec4 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packUnorm4x8.xml">GLSL packUnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint16 packUnorm1x16(float v);
+
+	/// First, unpacks a single 16-bit unsigned integer p into a of 16-bit unsigned integers.
+	/// Then, the value is converted to a normalized floating-point value to generate the returned scalar.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackUnorm1x16: f / 65535.0
+	///
+	/// @see gtc_packing
+	/// @see vec2 unpackUnorm2x16(uint32 p)
+	/// @see vec4 unpackUnorm4x16(uint64 p)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackUnorm2x16.xml">GLSL unpackUnorm2x16 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL float unpackUnorm1x16(uint16 p);
+
+	/// First, converts each component of the normalized floating-point value v into 16-bit integer values.
+	/// Then, the results are packed into the returned 64-bit unsigned integer.
+	///
+	/// The conversion for component c of v to fixed point is done as follows:
+	/// packUnorm4x16:	round(clamp(c, 0, +1) * 65535.0)
+	///
+	/// The first component of the vector will be written to the least significant bits of the output;
+	/// the last component will be written to the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint16 packUnorm1x16(float const& v)
+	/// @see uint32 packUnorm2x16(vec2 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packUnorm4x8.xml">GLSL packUnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint64 packUnorm4x16(vec4 const& v);
+
+	/// First, unpacks a single 64-bit unsigned integer p into four 16-bit unsigned integers.
+	/// Then, each component is converted to a normalized floating-point value to generate the returned four-component vector.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackUnormx4x16: f / 65535.0
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see float unpackUnorm1x16(uint16 p)
+	/// @see vec2 unpackUnorm2x16(uint32 p)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackUnorm2x16.xml">GLSL unpackUnorm2x16 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL vec4 unpackUnorm4x16(uint64 p);
+
+	/// First, converts the normalized floating-point value v into 16-bit integer value.
+	/// Then, the results are packed into the returned 16-bit unsigned integer.
+	///
+	/// The conversion to fixed point is done as follows:
+	/// packSnorm1x8:	round(clamp(s, -1, +1) * 32767.0)
+	///
+	/// @see gtc_packing
+	/// @see uint32 packSnorm2x16(vec2 const& v)
+	/// @see uint64 packSnorm4x16(vec4 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packSnorm4x8.xml">GLSL packSnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint16 packSnorm1x16(float v);
+
+	/// First, unpacks a single 16-bit unsigned integer p into a single 16-bit signed integers.
+	/// Then, each component is converted to a normalized floating-point value to generate the returned scalar.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackSnorm1x16: clamp(f / 32767.0, -1, +1)
+	///
+	/// @see gtc_packing
+	/// @see vec2 unpackSnorm2x16(uint32 p)
+	/// @see vec4 unpackSnorm4x16(uint64 p)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackSnorm1x16.xml">GLSL unpackSnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL float unpackSnorm1x16(uint16 p);
+
+	/// First, converts each component of the normalized floating-point value v into 16-bit integer values.
+	/// Then, the results are packed into the returned 64-bit unsigned integer.
+	///
+	/// The conversion for component c of v to fixed point is done as follows:
+	/// packSnorm2x8:	round(clamp(c, -1, +1) * 32767.0)
+	///
+	/// The first component of the vector will be written to the least significant bits of the output;
+	/// the last component will be written to the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint16 packSnorm1x16(float const& v)
+	/// @see uint32 packSnorm2x16(vec2 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packSnorm4x8.xml">GLSL packSnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint64 packSnorm4x16(vec4 const& v);
+
+	/// First, unpacks a single 64-bit unsigned integer p into four 16-bit signed integers.
+	/// Then, each component is converted to a normalized floating-point value to generate the returned four-component vector.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackSnorm4x16: clamp(f / 32767.0, -1, +1)
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see float unpackSnorm1x16(uint16 p)
+	/// @see vec2 unpackSnorm2x16(uint32 p)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackSnorm2x16.xml">GLSL unpackSnorm4x8 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL vec4 unpackSnorm4x16(uint64 p);
+
+	/// Returns an unsigned integer obtained by converting the components of a floating-point scalar
+	/// to the 16-bit floating-point representation found in the OpenGL Specification,
+	/// and then packing this 16-bit value into a 16-bit unsigned integer.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packHalf2x16(vec2 const& v)
+	/// @see uint64 packHalf4x16(vec4 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packHalf2x16.xml">GLSL packHalf2x16 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint16 packHalf1x16(float v);
+
+	/// Returns a floating-point scalar with components obtained by unpacking a 16-bit unsigned integer into a 16-bit value,
+	/// interpreted as a 16-bit floating-point number according to the OpenGL Specification,
+	/// and converting it to 32-bit floating-point values.
+	///
+	/// @see gtc_packing
+	/// @see vec2 unpackHalf2x16(uint32 const& v)
+	/// @see vec4 unpackHalf4x16(uint64 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackHalf2x16.xml">GLSL unpackHalf2x16 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL float unpackHalf1x16(uint16 v);
+
+	/// Returns an unsigned integer obtained by converting the components of a four-component floating-point vector
+	/// to the 16-bit floating-point representation found in the OpenGL Specification,
+	/// and then packing these four 16-bit values into a 64-bit unsigned integer.
+	/// The first vector component specifies the 16 least-significant bits of the result;
+	/// the forth component specifies the 16 most-significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint16 packHalf1x16(float const& v)
+	/// @see uint32 packHalf2x16(vec2 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/packHalf2x16.xml">GLSL packHalf2x16 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL uint64 packHalf4x16(vec4 const& v);
+
+	/// Returns a four-component floating-point vector with components obtained by unpacking a 64-bit unsigned integer into four 16-bit values,
+	/// interpreting those values as 16-bit floating-point numbers according to the OpenGL Specification,
+	/// and converting them to 32-bit floating-point values.
+	/// The first component of the vector is obtained from the 16 least-significant bits of v;
+	/// the forth component is obtained from the 16 most-significant bits of v.
+	///
+	/// @see gtc_packing
+	/// @see float unpackHalf1x16(uint16 const& v)
+	/// @see vec2 unpackHalf2x16(uint32 const& v)
+	/// @see <a href="http://www.opengl.org/sdk/docs/manglsl/xhtml/unpackHalf2x16.xml">GLSL unpackHalf2x16 man page</a>
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	GLM_FUNC_DECL vec4 unpackHalf4x16(uint64 p);
+
+	/// Returns an unsigned integer obtained by converting the components of a four-component signed integer vector
+	/// to the 10-10-10-2-bit signed integer representation found in the OpenGL Specification,
+	/// and then packing these four values into a 32-bit unsigned integer.
+	/// The first vector component specifies the 10 least-significant bits of the result;
+	/// the forth component specifies the 2 most-significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packI3x10_1x2(uvec4 const& v)
+	/// @see uint32 packSnorm3x10_1x2(vec4 const& v)
+	/// @see uint32 packUnorm3x10_1x2(vec4 const& v)
+	/// @see ivec4 unpackI3x10_1x2(uint32 const& p)
+	GLM_FUNC_DECL uint32 packI3x10_1x2(ivec4 const& v);
+
+	/// Unpacks a single 32-bit unsigned integer p into three 10-bit and one 2-bit signed integers.
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packU3x10_1x2(uvec4 const& v)
+	/// @see vec4 unpackSnorm3x10_1x2(uint32 const& p);
+	/// @see uvec4 unpackI3x10_1x2(uint32 const& p);
+	GLM_FUNC_DECL ivec4 unpackI3x10_1x2(uint32 p);
+
+	/// Returns an unsigned integer obtained by converting the components of a four-component unsigned integer vector
+	/// to the 10-10-10-2-bit unsigned integer representation found in the OpenGL Specification,
+	/// and then packing these four values into a 32-bit unsigned integer.
+	/// The first vector component specifies the 10 least-significant bits of the result;
+	/// the forth component specifies the 2 most-significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packI3x10_1x2(ivec4 const& v)
+	/// @see uint32 packSnorm3x10_1x2(vec4 const& v)
+	/// @see uint32 packUnorm3x10_1x2(vec4 const& v)
+	/// @see ivec4 unpackU3x10_1x2(uint32 const& p)
+	GLM_FUNC_DECL uint32 packU3x10_1x2(uvec4 const& v);
+
+	/// Unpacks a single 32-bit unsigned integer p into three 10-bit and one 2-bit unsigned integers.
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packU3x10_1x2(uvec4 const& v)
+	/// @see vec4 unpackSnorm3x10_1x2(uint32 const& p);
+	/// @see uvec4 unpackI3x10_1x2(uint32 const& p);
+	GLM_FUNC_DECL uvec4 unpackU3x10_1x2(uint32 p);
+
+	/// First, converts the first three components of the normalized floating-point value v into 10-bit signed integer values.
+	/// Then, converts the forth component of the normalized floating-point value v into 2-bit signed integer values.
+	/// Then, the results are packed into the returned 32-bit unsigned integer.
+	///
+	/// The conversion for component c of v to fixed point is done as follows:
+	/// packSnorm3x10_1x2(xyz):	round(clamp(c, -1, +1) * 511.0)
+	/// packSnorm3x10_1x2(w):	round(clamp(c, -1, +1) * 1.0)
+	///
+	/// The first vector component specifies the 10 least-significant bits of the result;
+	/// the forth component specifies the 2 most-significant bits.
