From 2877f4eda3d1b0c7431039e3142ecf1a282a34b1 Mon Sep 17 00:00:00 2001 From: LaG1924 <12997935+LaG1924@users.noreply.github.com> Date: Tue, 21 Aug 2018 20:40:38 +0500 Subject: Update glm to 0.9.9.0 --- external/include/glm/detail/type_vec4_simd.inl | 381 ++++++++++++------------- 1 file changed, 179 insertions(+), 202 deletions(-) (limited to 'external/include/glm/detail/type_vec4_simd.inl') diff --git a/external/include/glm/detail/type_vec4_simd.inl b/external/include/glm/detail/type_vec4_simd.inl index 90652fd..adfb20c 100644 --- a/external/include/glm/detail/type_vec4_simd.inl +++ b/external/include/glm/detail/type_vec4_simd.inl @@ -7,14 +7,14 @@ namespace glm{ namespace detail { # if GLM_SWIZZLE == GLM_SWIZZLE_ENABLED - template - struct _swizzle_base1<4, float, P, glm::tvec4, E0,E1,E2,E3, true> : public _swizzle_base0 - { - GLM_FUNC_QUALIFIER tvec4 operator ()() const + template + struct _swizzle_base1<4, float, Q, E0,E1,E2,E3, true> : public _swizzle_base0 + { + GLM_FUNC_QUALIFIER vec<4, float, Q> operator ()() const { __m128 data = *reinterpret_cast<__m128 const*>(&this->_buffer); - tvec4 Result(uninitialize); + vec<4, float, Q> Result; # if GLM_ARCH & GLM_ARCH_AVX_BIT Result.data = _mm_permute_ps(data, _MM_SHUFFLE(E3, E2, E1, E0)); # else @@ -24,457 +24,434 @@ namespace detail } }; - template - struct _swizzle_base1<4, int32, P, glm::tvec4, E0,E1,E2,E3, true> : public _swizzle_base0 - { - GLM_FUNC_QUALIFIER tvec4 operator ()() const + template + struct _swizzle_base1<4, int32, Q, E0,E1,E2,E3, true> : public _swizzle_base0 + { + GLM_FUNC_QUALIFIER vec<4, int32, Q> operator ()() const { __m128i data = *reinterpret_cast<__m128i const*>(&this->_buffer); - tvec4 Result(uninitialize); + vec<4, int32, Q> Result; Result.data = _mm_shuffle_epi32(data, _MM_SHUFFLE(E3, E2, E1, E0)); return Result; } }; - template - struct _swizzle_base1<4, uint32, P, glm::tvec4, E0,E1,E2,E3, true> : public _swizzle_base0 - { - GLM_FUNC_QUALIFIER tvec4 operator ()() const + template + struct _swizzle_base1<4, uint32, Q, E0,E1,E2,E3, true> : public _swizzle_base0 + { + GLM_FUNC_QUALIFIER vec<4, uint32, Q> operator ()() const { __m128i data = *reinterpret_cast<__m128i const*>(&this->_buffer); - tvec4 Result(uninitialize); + vec<4, uint32, Q> Result; Result.data = _mm_shuffle_epi32(data, _MM_SHUFFLE(E3, E2, E1, E0)); return Result; } }; # endif// GLM_SWIZZLE == GLM_SWIZZLE_ENABLED - template - struct compute_vec4_add + template + struct compute_vec4_add { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, float, Q> call(vec<4, float, Q> const& a, vec<4, float, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, float, Q> Result; Result.data = _mm_add_ps(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX_BIT - template - struct compute_vec4_add + template + struct compute_vec4_add { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, double, Q> call(vec<4, double, Q> const& a, vec<4, double, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, double, Q> Result; Result.data = _mm256_add_pd(a.data, b.data); return Result; } }; # endif - template - struct compute_vec4_sub + template + struct compute_vec4_sub { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, float, Q> call(vec<4, float, Q> const& a, vec<4, float, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, float, Q> Result; Result.data = _mm_sub_ps(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX_BIT - template - struct compute_vec4_sub + template + struct compute_vec4_sub { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, double, Q> call(vec<4, double, Q> const& a, vec<4, double, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, double, Q> Result; Result.data = _mm256_sub_pd(a.data, b.data); return Result; } }; # endif - template - struct compute_vec4_mul + template + struct compute_vec4_mul { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, float, Q> call(vec<4, float, Q> const& a, vec<4, float, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, float, Q> Result; Result.data = _mm_mul_ps(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX_BIT - template - struct compute_vec4_mul + template + struct compute_vec4_mul { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, double, Q> call(vec<4, double, Q> const& a, vec<4, double, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, double, Q> Result; Result.data = _mm256_mul_pd(a.data, b.data); return Result; } }; # endif - template - struct compute_vec4_div + template + struct compute_vec4_div { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, float, Q> call(vec<4, float, Q> const& a, vec<4, float, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, float, Q> Result; Result.data = _mm_div_ps(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX_BIT - template - struct compute_vec4_div + template + struct compute_vec4_div { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, double, Q> call(vec<4, double, Q> const& a, vec<4, double, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, double, Q> Result; Result.data = _mm256_div_pd(a.data, b.data); return