+	///
+	/// @see gtc_packing
+	/// @see vec4 unpackSnorm3x10_1x2(uint32 const& p)
+	/// @see uint32 packUnorm3x10_1x2(vec4 const& v)
+	/// @see uint32 packU3x10_1x2(uvec4 const& v)
+	/// @see uint32 packI3x10_1x2(ivec4 const& v)
+	GLM_FUNC_DECL uint32 packSnorm3x10_1x2(vec4 const& v);
+
+	/// First, unpacks a single 32-bit unsigned integer p into four 16-bit signed integers.
+	/// Then, each component is converted to a normalized floating-point value to generate the returned four-component vector.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackSnorm3x10_1x2(xyz): clamp(f / 511.0, -1, +1)
+	/// unpackSnorm3x10_1x2(w): clamp(f / 511.0, -1, +1)
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packSnorm3x10_1x2(vec4 const& v)
+	/// @see vec4 unpackUnorm3x10_1x2(uint32 const& p))
+	/// @see uvec4 unpackI3x10_1x2(uint32 const& p)
+	/// @see uvec4 unpackU3x10_1x2(uint32 const& p)
+	GLM_FUNC_DECL vec4 unpackSnorm3x10_1x2(uint32 p);
+
+	/// First, converts the first three components of the normalized floating-point value v into 10-bit unsigned integer values.
+	/// Then, converts the forth component of the normalized floating-point value v into 2-bit signed uninteger values.
+	/// Then, the results are packed into the returned 32-bit unsigned integer.
+	///
+	/// The conversion for component c of v to fixed point is done as follows:
+	/// packUnorm3x10_1x2(xyz):	round(clamp(c, 0, +1) * 1023.0)
+	/// packUnorm3x10_1x2(w):	round(clamp(c, 0, +1) * 3.0)
+	///
+	/// The first vector component specifies the 10 least-significant bits of the result;
+	/// the forth component specifies the 2 most-significant bits.
+	///
+	/// @see gtc_packing
+	/// @see vec4 unpackUnorm3x10_1x2(uint32 const& p)
+	/// @see uint32 packUnorm3x10_1x2(vec4 const& v)
+	/// @see uint32 packU3x10_1x2(uvec4 const& v)
+	/// @see uint32 packI3x10_1x2(ivec4 const& v)
+	GLM_FUNC_DECL uint32 packUnorm3x10_1x2(vec4 const& v);
+
+	/// First, unpacks a single 32-bit unsigned integer p into four 16-bit signed integers.
+	/// Then, each component is converted to a normalized floating-point value to generate the returned four-component vector.
+	///
+	/// The conversion for unpacked fixed-point value f to floating point is done as follows:
+	/// unpackSnorm3x10_1x2(xyz): clamp(f / 1023.0, 0, +1)
+	/// unpackSnorm3x10_1x2(w): clamp(f / 3.0, 0, +1)
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packSnorm3x10_1x2(vec4 const& v)
+	/// @see vec4 unpackInorm3x10_1x2(uint32 const& p))
+	/// @see uvec4 unpackI3x10_1x2(uint32 const& p)
+	/// @see uvec4 unpackU3x10_1x2(uint32 const& p)
+	GLM_FUNC_DECL vec4 unpackUnorm3x10_1x2(uint32 p);
+
+	/// First, converts the first two components of the normalized floating-point value v into 11-bit signless floating-point values.
+	/// Then, converts the third component of the normalized floating-point value v into a 10-bit signless floating-point value.
+	/// Then, the results are packed into the returned 32-bit unsigned integer.
+	///
+	/// The first vector component specifies the 11 least-significant bits of the result;
+	/// the last component specifies the 10 most-significant bits.
+	///
+	/// @see gtc_packing
+	/// @see vec3 unpackF2x11_1x10(uint32 const& p)
+	GLM_FUNC_DECL uint32 packF2x11_1x10(vec3 const& v);
+
+	/// First, unpacks a single 32-bit unsigned integer p into two 11-bit signless floating-point values and one 10-bit signless floating-point value .
+	/// Then, each component is converted to a normalized floating-point value to generate the returned three-component vector.
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packF2x11_1x10(vec3 const& v)
+	GLM_FUNC_DECL vec3 unpackF2x11_1x10(uint32 p);
+
+
+	/// First, converts the first two components of the normalized floating-point value v into 11-bit signless floating-point values.
+	/// Then, converts the third component of the normalized floating-point value v into a 10-bit signless floating-point value.
+	/// Then, the results are packed into the returned 32-bit unsigned integer.
+	///
+	/// The first vector component specifies the 11 least-significant bits of the result;
+	/// the last component specifies the 10 most-significant bits.
+	///
+	/// packF3x9_E1x5 allows encoding into RGBE / RGB9E5 format
+	///
+	/// @see gtc_packing
+	/// @see vec3 unpackF3x9_E1x5(uint32 const& p)
+	GLM_FUNC_DECL uint32 packF3x9_E1x5(vec3 const& v);
+
+	/// First, unpacks a single 32-bit unsigned integer p into two 11-bit signless floating-point values and one 10-bit signless floating-point value .
+	/// Then, each component is converted to a normalized floating-point value to generate the returned three-component vector.
+	///
+	/// The first component of the returned vector will be extracted from the least significant bits of the input;
+	/// the last component will be extracted from the most significant bits.
+	///
+	/// unpackF3x9_E1x5 allows decoding RGBE / RGB9E5 data
+	///
+	/// @see gtc_packing
+	/// @see uint32 packF3x9_E1x5(vec3 const& v)
+	GLM_FUNC_DECL vec3 unpackF3x9_E1x5(uint32 p);
+
+	/// Returns an unsigned integer vector obtained by converting the components of a floating-point vector
+	/// to the 16-bit floating-point representation found in the OpenGL Specification.
+	/// The first vector component specifies the 16 least-significant bits of the result;
+	/// the forth component specifies the 16 most-significant bits.
+	///
+	/// @see gtc_packing
+	/// @see vec<3, T, Q> unpackRGBM(vec<4, T, Q> const& p)
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<4, T, Q> packRGBM(vec<3, T, Q> const& rgb);
+
+	/// Returns a floating-point vector with components obtained by reinterpreting an integer vector as 16-bit floating-point numbers and converting them to 32-bit floating-point values.
+	/// The first component of the vector is obtained from the 16 least-significant bits of v;
+	/// the forth component is obtained from the 16 most-significant bits of v.
+	///
+	/// @see gtc_packing
+	/// @see vec<4, T, Q> packRGBM(vec<3, float, Q> const& v)
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<3, T, Q> unpackRGBM(vec<4, T, Q> const& rgbm);
+
+	/// Returns an unsigned integer vector obtained by converting the components of a floating-point vector
+	/// to the 16-bit floating-point representation found in the OpenGL Specification.
+	/// The first vector component specifies the 16 least-significant bits of the result;
+	/// the forth component specifies the 16 most-significant bits.
+	///
+	/// @see gtc_packing
+	/// @see vec<L, float, Q> unpackHalf(vec<L, uint16, Q> const& p)
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	template<length_t L, qualifier Q>
+	GLM_FUNC_DECL vec<L, uint16, Q> packHalf(vec<L, float, Q> const& v);
+
+	/// Returns a floating-point vector with components obtained by reinterpreting an integer vector as 16-bit floating-point numbers and converting them to 32-bit floating-point values.
+	/// The first component of the vector is obtained from the 16 least-significant bits of v;
+	/// the forth component is obtained from the 16 most-significant bits of v.
+	///
+	/// @see gtc_packing
+	/// @see vec<L, uint16, Q> packHalf(vec<L, float, Q> const& v)
+	/// @see <a href="http://www.opengl.org/registry/doc/GLSLangSpec.4.20.8.pdf">GLSL 4.20.8 specification, section 8.4 Floating-Point Pack and Unpack Functions</a>
+	template<length_t L, qualifier Q>
+	GLM_FUNC_DECL vec<L, float, Q> unpackHalf(vec<L, uint16, Q> const& p);
+
+	/// Convert each component of the normalized floating-point vector into unsigned integer values.
+	///
+	/// @see gtc_packing
+	/// @see vec<L, floatType, Q> unpackUnorm(vec<L, intType, Q> const& p);
+	template<typename uintType, length_t L, typename floatType, qualifier Q>
+	GLM_FUNC_DECL vec<L, uintType, Q> packUnorm(vec<L, floatType, Q> const& v);
+
+	/// Convert a packed integer to a normalized floating-point vector.
+	///
+	/// @see gtc_packing
+	/// @see vec<L, intType, Q> packUnorm(vec<L, floatType, Q> const& v)
+	template<typename floatType, length_t L, typename uintType, qualifier Q>
+	GLM_FUNC_DECL vec<L, floatType, Q> unpackUnorm(vec<L, uintType, Q> const& v);
+
+	/// Convert each component of the normalized floating-point vector into signed integer values.
+	///
+	/// @see gtc_packing
+	/// @see vec<L, floatType, Q> unpackSnorm(vec<L, intType, Q> const& p);
+	template<typename intType, length_t L, typename floatType, qualifier Q>
+	GLM_FUNC_DECL vec<L, intType, Q> packSnorm(vec<L, floatType, Q> const& v);
+
+	/// Convert a packed integer to a normalized floating-point vector.
+	///
+	/// @see gtc_packing
+	/// @see vec<L, intType, Q> packSnorm(vec<L, floatType, Q> const& v)
+	template<typename floatType, length_t L, typename intType, qualifier Q>
+	GLM_FUNC_DECL vec<L, floatType, Q> unpackSnorm(vec<L, intType, Q> const& v);
+
+	/// Convert each component of the normalized floating-point vector into unsigned integer values.
+	///
+	/// @see gtc_packing
+	/// @see vec2 unpackUnorm2x4(uint8 p)
+	GLM_FUNC_DECL uint8 packUnorm2x4(vec2 const& v);
+
+	/// Convert a packed integer to a normalized floating-point vector.
+	///
+	/// @see gtc_packing
+	/// @see uint8 packUnorm2x4(vec2 const& v)
+	GLM_FUNC_DECL vec2 unpackUnorm2x4(uint8 p);
+
+	/// Convert each component of the normalized floating-point vector into unsigned integer values.
+	///
+	/// @see gtc_packing
+	/// @see vec4 unpackUnorm4x4(uint16 p)
+	GLM_FUNC_DECL uint16 packUnorm4x4(vec4 const& v);
+
+	/// Convert a packed integer to a normalized floating-point vector.
+	///
+	/// @see gtc_packing
+	/// @see uint16 packUnorm4x4(vec4 const& v)
+	GLM_FUNC_DECL vec4 unpackUnorm4x4(uint16 p);
+
+	/// Convert each component of the normalized floating-point vector into unsigned integer values.
+	///
+	/// @see gtc_packing
+	/// @see vec3 unpackUnorm1x5_1x6_1x5(uint16 p)
+	GLM_FUNC_DECL uint16 packUnorm1x5_1x6_1x5(vec3 const& v);
+
+	/// Convert a packed integer to a normalized floating-point vector.
+	///
+	/// @see gtc_packing
+	/// @see uint16 packUnorm1x5_1x6_1x5(vec3 const& v)
+	GLM_FUNC_DECL vec3 unpackUnorm1x5_1x6_1x5(uint16 p);
+
+	/// Convert each component of the normalized floating-point vector into unsigned integer values.
+	///
+	/// @see gtc_packing
+	/// @see vec4 unpackUnorm3x5_1x1(uint16 p)
+	GLM_FUNC_DECL uint16 packUnorm3x5_1x1(vec4 const& v);
+
+	/// Convert a packed integer to a normalized floating-point vector.
+	///
+	/// @see gtc_packing
+	/// @see uint16 packUnorm3x5_1x1(vec4 const& v)
+	GLM_FUNC_DECL vec4 unpackUnorm3x5_1x1(uint16 p);
+
+	/// Convert each component of the normalized floating-point vector into unsigned integer values.
+	///
+	/// @see gtc_packing
+	/// @see vec3 unpackUnorm2x3_1x2(uint8 p)
+	GLM_FUNC_DECL uint8 packUnorm2x3_1x2(vec3 const& v);
+
+	/// Convert a packed integer to a normalized floating-point vector.
+	///
+	/// @see gtc_packing
+	/// @see uint8 packUnorm2x3_1x2(vec3 const& v)
+	GLM_FUNC_DECL vec3 unpackUnorm2x3_1x2(uint8 p);
+
+
+
+	/// Convert each component from an integer vector into a packed integer.
+	///
+	/// @see gtc_packing
+	/// @see i8vec2 unpackInt2x8(int16 p)
+	GLM_FUNC_DECL int16 packInt2x8(i8vec2 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see int16 packInt2x8(i8vec2 const& v)
+	GLM_FUNC_DECL i8vec2 unpackInt2x8(int16 p);
+
+	/// Convert each component from an integer vector into a packed unsigned integer.
+	///
+	/// @see gtc_packing
+	/// @see u8vec2 unpackInt2x8(uint16 p)
+	GLM_FUNC_DECL uint16 packUint2x8(u8vec2 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see uint16 packInt2x8(u8vec2 const& v)
+	GLM_FUNC_DECL u8vec2 unpackUint2x8(uint16 p);
+
+	/// Convert each component from an integer vector into a packed integer.
+	///
+	/// @see gtc_packing
+	/// @see i8vec4 unpackInt4x8(int32 p)
+	GLM_FUNC_DECL int32 packInt4x8(i8vec4 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see int32 packInt2x8(i8vec4 const& v)
+	GLM_FUNC_DECL i8vec4 unpackInt4x8(int32 p);
+
+	/// Convert each component from an integer vector into a packed unsigned integer.
+	///
+	/// @see gtc_packing
+	/// @see u8vec4 unpackUint4x8(uint32 p)
+	GLM_FUNC_DECL uint32 packUint4x8(u8vec4 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see uint32 packUint4x8(u8vec2 const& v)
+	GLM_FUNC_DECL u8vec4 unpackUint4x8(uint32 p);
+
+	/// Convert each component from an integer vector into a packed integer.
+	///
+	/// @see gtc_packing
+	/// @see i16vec2 unpackInt2x16(int p)
+	GLM_FUNC_DECL int packInt2x16(i16vec2 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see int packInt2x16(i16vec2 const& v)
+	GLM_FUNC_DECL i16vec2 unpackInt2x16(int p);
+
+	/// Convert each component from an integer vector into a packed integer.
+	///
+	/// @see gtc_packing
+	/// @see i16vec4 unpackInt4x16(int64 p)
+	GLM_FUNC_DECL int64 packInt4x16(i16vec4 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see int64 packInt4x16(i16vec4 const& v)
+	GLM_FUNC_DECL i16vec4 unpackInt4x16(int64 p);
+
+	/// Convert each component from an integer vector into a packed unsigned integer.
+	///
+	/// @see gtc_packing
+	/// @see u16vec2 unpackUint2x16(uint p)
+	GLM_FUNC_DECL uint packUint2x16(u16vec2 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see uint packUint2x16(u16vec2 const& v)
+	GLM_FUNC_DECL u16vec2 unpackUint2x16(uint p);
+
+	/// Convert each component from an integer vector into a packed unsigned integer.
+	///
+	/// @see gtc_packing
+	/// @see u16vec4 unpackUint4x16(uint64 p)
+	GLM_FUNC_DECL uint64 packUint4x16(u16vec4 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see uint64 packUint4x16(u16vec4 const& v)
+	GLM_FUNC_DECL u16vec4 unpackUint4x16(uint64 p);
+
+	/// Convert each component from an integer vector into a packed integer.
+	///
+	/// @see gtc_packing
+	/// @see i32vec2 unpackInt2x32(int p)
+	GLM_FUNC_DECL int64 packInt2x32(i32vec2 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see int packInt2x16(i32vec2 const& v)
+	GLM_FUNC_DECL i32vec2 unpackInt2x32(int64 p);
+
+	/// Convert each component from an integer vector into a packed unsigned integer.
+	///
+	/// @see gtc_packing
+	/// @see u32vec2 unpackUint2x32(int p)
+	GLM_FUNC_DECL uint64 packUint2x32(u32vec2 const& v);
+
+	/// Convert a packed integer into an integer vector.
+	///
+	/// @see gtc_packing
+	/// @see int packUint2x16(u32vec2 const& v)
+	GLM_FUNC_DECL u32vec2 unpackUint2x32(uint64 p);
+
+	/// @}
+}// namespace glm
+
+#include "packing.inl"

cuda_extensions/third_party/glm/glm/gtc/packing.inl

@@ -0,0 +1,938 @@
+/// @ref gtc_packing
+
+#include "../ext/scalar_relational.hpp"
+#include "../ext/vector_relational.hpp"
+#include "../common.hpp"
+#include "../vec2.hpp"
+#include "../vec3.hpp"
+#include "../vec4.hpp"
+#include "../detail/type_half.hpp"
+#include <cstring>
+#include <limits>
+
+namespace glm{
+namespace detail
+{
+	GLM_FUNC_QUALIFIER glm::uint16 float2half(glm::uint32 f)
+	{
+		// 10 bits    =>                         EE EEEFFFFF
+		// 11 bits    =>                        EEE EEFFFFFF
+		// Half bits  =>                   SEEEEEFF FFFFFFFF
+		// Float bits => SEEEEEEE EFFFFFFF FFFFFFFF FFFFFFFF
+
+		// 0x00007c00 => 00000000 00000000 01111100 00000000
+		// 0x000003ff => 00000000 00000000 00000011 11111111
+		// 0x38000000 => 00111000 00000000 00000000 00000000
+		// 0x7f800000 => 01111111 10000000 00000000 00000000
+		// 0x00008000 => 00000000 00000000 10000000 00000000
+		return
+			((f >> 16) & 0x8000) | // sign
+			((((f & 0x7f800000) - 0x38000000) >> 13) & 0x7c00) | // exponential
+			((f >> 13) & 0x03ff); // Mantissa
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint32 float2packed11(glm::uint32 f)
+	{
+		// 10 bits    =>                         EE EEEFFFFF
+		// 11 bits    =>                        EEE EEFFFFFF
+		// Half bits  =>                   SEEEEEFF FFFFFFFF
+		// Float bits => SEEEEEEE EFFFFFFF FFFFFFFF FFFFFFFF
+
+		// 0x000007c0 => 00000000 00000000 00000111 11000000
+		// 0x00007c00 => 00000000 00000000 01111100 00000000
+		// 0x000003ff => 00000000 00000000 00000011 11111111
+		// 0x38000000 => 00111000 00000000 00000000 00000000
+		// 0x7f800000 => 01111111 10000000 00000000 00000000
+		// 0x00008000 => 00000000 00000000 10000000 00000000
+		return
+			((((f & 0x7f800000) - 0x38000000) >> 17) & 0x07c0) | // exponential
+			((f >> 17) & 0x003f); // Mantissa
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint32 packed11ToFloat(glm::uint32 p)
+	{
+		// 10 bits    =>                         EE EEEFFFFF
+		// 11 bits    =>                        EEE EEFFFFFF
+		// Half bits  =>                   SEEEEEFF FFFFFFFF
+		// Float bits => SEEEEEEE EFFFFFFF FFFFFFFF FFFFFFFF
+
+		// 0x000007c0 => 00000000 00000000 00000111 11000000
+		// 0x00007c00 => 00000000 00000000 01111100 00000000
+		// 0x000003ff => 00000000 00000000 00000011 11111111