Result; } }; # endif - template <> + template<> struct compute_vec4_div { - static tvec4 call(tvec4 const & a, tvec4 const & b) + static vec<4, float, aligned_lowp> call(vec<4, float, aligned_lowp> const& a, vec<4, float, aligned_lowp> const& b) { - tvec4 Result(uninitialize); + vec<4, float, aligned_lowp> Result; Result.data = _mm_mul_ps(a.data, _mm_rcp_ps(b.data)); return Result; } }; - template - struct compute_vec4_and + template + struct compute_vec4_and { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm_and_si128(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX2_BIT - template - struct compute_vec4_and + template + struct compute_vec4_and { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm256_and_si256(a.data, b.data); return Result; } }; # endif - template - struct compute_vec4_or + template + struct compute_vec4_or { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm_or_si128(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX2_BIT - template - struct compute_vec4_or + template + struct compute_vec4_or { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm256_or_si256(a.data, b.data); return Result; } }; # endif - template - struct compute_vec4_xor + template + struct compute_vec4_xor { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm_xor_si128(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX2_BIT - template - struct compute_vec4_xor + template + struct compute_vec4_xor { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm256_xor_si256(a.data, b.data); return Result; } }; # endif - template - struct compute_vec4_shift_left + template + struct compute_vec4_shift_left { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm_sll_epi32(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX2_BIT - template - struct compute_vec4_shift_left + template + struct compute_vec4_shift_left { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm256_sll_epi64(a.data, b.data); return Result; } }; # endif - template - struct compute_vec4_shift_right + template + struct compute_vec4_shift_right { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm_srl_epi32(a.data, b.data); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX2_BIT - template - struct compute_vec4_shift_right + template + struct compute_vec4_shift_right { - static tvec4 call(tvec4 const& a, tvec4 const& b) + static vec<4, T, Q> call(vec<4, T, Q> const& a, vec<4, T, Q> const& b) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm256_srl_epi64(a.data, b.data); return Result; } }; # endif - template - struct compute_vec4_bitwise_not + template + struct compute_vec4_bitwise_not { - static tvec4 call(tvec4 const & v) + static vec<4, T, Q> call(vec<4, T, Q> const& v) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm_xor_si128(v.data, _mm_set1_epi32(-1)); return Result; } }; # if GLM_ARCH & GLM_ARCH_AVX2_BIT - template - struct compute_vec4_bitwise_not + template + struct compute_vec4_bitwise_not { - static tvec4 call(tvec4 const & v) + static vec<4, T, Q> call(vec<4, T, Q> const& v) { - tvec4 Result(uninitialize); + vec<4, T, Q> Result; Result.data = _mm256_xor_si256(v.data, _mm_set1_epi32(-1)); return Result; } }; # endif - template - struct compute_vec4_equal + template + struct compute_vec4_equal { - static bool call(tvec4 const & v1, tvec4 const & v2) + static bool call(vec<4, float, Q> const& v1, vec<4, float, Q> const& v2) { return _mm_movemask_ps(_mm_cmpeq_ps(v1.data, v2.data)) != 0; } }; - template - struct compute_vec4_equal + template + struct compute_vec4_equal { - static bool call(tvec4 const & v1, tvec4 const & v2) + static bool call(vec<4, int32, Q> const& v1, vec<4, int32, Q> const& v2) { return _mm_movemask_epi8(_mm_cmpeq_epi32(v1.data, v2.data)) != 0; } }; - template - struct compute_vec4_nequal + template + struct compute_vec4_nequal { - static bool call(tvec4 const & v1, tvec4 const & v2) + static bool call(vec<4, float, Q> const& v1, vec<4, float, Q> const& v2) { return _mm_movemask_ps(_mm_cmpneq_ps(v1.data, v2.data)) != 0; } }; - template - struct compute_vec4_nequal + template + struct compute_vec4_nequal { - static bool call(tvec4 const & v1, tvec4 const & v2) + static bool call(vec<4, int32, Q> const& v1, vec<4, int32, Q> const& v2) { return _mm_movemask_epi8(_mm_cmpneq_epi32(v1.data, v2.data)) != 0; } }; }//namespace detail -# if !GLM_HAS_DEFAULTED_FUNCTIONS - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4() -# ifndef GLM_FORCE_NO_CTOR_INIT - : data(_mm_setzero_ps()) -# endif - {} - - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4() -# ifndef GLM_FORCE_NO_CTOR_INIT - : data(_mm_setzero_ps()) -# endif - {} - - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4() -# ifndef GLM_FORCE_NO_CTOR_INIT - : data(_mm_setzero_ps()) -# endif - {} -# endif//!GLM_HAS_DEFAULTED_FUNCTIONS - - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(float s) : - data(_mm_set1_ps(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_lowp>::vec(float _s) : + data(_mm_set1_ps(_s)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(float