+		// 0x38000000 => 00111000 00000000 00000000 00000000
+		// 0x7f800000 => 01111111 10000000 00000000 00000000
+		// 0x00008000 => 00000000 00000000 10000000 00000000
+		return
+			((((p & 0x07c0) << 17) + 0x38000000) & 0x7f800000) | // exponential
+			((p & 0x003f) << 17); // Mantissa
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint32 float2packed10(glm::uint32 f)
+	{
+		// 10 bits    =>                         EE EEEFFFFF
+		// 11 bits    =>                        EEE EEFFFFFF
+		// Half bits  =>                   SEEEEEFF FFFFFFFF
+		// Float bits => SEEEEEEE EFFFFFFF FFFFFFFF FFFFFFFF
+
+		// 0x0000001F => 00000000 00000000 00000000 00011111
+		// 0x0000003F => 00000000 00000000 00000000 00111111
+		// 0x000003E0 => 00000000 00000000 00000011 11100000
+		// 0x000007C0 => 00000000 00000000 00000111 11000000
+		// 0x00007C00 => 00000000 00000000 01111100 00000000
+		// 0x000003FF => 00000000 00000000 00000011 11111111
+		// 0x38000000 => 00111000 00000000 00000000 00000000
+		// 0x7f800000 => 01111111 10000000 00000000 00000000
+		// 0x00008000 => 00000000 00000000 10000000 00000000
+		return
+			((((f & 0x7f800000) - 0x38000000) >> 18) & 0x03E0) | // exponential
+			((f >> 18) & 0x001f); // Mantissa
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint32 packed10ToFloat(glm::uint32 p)
+	{
+		// 10 bits    =>                         EE EEEFFFFF
+		// 11 bits    =>                        EEE EEFFFFFF
+		// Half bits  =>                   SEEEEEFF FFFFFFFF
+		// Float bits => SEEEEEEE EFFFFFFF FFFFFFFF FFFFFFFF
+
+		// 0x0000001F => 00000000 00000000 00000000 00011111
+		// 0x0000003F => 00000000 00000000 00000000 00111111
+		// 0x000003E0 => 00000000 00000000 00000011 11100000
+		// 0x000007C0 => 00000000 00000000 00000111 11000000
+		// 0x00007C00 => 00000000 00000000 01111100 00000000
+		// 0x000003FF => 00000000 00000000 00000011 11111111
+		// 0x38000000 => 00111000 00000000 00000000 00000000
+		// 0x7f800000 => 01111111 10000000 00000000 00000000
+		// 0x00008000 => 00000000 00000000 10000000 00000000
+		return
+			((((p & 0x03E0) << 18) + 0x38000000) & 0x7f800000) | // exponential
+			((p & 0x001f) << 18); // Mantissa
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint half2float(glm::uint h)
+	{
+		return ((h & 0x8000) << 16) | ((( h & 0x7c00) + 0x1C000) << 13) | ((h & 0x03FF) << 13);
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint floatTo11bit(float x)
+	{
+		if(x == 0.0f)
+			return 0u;
+		else if(glm::isnan(x))
+			return ~0u;
+		else if(glm::isinf(x))
+			return 0x1Fu << 6u;
+
+		uint Pack = 0u;
+		memcpy(&Pack, &x, sizeof(Pack));
+		return float2packed11(Pack);
+	}
+
+	GLM_FUNC_QUALIFIER float packed11bitToFloat(glm::uint x)
+	{
+		if(x == 0)
+			return 0.0f;
+		else if(x == ((1 << 11) - 1))
+			return ~0;//NaN
+		else if(x == (0x1f << 6))
+			return ~0;//Inf
+
+		uint Result = packed11ToFloat(x);
+
+		float Temp = 0;
+		memcpy(&Temp, &Result, sizeof(Temp));
+		return Temp;
+	}
+
+	GLM_FUNC_QUALIFIER glm::uint floatTo10bit(float x)
+	{
+		if(x == 0.0f)
+			return 0u;
+		else if(glm::isnan(x))
+			return ~0u;
+		else if(glm::isinf(x))
+			return 0x1Fu << 5u;
+
+		uint Pack = 0;
+		memcpy(&Pack, &x, sizeof(Pack));
+		return float2packed10(Pack);
+	}
+
+	GLM_FUNC_QUALIFIER float packed10bitToFloat(glm::uint x)
+	{
+		if(x == 0)
+			return 0.0f;
+		else if(x == ((1 << 10) - 1))
+			return ~0;//NaN
+		else if(x == (0x1f << 5))
+			return ~0;//Inf
+
+		uint Result = packed10ToFloat(x);
+
+		float Temp = 0;
+		memcpy(&Temp, &Result, sizeof(Temp));
+		return Temp;
+	}
+
+//	GLM_FUNC_QUALIFIER glm::uint f11_f11_f10(float x, float y, float z)
+//	{
+//		return ((floatTo11bit(x) & ((1 << 11) - 1)) << 0) |  ((floatTo11bit(y) & ((1 << 11) - 1)) << 11) | ((floatTo10bit(z) & ((1 << 10) - 1)) << 22);
+//	}
+
+	union u3u3u2
+	{
+		struct
+		{
+			uint x : 3;
+			uint y : 3;
+			uint z : 2;
+		} data;
+		uint8 pack;
+	};
+
+	union u4u4
+	{
+		struct
+		{
+			uint x : 4;
+			uint y : 4;
+		} data;
+		uint8 pack;
+	};
+
+	union u4u4u4u4
+	{
+		struct
+		{
+			uint x : 4;
+			uint y : 4;
+			uint z : 4;
+			uint w : 4;
+		} data;
+		uint16 pack;
+	};
+
+	union u5u6u5
+	{
+		struct
+		{
+			uint x : 5;
+			uint y : 6;
+			uint z : 5;
+		} data;
+		uint16 pack;
+	};
+
+	union u5u5u5u1
+	{
+		struct
+		{
+			uint x : 5;
+			uint y : 5;
+			uint z : 5;
+			uint w : 1;
+		} data;
+		uint16 pack;
+	};
+
+	union u10u10u10u2
+	{
+		struct
+		{
+			uint x : 10;
+			uint y : 10;
+			uint z : 10;
+			uint w : 2;
+		} data;
+		uint32 pack;
+	};
+
+	union i10i10i10i2
+	{
+		struct
+		{
+			int x : 10;
+			int y : 10;
+			int z : 10;
+			int w : 2;
+		} data;
+		uint32 pack;
+	};
+
+	union u9u9u9e5
+	{
+		struct
+		{
+			uint x : 9;
+			uint y : 9;
+			uint z : 9;
+			uint w : 5;
+		} data;
+		uint32 pack;
+	};
+
+	template<length_t L, qualifier Q>
+	struct compute_half
+	{};
+
+	template<qualifier Q>
+	struct compute_half<1, Q>
+	{
+		GLM_FUNC_QUALIFIER static vec<1, uint16, Q> pack(vec<1, float, Q> const& v)
+		{
+			int16 const Unpack(detail::toFloat16(v.x));
+			u16vec1 Packed;
+			memcpy(&Packed, &Unpack, sizeof(Packed));
+			return Packed;
+		}
+
+		GLM_FUNC_QUALIFIER static vec<1, float, Q> unpack(vec<1, uint16, Q> const& v)
+		{
+			i16vec1 Unpack;
+			memcpy(&Unpack, &v, sizeof(Unpack));
+			return vec<1, float, Q>(detail::toFloat32(v.x));
+		}
+	};
+
+	template<qualifier Q>
+	struct compute_half<2, Q>
+	{
+		GLM_FUNC_QUALIFIER static vec<2, uint16, Q> pack(vec<2, float, Q> const& v)
+		{
+			vec<2, int16, Q> const Unpack(detail::toFloat16(v.x), detail::toFloat16(v.y));
+			u16vec2 Packed;
+			memcpy(&Packed, &Unpack, sizeof(Packed));
+			return Packed;
+		}
+
+		GLM_FUNC_QUALIFIER static vec<2, float, Q> unpack(vec<2, uint16, Q> const& v)
+		{
+			i16vec2 Unpack;
+			memcpy(&Unpack, &v, sizeof(Unpack));
+			return vec<2, float, Q>(detail::toFloat32(v.x), detail::toFloat32(v.y));
+		}
+	};
+
+	template<qualifier Q>
+	struct compute_half<3, Q>
+	{
+		GLM_FUNC_QUALIFIER static vec<3, uint16, Q> pack(vec<3, float, Q> const& v)
+		{
+			vec<3, int16, Q> const Unpack(detail::toFloat16(v.x), detail::toFloat16(v.y), detail::toFloat16(v.z));
+			u16vec3 Packed;
+			memcpy(&Packed, &Unpack, sizeof(Packed));
+			return Packed;
+		}
+
+		GLM_FUNC_QUALIFIER static vec<3, float, Q> unpack(vec<3, uint16, Q> const& v)
+		{
+			i16vec3 Unpack;
+			memcpy(&Unpack, &v, sizeof(Unpack));
+			return vec<3, float, Q>(detail::toFloat32(v.x), detail::toFloat32(v.y), detail::toFloat32(v.z));
+		}
+	};
+
+	template<qualifier Q>
+	struct compute_half<4, Q>
+	{
+		GLM_FUNC_QUALIFIER static vec<4, uint16, Q> pack(vec<4, float, Q> const& v)
+		{
+			vec<4, int16, Q> const Unpack(detail::toFloat16(v.x), detail::toFloat16(v.y), detail::toFloat16(v.z), detail::toFloat16(v.w));
+			u16vec4 Packed;
+			memcpy(&Packed, &Unpack, sizeof(Packed));
+			return Packed;
+		}
+
+		GLM_FUNC_QUALIFIER static vec<4, float, Q> unpack(vec<4, uint16, Q> const& v)
+		{
+			i16vec4 Unpack;
+			memcpy(&Unpack, &v, sizeof(Unpack));
+			return vec<4, float, Q>(detail::toFloat32(v.x), detail::toFloat32(v.y), detail::toFloat32(v.z), detail::toFloat32(v.w));
+		}
+	};
+}//namespace detail
+
+	GLM_FUNC_QUALIFIER uint8 packUnorm1x8(float v)
+	{
+		return static_cast<uint8>(round(clamp(v, 0.0f, 1.0f) * 255.0f));
+	}
+
+	GLM_FUNC_QUALIFIER float unpackUnorm1x8(uint8 p)
+	{
+		float const Unpack(p);
+		return Unpack * static_cast<float>(0.0039215686274509803921568627451); // 1 / 255
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packUnorm2x8(vec2 const& v)
+	{
+		u8vec2 const Topack(round(clamp(v, 0.0f, 1.0f) * 255.0f));
+
+		uint16 Unpack = 0;
+		memcpy(&Unpack, &Topack, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER vec2 unpackUnorm2x8(uint16 p)
+	{
+		u8vec2 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return vec2(Unpack) * float(0.0039215686274509803921568627451); // 1 / 255
+	}
+
+	GLM_FUNC_QUALIFIER uint8 packSnorm1x8(float v)
+	{
+		int8 const Topack(static_cast<int8>(round(clamp(v ,-1.0f, 1.0f) * 127.0f)));
+		uint8 Packed = 0;
+		memcpy(&Packed, &Topack, sizeof(Packed));
+		return Packed;
+	}
+
+	GLM_FUNC_QUALIFIER float unpackSnorm1x8(uint8 p)
+	{
+		int8 Unpack = 0;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return clamp(
+			static_cast<float>(Unpack) * 0.00787401574803149606299212598425f, // 1.0f / 127.0f
+			-1.0f, 1.0f);
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packSnorm2x8(vec2 const& v)
+	{
+		i8vec2 const Topack(round(clamp(v, -1.0f, 1.0f) * 127.0f));
+		uint16 Packed = 0;
+		memcpy(&Packed, &Topack, sizeof(Packed));
+		return Packed;
+	}
+
+	GLM_FUNC_QUALIFIER vec2 unpackSnorm2x8(uint16 p)
+	{
+		i8vec2 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return clamp(
+			vec2(Unpack) * 0.00787401574803149606299212598425f, // 1.0f / 127.0f
+			-1.0f, 1.0f);
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packUnorm1x16(float s)
+	{
+		return static_cast<uint16>(round(clamp(s, 0.0f, 1.0f) * 65535.0f));
+	}
+
+	GLM_FUNC_QUALIFIER float unpackUnorm1x16(uint16 p)
+	{
+		float const Unpack(p);
+		return Unpack * 1.5259021896696421759365224689097e-5f; // 1.0 / 65535.0
+	}
+
+	GLM_FUNC_QUALIFIER uint64 packUnorm4x16(vec4 const& v)
+	{
+		u16vec4 const Topack(round(clamp(v , 0.0f, 1.0f) * 65535.0f));
+		uint64 Packed = 0;
+		memcpy(&Packed, &Topack, sizeof(Packed));
+		return Packed;
+	}
+
+	GLM_FUNC_QUALIFIER vec4 unpackUnorm4x16(uint64 p)
+	{
+		u16vec4 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return vec4(Unpack) * 1.5259021896696421759365224689097e-5f; // 1.0 / 65535.0
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packSnorm1x16(float v)
+	{
+		int16 const Topack = static_cast<int16>(round(clamp(v ,-1.0f, 1.0f) * 32767.0f));
+		uint16 Packed = 0;
+		memcpy(&Packed, &Topack, sizeof(Packed));
+		return Packed;
+	}
+
+	GLM_FUNC_QUALIFIER float unpackSnorm1x16(uint16 p)
+	{
+		int16 Unpack = 0;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return clamp(
+			static_cast<float>(Unpack) * 3.0518509475997192297128208258309e-5f, //1.0f / 32767.0f,
+			-1.0f, 1.0f);
+	}
+
+	GLM_FUNC_QUALIFIER uint64 packSnorm4x16(vec4 const& v)
+	{
+		i16vec4 const Topack(round(clamp(v ,-1.0f, 1.0f) * 32767.0f));
+		uint64 Packed = 0;
+		memcpy(&Packed, &Topack, sizeof(Packed));
+		return Packed;
+	}
+
+	GLM_FUNC_QUALIFIER vec4 unpackSnorm4x16(uint64 p)
+	{
+		i16vec4 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return clamp(
+			vec4(Unpack) * 3.0518509475997192297128208258309e-5f, //1.0f / 32767.0f,
+			-1.0f, 1.0f);
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packHalf1x16(float v)
+	{
+		int16 const Topack(detail::toFloat16(v));
+		uint16 Packed = 0;
+		memcpy(&Packed, &Topack, sizeof(Packed));
+		return Packed;
+	}
+
+	GLM_FUNC_QUALIFIER float unpackHalf1x16(uint16 v)
+	{
+		int16 Unpack = 0;
+		memcpy(&Unpack, &v, sizeof(Unpack));
+		return detail::toFloat32(Unpack);
+	}
+
+	GLM_FUNC_QUALIFIER uint64 packHalf4x16(glm::vec4 const& v)
+	{
+		i16vec4 const Unpack(
+			detail::toFloat16(v.x),
+			detail::toFloat16(v.y),
+			detail::toFloat16(v.z),
+			detail::toFloat16(v.w));
+		uint64 Packed = 0;
+		memcpy(&Packed, &Unpack, sizeof(Packed));
+		return Packed;
+	}
+
+	GLM_FUNC_QUALIFIER glm::vec4 unpackHalf4x16(uint64 v)
+	{
+		i16vec4 Unpack;
+		memcpy(&Unpack, &v, sizeof(Unpack));
+		return vec4(
+			detail::toFloat32(Unpack.x),
+			detail::toFloat32(Unpack.y),
+			detail::toFloat32(Unpack.z),
+			detail::toFloat32(Unpack.w));
+	}
+
+	GLM_FUNC_QUALIFIER uint32 packI3x10_1x2(ivec4 const& v)
+	{
+		detail::i10i10i10i2 Result;
+		Result.data.x = v.x;
+		Result.data.y = v.y;
+		Result.data.z = v.z;
+		Result.data.w = v.w;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER ivec4 unpackI3x10_1x2(uint32 v)
+	{
+		detail::i10i10i10i2 Unpack;
+		Unpack.pack = v;
+		return ivec4(
+			Unpack.data.x,
+			Unpack.data.y,
+			Unpack.data.z,
+			Unpack.data.w);
+	}
+
+	GLM_FUNC_QUALIFIER uint32 packU3x10_1x2(uvec4 const& v)
+	{
+		detail::u10u10u10u2 Result;
+		Result.data.x = v.x;
+		Result.data.y = v.y;
+		Result.data.z = v.z;
+		Result.data.w = v.w;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER uvec4 unpackU3x10_1x2(uint32 v)
+	{
+		detail::u10u10u10u2 Unpack;
+		Unpack.pack = v;
+		return uvec4(
+			Unpack.data.x,
+			Unpack.data.y,
+			Unpack.data.z,
+			Unpack.data.w);
+	}
+
+	GLM_FUNC_QUALIFIER uint32 packSnorm3x10_1x2(vec4 const& v)
+	{
+		ivec4 const Pack(round(clamp(v,-1.0f, 1.0f) * vec4(511.f, 511.f, 511.f, 1.f)));
+
+		detail::i10i10i10i2 Result;
+		Result.data.x = Pack.x;
+		Result.data.y = Pack.y;
+		Result.data.z = Pack.z;
+		Result.data.w = Pack.w;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER vec4 unpackSnorm3x10_1x2(uint32 v)
+	{
+		detail::i10i10i10i2 Unpack;
+		Unpack.pack = v;
+
+		vec4 const Result(Unpack.data.x, Unpack.data.y, Unpack.data.z, Unpack.data.w);
+
+		return clamp(Result * vec4(1.f / 511.f, 1.f / 511.f, 1.f / 511.f, 1.f), -1.0f, 1.0f);
+	}
+
+	GLM_FUNC_QUALIFIER uint32 packUnorm3x10_1x2(vec4 const& v)
+	{
+		uvec4 const Unpack(round(clamp(v, 0.0f, 1.0f) * vec4(1023.f, 1023.f, 1023.f, 3.f)));
+
+		detail::u10u10u10u2 Result;
+		Result.data.x = Unpack.x;
+		Result.data.y = Unpack.y;
+		Result.data.z = Unpack.z;
+		Result.data.w = Unpack.w;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER vec4 unpackUnorm3x10_1x2(uint32 v)
+	{
+		vec4 const ScaleFactors(1.0f / 1023.f, 1.0f / 1023.f, 1.0f / 1023.f, 1.0f / 3.f);
+
+		detail::u10u10u10u2 Unpack;
+		Unpack.pack = v;
+		return vec4(Unpack.data.x, Unpack.data.y, Unpack.data.z, Unpack.data.w) * ScaleFactors;
+	}
+
+	GLM_FUNC_QUALIFIER uint32 packF2x11_1x10(vec3 const& v)
+	{
+		return
+			((detail::floatTo11bit(v.x) & ((1 << 11) - 1)) <<  0) |
+			((detail::floatTo11bit(v.y) & ((1 << 11) - 1)) << 11) |
+			((detail::floatTo10bit(v.z) & ((1 << 10) - 1)) << 22);
+	}
+
+	GLM_FUNC_QUALIFIER vec3 unpackF2x11_1x10(uint32 v)
+	{
+		return vec3(
+			detail::packed11bitToFloat(v >> 0),
+			detail::packed11bitToFloat(v >> 11),
+			detail::packed10bitToFloat(v >> 22));
+	}
+
+	GLM_FUNC_QUALIFIER uint32 packF3x9_E1x5(vec3 const& v)
+	{
+		float const SharedExpMax = (pow(2.0f, 9.0f - 1.0f) / pow(2.0f, 9.0f)) * pow(2.0f, 31.f - 15.f);
+		vec3 const Color = clamp(v, 0.0f, SharedExpMax);
+		float const MaxColor = max(Color.x, max(Color.y, Color.z));
+
+		float const ExpSharedP = max(-15.f - 1.f, floor(log2(MaxColor))) + 1.0f + 15.f;
+		float const MaxShared = floor(MaxColor / pow(2.0f, (ExpSharedP - 15.f - 9.f)) + 0.5f);
+		float const ExpShared = equal(MaxShared, pow(2.0f, 9.0f), epsilon<float>()) ? ExpSharedP + 1.0f : ExpSharedP;
+
+		uvec3 const ColorComp(floor(Color / pow(2.f, (ExpShared - 15.f - 9.f)) + 0.5f));
+
+		detail::u9u9u9e5 Unpack;
+		Unpack.data.x = ColorComp.x;
+		Unpack.data.y = ColorComp.y;
+		Unpack.data.z = ColorComp.z;
+		Unpack.data.w = uint(ExpShared);
+		return Unpack.pack;
+	}
+
+	GLM_FUNC_QUALIFIER vec3 unpackF3x9_E1x5(uint32 v)
+	{
+		detail::u9u9u9e5 Unpack;
+		Unpack.pack = v;
+
+		return vec3(Unpack.data.x, Unpack.data.y, Unpack.data.z) * pow(2.0f, Unpack.data.w - 15.f - 9.f);
+	}
+
+	// Based on Brian Karis http://graphicrants.blogspot.fr/2009/04/rgbm-color-encoding.html
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<4, T, Q> packRGBM(vec<3, T, Q> const& rgb)
+	{
+		vec<3, T, Q> const Color(rgb * static_cast<T>(1.0 / 6.0));
+		T Alpha = clamp(max(max(Color.x, Color.y), max(Color.z, static_cast<T>(1e-6))), static_cast<T>(0), static_cast<T>(1));
+		Alpha = ceil(Alpha * static_cast<T>(255.0)) / static_cast<T>(255.0);
+		return vec<4, T, Q>(Color / Alpha, Alpha);
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<3, T, Q> unpackRGBM(vec<4, T, Q> const& rgbm)
+	{
+		return vec<3, T, Q>(rgbm.x, rgbm.y, rgbm.z) * rgbm.w * static_cast<T>(6);