s) : - data(_mm_set1_ps(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_mediump>::vec(float _s) : + data(_mm_set1_ps(_s)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(float s) : - data(_mm_set1_ps(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_highp>::vec(float _s) : + data(_mm_set1_ps(_s)) {} # if GLM_ARCH & GLM_ARCH_AVX_BIT - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(double s) : - data(_mm256_set1_pd(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, double, aligned_lowp>::vec(double _s) : + data(_mm256_set1_pd(_s)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(double s) : - data(_mm256_set1_pd(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, double, aligned_mediump>::vec(double _s) : + data(_mm256_set1_pd(_s)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(double s) : - data(_mm256_set1_pd(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, double, aligned_highp>::vec(double _s) : + data(_mm256_set1_pd(_s)) {} # endif - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 s) : - data(_mm_set1_epi32(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int32, aligned_lowp>::vec(int32 _s) : + data(_mm_set1_epi32(_s)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 s) : - data(_mm_set1_epi32(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int32, aligned_mediump>::vec(int32 _s) : + data(_mm_set1_epi32(_s)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 s) : - data(_mm_set1_epi32(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int32, aligned_highp>::vec(int32 _s) : + data(_mm_set1_epi32(_s)) {} # if GLM_ARCH & GLM_ARCH_AVX2_BIT - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int64 s) : - data(_mm256_set1_epi64x(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int64, aligned_lowp>::vec(int64 _s) : + data(_mm256_set1_epi64x(_s)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int64 s) : - data(_mm256_set1_epi64x(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int64, aligned_mediump>::vec(int64 _s) : + data(_mm256_set1_epi64x(_s)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int64 s) : - data(_mm256_set1_epi64x(s)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int64, aligned_highp>::vec(int64 _s) : + data(_mm256_set1_epi64x(_s)) {} # endif - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(float a, float b, float c, float d) : - data(_mm_set_ps(d, c, b, a)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_lowp>::vec(float _x, float _y, float _z, float _w) : + data(_mm_set_ps(_w, _z, _y, _x)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(float a, float b, float c, float d) : - data(_mm_set_ps(d, c, b, a)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_mediump>::vec(float _x, float _y, float _z, float _w) : + data(_mm_set_ps(_w, _z, _y, _x)) {} - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(float a, float b, float c, float d) : - data(_mm_set_ps(d, c, b, a)) + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_highp>::vec(float _x, float _y, float _z, float _w) : + data(_mm_set_ps(_w, _z, _y, _x)) {} - template <> - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 a, int32 b, int32 c, int32 d) : - data(_mm_set_epi32(d, c, b, a)) + template<> + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int32, aligned_lowp>::vec(int32 _x, int32 _y, int32 _z, int32 _w) : + data(_mm_set_epi32(_w, _z, _y, _x)) {} - template <> - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 a, int32 b, int32 c, int32 d) : - data(_mm_set_epi32(d, c, b, a)) + template<> + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int32, aligned_mediump>::vec(int32 _x, int32 _y, int32 _z, int32 _w) : + data(_mm_set_epi32(_w, _z, _y, _x)) {} - template <> - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 a, int32 b, int32 c, int32 d) : - data(_mm_set_epi32(d, c, b, a)) + template<> + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, int32, aligned_highp>::vec(int32 _x, int32 _y, int32 _z, int32 _w) : + data(_mm_set_epi32(_w, _z, _y, _x)) {} - template <> - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 a, int32 b, int32 c, int32 d) : - data(_mm_castsi128_ps(_mm_set_epi32(d, c, b, a))) + template<> + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_lowp>::vec(int32 _x, int32 _y, int32 _z, int32 _w) : + data(_mm_castsi128_ps(_mm_set_epi32(_w, _z, _y, _x))) {} - template <> - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 a, int32 b, int32 c, int32 d) : - data(_mm_castsi128_ps(_mm_set_epi32(d, c, b, a))) + template<> + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_mediump>::vec(int32 _x, int32 _y, int32 _z, int32 _w) : + data(_mm_castsi128_ps(_mm_set_epi32(_w, _z, _y, _x))) {} - template <> - template <> - GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD tvec4::tvec4(int32 a, int32 b, int32 c, int32 d) : - data(_mm_castsi128_ps(_mm_set_epi32(d, c, b, a))) + template<> + template<> + GLM_FUNC_QUALIFIER GLM_CONSTEXPR_SIMD vec<4, float, aligned_highp>::vec(int32 _x, int32 _y, int32 _z, int32 _w) : + data(_mm_castsi128_ps(_mm_set_epi32(_w, _z, _y, _x))) {} }//namespace glm -- cgit v1.2.3