+	}
+
+	template<length_t L, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, uint16, Q> packHalf(vec<L, float, Q> const& v)
+	{
+		return detail::compute_half<L, Q>::pack(v);
+	}
+
+	template<length_t L, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, float, Q> unpackHalf(vec<L, uint16, Q> const& v)
+	{
+		return detail::compute_half<L, Q>::unpack(v);
+	}
+
+	template<typename uintType, length_t L, typename floatType, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, uintType, Q> packUnorm(vec<L, floatType, Q> const& v)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<uintType>::is_integer, "uintType must be an integer type");
+		GLM_STATIC_ASSERT(std::numeric_limits<floatType>::is_iec559, "floatType must be a floating point type");
+
+		return vec<L, uintType, Q>(round(clamp(v, static_cast<floatType>(0), static_cast<floatType>(1)) * static_cast<floatType>(std::numeric_limits<uintType>::max())));
+	}
+
+	template<typename floatType, length_t L, typename uintType, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, floatType, Q> unpackUnorm(vec<L, uintType, Q> const& v)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<uintType>::is_integer, "uintType must be an integer type");
+		GLM_STATIC_ASSERT(std::numeric_limits<floatType>::is_iec559, "floatType must be a floating point type");
+
+		return vec<L, floatType, Q>(v) * (static_cast<floatType>(1) / static_cast<floatType>(std::numeric_limits<uintType>::max()));
+	}
+
+	template<typename intType, length_t L, typename floatType, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, intType, Q> packSnorm(vec<L, floatType, Q> const& v)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<intType>::is_integer, "uintType must be an integer type");
+		GLM_STATIC_ASSERT(std::numeric_limits<floatType>::is_iec559, "floatType must be a floating point type");
+
+		return vec<L, intType, Q>(round(clamp(v , static_cast<floatType>(-1), static_cast<floatType>(1)) * static_cast<floatType>(std::numeric_limits<intType>::max())));
+	}
+
+	template<typename floatType, length_t L, typename intType, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<L, floatType, Q> unpackSnorm(vec<L, intType, Q> const& v)
+	{
+		GLM_STATIC_ASSERT(std::numeric_limits<intType>::is_integer, "uintType must be an integer type");
+		GLM_STATIC_ASSERT(std::numeric_limits<floatType>::is_iec559, "floatType must be a floating point type");
+
+		return clamp(vec<L, floatType, Q>(v) * (static_cast<floatType>(1) / static_cast<floatType>(std::numeric_limits<intType>::max())), static_cast<floatType>(-1), static_cast<floatType>(1));
+	}
+
+	GLM_FUNC_QUALIFIER uint8 packUnorm2x4(vec2 const& v)
+	{
+		u32vec2 const Unpack(round(clamp(v, 0.0f, 1.0f) * 15.0f));
+		detail::u4u4 Result;
+		Result.data.x = Unpack.x;
+		Result.data.y = Unpack.y;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER vec2 unpackUnorm2x4(uint8 v)
+	{
+		float const ScaleFactor(1.f / 15.f);
+		detail::u4u4 Unpack;
+		Unpack.pack = v;
+		return vec2(Unpack.data.x, Unpack.data.y) * ScaleFactor;
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packUnorm4x4(vec4 const& v)
+	{
+		u32vec4 const Unpack(round(clamp(v, 0.0f, 1.0f) * 15.0f));
+		detail::u4u4u4u4 Result;
+		Result.data.x = Unpack.x;
+		Result.data.y = Unpack.y;
+		Result.data.z = Unpack.z;
+		Result.data.w = Unpack.w;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER vec4 unpackUnorm4x4(uint16 v)
+	{
+		float const ScaleFactor(1.f / 15.f);
+		detail::u4u4u4u4 Unpack;
+		Unpack.pack = v;
+		return vec4(Unpack.data.x, Unpack.data.y, Unpack.data.z, Unpack.data.w) * ScaleFactor;
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packUnorm1x5_1x6_1x5(vec3 const& v)
+	{
+		u32vec3 const Unpack(round(clamp(v, 0.0f, 1.0f) * vec3(31.f, 63.f, 31.f)));
+		detail::u5u6u5 Result;
+		Result.data.x = Unpack.x;
+		Result.data.y = Unpack.y;
+		Result.data.z = Unpack.z;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER vec3 unpackUnorm1x5_1x6_1x5(uint16 v)
+	{
+		vec3 const ScaleFactor(1.f / 31.f, 1.f / 63.f, 1.f / 31.f);
+		detail::u5u6u5 Unpack;
+		Unpack.pack = v;
+		return vec3(Unpack.data.x, Unpack.data.y, Unpack.data.z) * ScaleFactor;
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packUnorm3x5_1x1(vec4 const& v)
+	{
+		u32vec4 const Unpack(round(clamp(v, 0.0f, 1.0f) * vec4(31.f, 31.f, 31.f, 1.f)));
+		detail::u5u5u5u1 Result;
+		Result.data.x = Unpack.x;
+		Result.data.y = Unpack.y;
+		Result.data.z = Unpack.z;
+		Result.data.w = Unpack.w;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER vec4 unpackUnorm3x5_1x1(uint16 v)
+	{
+		vec4 const ScaleFactor(1.f / 31.f, 1.f / 31.f, 1.f / 31.f, 1.f);
+		detail::u5u5u5u1 Unpack;
+		Unpack.pack = v;
+		return vec4(Unpack.data.x, Unpack.data.y, Unpack.data.z, Unpack.data.w) * ScaleFactor;
+	}
+
+	GLM_FUNC_QUALIFIER uint8 packUnorm2x3_1x2(vec3 const& v)
+	{
+		u32vec3 const Unpack(round(clamp(v, 0.0f, 1.0f) * vec3(7.f, 7.f, 3.f)));
+		detail::u3u3u2 Result;
+		Result.data.x = Unpack.x;
+		Result.data.y = Unpack.y;
+		Result.data.z = Unpack.z;
+		return Result.pack;
+	}
+
+	GLM_FUNC_QUALIFIER vec3 unpackUnorm2x3_1x2(uint8 v)
+	{
+		vec3 const ScaleFactor(1.f / 7.f, 1.f / 7.f, 1.f / 3.f);
+		detail::u3u3u2 Unpack;
+		Unpack.pack = v;
+		return vec3(Unpack.data.x, Unpack.data.y, Unpack.data.z) * ScaleFactor;
+	}
+
+	GLM_FUNC_QUALIFIER int16 packInt2x8(i8vec2 const& v)
+	{
+		int16 Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER i8vec2 unpackInt2x8(int16 p)
+	{
+		i8vec2 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER uint16 packUint2x8(u8vec2 const& v)
+	{
+		uint16 Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER u8vec2 unpackUint2x8(uint16 p)
+	{
+		u8vec2 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER int32 packInt4x8(i8vec4 const& v)
+	{
+		int32 Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER i8vec4 unpackInt4x8(int32 p)
+	{
+		i8vec4 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER uint32 packUint4x8(u8vec4 const& v)
+	{
+		uint32 Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER u8vec4 unpackUint4x8(uint32 p)
+	{
+		u8vec4 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER int packInt2x16(i16vec2 const& v)
+	{
+		int Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER i16vec2 unpackInt2x16(int p)
+	{
+		i16vec2 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER int64 packInt4x16(i16vec4 const& v)
+	{
+		int64 Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER i16vec4 unpackInt4x16(int64 p)
+	{
+		i16vec4 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER uint packUint2x16(u16vec2 const& v)
+	{
+		uint Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER u16vec2 unpackUint2x16(uint p)
+	{
+		u16vec2 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER uint64 packUint4x16(u16vec4 const& v)
+	{
+		uint64 Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER u16vec4 unpackUint4x16(uint64 p)
+	{
+		u16vec4 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER int64 packInt2x32(i32vec2 const& v)
+	{
+		int64 Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER i32vec2 unpackInt2x32(int64 p)
+	{
+		i32vec2 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+
+	GLM_FUNC_QUALIFIER uint64 packUint2x32(u32vec2 const& v)
+	{
+		uint64 Pack = 0;
+		memcpy(&Pack, &v, sizeof(Pack));
+		return Pack;
+	}
+
+	GLM_FUNC_QUALIFIER u32vec2 unpackUint2x32(uint64 p)
+	{
+		u32vec2 Unpack;
+		memcpy(&Unpack, &p, sizeof(Unpack));
+		return Unpack;
+	}
+}//namespace glm
+

cuda_extensions/third_party/glm/glm/gtc/quaternion.hpp

@@ -0,0 +1,173 @@
+/// @ref gtc_quaternion
+/// @file glm/gtc/quaternion.hpp
+///
+/// @see core (dependence)
+/// @see gtc_constants (dependence)
+///
+/// @defgroup gtc_quaternion GLM_GTC_quaternion
+/// @ingroup gtc
+///
+/// Include <glm/gtc/quaternion.hpp> to use the features of this extension.
+///
+/// Defines a templated quaternion type and several quaternion operations.
+
+#pragma once
+
+// Dependency:
+#include "../gtc/constants.hpp"
+#include "../gtc/matrix_transform.hpp"
+#include "../ext/vector_relational.hpp"
+#include "../ext/quaternion_common.hpp"
+#include "../ext/quaternion_float.hpp"
+#include "../ext/quaternion_float_precision.hpp"
+#include "../ext/quaternion_double.hpp"
+#include "../ext/quaternion_double_precision.hpp"
+#include "../ext/quaternion_relational.hpp"
+#include "../ext/quaternion_geometric.hpp"
+#include "../ext/quaternion_trigonometric.hpp"
+#include "../ext/quaternion_transform.hpp"
+#include "../detail/type_mat3x3.hpp"
+#include "../detail/type_mat4x4.hpp"
+#include "../detail/type_vec3.hpp"
+#include "../detail/type_vec4.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_quaternion extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_quaternion
+	/// @{
+
+	/// Returns euler angles, pitch as x, yaw as y, roll as z.
+	/// The result is expressed in radians.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see gtc_quaternion
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL vec<3, T, Q> eulerAngles(qua<T, Q> const& x);
+
+	/// Returns roll value of euler angles expressed in radians.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see gtc_quaternion
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL T roll(qua<T, Q> const& x);
+
+	/// Returns pitch value of euler angles expressed in radians.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see gtc_quaternion
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL T pitch(qua<T, Q> const& x);
+
+	/// Returns yaw value of euler angles expressed in radians.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see gtc_quaternion
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL T yaw(qua<T, Q> const& x);
+
+	/// Converts a quaternion to a 3 * 3 matrix.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see gtc_quaternion
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL mat<3, 3, T, Q> mat3_cast(qua<T, Q> const& x);
+
+	/// Converts a quaternion to a 4 * 4 matrix.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see gtc_quaternion
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL mat<4, 4, T, Q> mat4_cast(qua<T, Q> const& x);
+
+	/// Converts a pure rotation 3 * 3 matrix to a quaternion.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see gtc_quaternion
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL qua<T, Q> quat_cast(mat<3, 3, T, Q> const& x);
+
+	/// Converts a pure rotation 4 * 4 matrix to a quaternion.
+	///
+	/// @tparam T Floating-point scalar types.
+	///
+	/// @see gtc_quaternion
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL qua<T, Q> quat_cast(mat<4, 4, T, Q> const& x);
+
+	/// Returns the component-wise comparison result of x < y.
+	///
+	/// @tparam T Floating-point scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_quaternion_relational
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL vec<4, bool, Q> lessThan(qua<T, Q> const& x, qua<T, Q> const& y);
+
+	/// Returns the component-wise comparison of result x <= y.
+	///
+	/// @tparam T Floating-point scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_quaternion_relational
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL vec<4, bool, Q> lessThanEqual(qua<T, Q> const& x, qua<T, Q> const& y);
+
+	/// Returns the component-wise comparison of result x > y.
+	///
+	/// @tparam T Floating-point scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_quaternion_relational
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL vec<4, bool, Q> greaterThan(qua<T, Q> const& x, qua<T, Q> const& y);
+
+	/// Returns the component-wise comparison of result x >= y.
+	///
+	/// @tparam T Floating-point scalar types
+	/// @tparam Q Value from qualifier enum
+	///
+	/// @see ext_quaternion_relational
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL vec<4, bool, Q> greaterThanEqual(qua<T, Q> const& x, qua<T, Q> const& y);
+
+	/// Build a look at quaternion based on the default handedness.
+	///
+	/// @param direction Desired forward direction. Needs to be normalized.
+	/// @param up Up vector, how the camera is oriented. Typically (0, 1, 0).
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL qua<T, Q> quatLookAt(
+		vec<3, T, Q> const& direction,
+		vec<3, T, Q> const& up);
+
+	/// Build a right-handed look at quaternion.
+	///
+	/// @param direction Desired forward direction onto which the -z-axis gets mapped. Needs to be normalized.
+	/// @param up Up vector, how the camera is oriented. Typically (0, 1, 0).
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL qua<T, Q> quatLookAtRH(
+		vec<3, T, Q> const& direction,
+		vec<3, T, Q> const& up);
+
+	/// Build a left-handed look at quaternion.
+	///
+	/// @param direction Desired forward direction onto which the +z-axis gets mapped. Needs to be normalized.
+	/// @param up Up vector, how the camera is oriented. Typically (0, 1, 0).
+	template<typename T, qualifier Q>
+	GLM_FUNC_DECL qua<T, Q> quatLookAtLH(
+		vec<3, T, Q> const& direction,
+		vec<3, T, Q> const& up);
+	/// @}
+} //namespace glm
+
+#include "quaternion.inl"

cuda_extensions/third_party/glm/glm/gtc/quaternion.inl

@@ -0,0 +1,208 @@
+#include "../trigonometric.hpp"
+#include "../geometric.hpp"
+#include "../exponential.hpp"
+#include "epsilon.hpp"
+#include <limits>
+
+namespace glm
+{
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<3, T, Q> eulerAngles(qua<T, Q> const& x)
+	{
+		return vec<3, T, Q>(pitch(x), yaw(x), roll(x));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T roll(qua<T, Q> const& q)
+	{
+		T const y = static_cast<T>(2) * (q.x * q.y + q.w * q.z);
+		T const x = q.w * q.w + q.x * q.x - q.y * q.y - q.z * q.z;
+
+		if(all(equal(vec<2, T, Q>(x, y), vec<2, T, Q>(0), epsilon<T>()))) //avoid atan2(0,0) - handle singularity - Matiis
+			return static_cast<T>(0);
+
+		return static_cast<T>(atan(y, x));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T pitch(qua<T, Q> const& q)
+	{
+		//return T(atan(T(2) * (q.y * q.z + q.w * q.x), q.w * q.w - q.x * q.x - q.y * q.y + q.z * q.z));
+		T const y = static_cast<T>(2) * (q.y * q.z + q.w * q.x);
+		T const x = q.w * q.w - q.x * q.x - q.y * q.y + q.z * q.z;
+
+		if(all(equal(vec<2, T, Q>(x, y), vec<2, T, Q>(0), epsilon<T>()))) //avoid atan2(0,0) - handle singularity - Matiis
+			return static_cast<T>(static_cast<T>(2) * atan(q.x, q.w));
+
+		return static_cast<T>(atan(y, x));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER T yaw(qua<T, Q> const& q)
+	{
+		return asin(clamp(static_cast<T>(-2) * (q.x * q.z - q.w * q.y), static_cast<T>(-1), static_cast<T>(1)));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER mat<3, 3, T, Q> mat3_cast(qua<T, Q> const& q)
+	{
+		mat<3, 3, T, Q> Result(T(1));
+		T qxx(q.x * q.x);
+		T qyy(q.y * q.y);
+		T qzz(q.z * q.z);
+		T qxz(q.x * q.z);
+		T qxy(q.x * q.y);
+		T qyz(q.y * q.z);
+		T qwx(q.w * q.x);
+		T qwy(q.w * q.y);
+		T qwz(q.w * q.z);
+
+		Result[0][0] = T(1) - T(2) * (qyy +  qzz);
+		Result[0][1] = T(2) * (qxy + qwz);
+		Result[0][2] = T(2) * (qxz - qwy);
+
+		Result[1][0] = T(2) * (qxy - qwz);
+		Result[1][1] = T(1) - T(2) * (qxx +  qzz);
+		Result[1][2] = T(2) * (qyz + qwx);
+
+		Result[2][0] = T(2) * (qxz + qwy);
+		Result[2][1] = T(2) * (qyz - qwx);
+		Result[2][2] = T(1) - T(2) * (qxx +  qyy);
+		return Result;
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER mat<4, 4, T, Q> mat4_cast(qua<T, Q> const& q)
+	{
+		return mat<4, 4, T, Q>(mat3_cast(q));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER qua<T, Q> quat_cast(mat<3, 3, T, Q> const& m)
+	{
+		T fourXSquaredMinus1 = m[0][0] - m[1][1] - m[2][2];
+		T fourYSquaredMinus1 = m[1][1] - m[0][0] - m[2][2];
+		T fourZSquaredMinus1 = m[2][2] - m[0][0] - m[1][1];
+		T fourWSquaredMinus1 = m[0][0] + m[1][1] + m[2][2];
+
+		int biggestIndex = 0;
+		T fourBiggestSquaredMinus1 = fourWSquaredMinus1;
+		if(fourXSquaredMinus1 > fourBiggestSquaredMinus1)
+		{
+			fourBiggestSquaredMinus1 = fourXSquaredMinus1;
+			biggestIndex = 1;
+		}
+		if(fourYSquaredMinus1 > fourBiggestSquaredMinus1)
+		{
+			fourBiggestSquaredMinus1 = fourYSquaredMinus1;
+			biggestIndex = 2;
+		}
+		if(fourZSquaredMinus1 > fourBiggestSquaredMinus1)
+		{
+			fourBiggestSquaredMinus1 = fourZSquaredMinus1;
+			biggestIndex = 3;
+		}
+
+		T biggestVal = sqrt(fourBiggestSquaredMinus1 + static_cast<T>(1)) * static_cast<T>(0.5);
+		T mult = static_cast<T>(0.25) / biggestVal;
+
+		switch(biggestIndex)
+		{
+		case 0:
+			return qua<T, Q>(biggestVal, (m[1][2] - m[2][1]) * mult, (m[2][0] - m[0][2]) * mult, (m[0][1] - m[1][0]) * mult);
+		case 1:
+			return qua<T, Q>((m[1][2] - m[2][1]) * mult, biggestVal, (m[0][1] + m[1][0]) * mult, (m[2][0] + m[0][2]) * mult);
+		case 2:
+			return qua<T, Q>((m[2][0] - m[0][2]) * mult, (m[0][1] + m[1][0]) * mult, biggestVal, (m[1][2] + m[2][1]) * mult);
+		case 3:
+			return qua<T, Q>((m[0][1] - m[1][0]) * mult, (m[2][0] + m[0][2]) * mult, (m[1][2] + m[2][1]) * mult, biggestVal);
+		default: // Silence a -Wswitch-default warning in GCC. Should never actually get here. Assert is just for sanity.
+			assert(false);
+			return qua<T, Q>(1, 0, 0, 0);
+		}
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER qua<T, Q> quat_cast(mat<4, 4, T, Q> const& m4)
+	{
+		return quat_cast(mat<3, 3, T, Q>(m4));
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<4, bool, Q> lessThan(qua<T, Q> const& x, qua<T, Q> const& y)
+	{
+		vec<4, bool, Q> Result;
+		for(length_t i = 0; i < x.length(); ++i)
+			Result[i] = x[i] < y[i];
+		return Result;
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<4, bool, Q> lessThanEqual(qua<T, Q> const& x, qua<T, Q> const& y)
+	{
+		vec<4, bool, Q> Result;
+		for(length_t i = 0; i < x.length(); ++i)
+			Result[i] = x[i] <= y[i];
+		return Result;
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<4, bool, Q> greaterThan(qua<T, Q> const& x, qua<T, Q> const& y)
+	{
+		vec<4, bool, Q> Result;
+		for(length_t i = 0; i < x.length(); ++i)
+			Result[i] = x[i] > y[i];
+		return Result;
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER vec<4, bool, Q> greaterThanEqual(qua<T, Q> const& x, qua<T, Q> const& y)
+	{
+		vec<4, bool, Q> Result;
+		for(length_t i = 0; i < x.length(); ++i)
+			Result[i] = x[i] >= y[i];
+		return Result;
+	}
+
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER qua<T, Q> quatLookAt(vec<3, T, Q> const& direction, vec<3, T, Q> const& up)
+	{
+#		if GLM_CONFIG_CLIP_CONTROL & GLM_CLIP_CONTROL_LH_BIT
+			return quatLookAtLH(direction, up);
+#		else
+			return quatLookAtRH(direction, up);
+# 		endif
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER qua<T, Q> quatLookAtRH(vec<3, T, Q> const& direction, vec<3, T, Q> const& up)
+	{
+		mat<3, 3, T, Q> Result;
+
+		Result[2] = -direction;
+		vec<3, T, Q> const& Right = cross(up, Result[2]);
+		Result[0] = Right * inversesqrt(max(static_cast<T>(0.00001), dot(Right, Right)));
+		Result[1] = cross(Result[2], Result[0]);
+
+		return quat_cast(Result);
+	}
+
+	template<typename T, qualifier Q>
+	GLM_FUNC_QUALIFIER qua<T, Q> quatLookAtLH(vec<3, T, Q> const& direction, vec<3, T, Q> const& up)
+	{
+		mat<3, 3, T, Q> Result;
+
+		Result[2] = direction;
+		vec<3, T, Q> const& Right = cross(up, Result[2]);
+		Result[0] = Right * inversesqrt(max(static_cast<T>(0.00001), dot(Right, Right)));
+		Result[1] = cross(Result[2], Result[0]);
+
+		return quat_cast(Result);
+	}
+}//namespace glm
+
+#if GLM_CONFIG_SIMD == GLM_ENABLE
+#	include "quaternion_simd.inl"
+#endif
+

cuda_extensions/third_party/glm/glm/gtc/random.hpp

@@ -0,0 +1,82 @@
+/// @ref gtc_random
+/// @file glm/gtc/random.hpp
+///
+/// @see core (dependence)
+/// @see gtx_random (extended)
+///
+/// @defgroup gtc_random GLM_GTC_random
+/// @ingroup gtc
+///
+/// Include <glm/gtc/random.hpp> to use the features of this extension.
+///
+/// Generate random number from various distribution methods.
+
+#pragma once
+
+// Dependency:
+#include "../ext/scalar_int_sized.hpp"
+#include "../ext/scalar_uint_sized.hpp"
+#include "../detail/qualifier.hpp"
+
+#if GLM_MESSAGES == GLM_ENABLE && !defined(GLM_EXT_INCLUDED)
+#	pragma message("GLM: GLM_GTC_random extension included")
+#endif
+
+namespace glm
+{
+	/// @addtogroup gtc_random
+	/// @{
+
+	/// Generate random numbers in the interval [Min, Max], according a linear distribution
+	///
+	/// @param Min Minimum value included in the sampling
+	/// @param Max Maximum value included in the sampling
+	/// @tparam genType Value type. Currently supported: float or double scalars.
+	/// @see gtc_random
+	template<typename genType>
+	GLM_FUNC_DECL genType linearRand(genType Min, genType Max);
+
+	/// Generate random numbers in the interval [Min, Max], according a linear distribution
+	///
+	/// @param Min Minimum value included in the sampling
+	/// @param Max Maximum value included in the sampling
+	/// @tparam T Value type. Currently supported: float or double.
+	///
+	/// @see gtc_random
+	template<length_t L, typename T, qualifier Q>
+	GLM_FUNC_DECL vec<L, T, Q> linearRand(vec<L, T, Q> const& Min, vec<L, T, Q> const& Max);
+
+	/// Generate random numbers in the interval [Min, Max], according a gaussian distribution
+	///
+	/// @see gtc_random
+	template<typename genType>
+	GLM_FUNC_DECL genType gaussRand(genType Mean, genType Deviation);
+
+	/// Generate a random 2D vector which coordinates are regularly distributed on a circle of a given radius
+	///
+	/// @see gtc_random
+	template<typename T>
+	GLM_FUNC_DECL vec<2, T, defaultp> circularRand(T Radius);
+
+	/// Generate a random 3D vector which coordinates are regularly distributed on a sphere of a given radius
+	///
+	/// @see gtc_random
+	template<typename T>
+	GLM_FUNC_DECL vec<3, T, defaultp> sphericalRand(T Radius);
+
+	/// Generate a random 2D vector which coordinates are regularly distributed within the area of a disk of a given radius
+	///
+	/// @see gtc_random
+	template<typename T>
+	GLM_FUNC_DECL vec<2, T, defaultp> diskRand(T Radius);
+
+	/// Generate a random 3D vector which coordinates are regularly distributed within the volume of a ball of a given radius
+	///
+	/// @see gtc_random
+	template<typename T>
+	GLM_FUNC_DECL vec<3, T, defaultp> ballRand(T Radius);
+
+	/// @}
+}//namespace glm
+
+#include "random.inl"