Add quick simd implementation, needs a lot more work still.

include/vectorwrapper/vectorwrapper_simd.hpp

@@ -0,0 +1,55 @@
+/*
+ * Copyright 2015, 2016 Michele "King_DuckZ" Santullo
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef id2996A9D174804F7B88936D6C363D235B
+#define id2996A9D174804F7B88936D6C363D235B
+
+#include "vectorwrapper/vectorwrapper.hpp"
+#include <type_traits>
+#include <emmintrin.h>
+
+namespace vwr {
+	namespace simd {
+		template <typename V, std::size_t=::vwr::Vec<V>::dimensions, typename=typename ::vwr::Vec<V>::scalar_type>
+		class Vec;
+
+		template <typename V>
+		class Vec<V, 3, float> : public ::vwr::Vec<V, 3> {
+			typedef ::vwr::Vec<V, 3> base_class;
+		public:
+			static_assert(alignof(V) % 16 == 0, "Wrapped type must be aligned to 16");
+			static_assert(base_class::is_interleaved_mem == 0, "Expected tightly packed vector_type");
+
+			using typename base_class::vector_type;
+			typedef float scalar_type;
+
+			Vec ( void ) = default;
+			Vec ( const Vec& ) = default;
+			explicit Vec ( const vector_type& parIn ) : base_class(parIn) { }
+			explicit Vec ( const scalar_type parX ) : base_class(parX) { }
+			explicit Vec ( const base_class& parIn ) : base_class(parIn) { }
+
+			template <typename V2> Vec& operator+= ( const Vec<V2, 3, scalar_type>& parOther );
+			template <typename V2> Vec& operator-= ( const Vec<V2, 3, scalar_type>& parOther );
+			template <typename V2> Vec& operator*= ( const Vec<V2, 3, scalar_type>& parOther );
+			template <typename V2> Vec& operator/= ( const Vec<V2, 3, scalar_type>& parOther );
+		};
+	} //namespace simd
+} //namespace vwr
+
+#include "vectorwrapper/vectorwrapper_simd.inl"
+
+#endif

include/vectorwrapper/vectorwrapper_simd.inl

@@ -0,0 +1,47 @@
+/*
+ * Copyright 2015, 2016 Michele "King_DuckZ" Santullo
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+namespace vwr {
+	namespace simd {
+		template <typename V>
+		template <typename V2>
+		auto Vec<V, 3, float>::operator+= (const Vec<V2, 3, scalar_type>& parOther) -> Vec& {
+			__m128 pack_this;
+			__m128 pack_other;
+
+			pack_this = _mm_load_ps(&this->x());
+			pack_other = _mm_load_ps(&parOther.x());
+			pack_this = _mm_add_ps(pack_this, pack_other);
+			_mm_store_ps(&this->x(), pack_this);
+
+			return *this;
+		}
+
+		template <typename V>
+		template <typename V2>
+		auto Vec<V, 3, float>::operator-= (const Vec<V2, 3, scalar_type>& parOther) -> Vec& {
+			__m128 pack_this;
+			__m128 pack_other;
+
+			pack_this = _mm_load_ps(&this->x());
+			pack_other = _mm_load_ps(&parOther.x());
+			pack_this = _mm_sub_ps(pack_this, pack_other);
+			_mm_store_ps(&this->x(), pack_this);
+
+			return *this;
+		}
+	} //namespace simd
+} //namespace vwr

test/CMakeLists.txt

@@ -1,5 +1,7 @@
 cmake_minimum_required(VERSION 2.6.4 FATAL_ERROR)
 
+include(CTest)
+
 set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -g -O0 -std=c++11")
 set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -O3 -std=c++11")
 
@@ -12,3 +14,4 @@ include_directories(SYSTEM
 )
 include_directories(../include)
 add_subdirectory(unit)
+add_subdirectory(speed)

test/speed/CMakeLists.txt

@@ -0,0 +1,18 @@
+project(speed CXX)
+
+set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -mfpmath=sse -mmmx -msse -msse2")
+set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -mfpmath=sse -mmmx -msse -msse2")
+
+add_executable(${PROJECT_NAME}
+	${GTEST_MAIN_CPP}
+	speed.cpp
+)
+
+target_link_libraries(${PROJECT_NAME}
+	gtest
+)
+add_test(
+	NAME SpeedTest
+	WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
+	COMMAND ${PROJECT_NAME}
+)

test/speed/speed.cpp

@@ -0,0 +1,34 @@
+#include "vectorwrapper/vectorops.hpp"
+#include "vectorwrapper/vectorwrapper_simd.hpp"
+#include <gtest/gtest.h>
+#include <memory>
+
+namespace vwr {
+	typedef std::aligned_storage<sizeof(float) * 3, 16>::type float3_storage;
+
+	template <>
+	struct VectorWrapperInfo<float3_storage> {
+		enum { dimensions = 3 };
+		typedef float scalar_type;
+
+		enum {
+			offset_x = 0,
+			offset_y = offset_x + sizeof(scalar_type),
+			offset_z = offset_y + sizeof(scalar_type)
+		};
+	};
+} //namespace vwr
+
+TEST(vwr_speed, speed) {
+	typedef vwr::simd::Vec<vwr::float3_storage> simd_vec3;
+	static_assert(sizeof(simd_vec3) >= sizeof(float) * 3, "SIMD vector too small");
+
+	simd_vec3 v1(1.0f);
+	simd_vec3 v2(0.5f);
+
+	v1 += v2;
+
+	EXPECT_FLOAT_EQ(1.5f, v1.x());
+	EXPECT_FLOAT_EQ(1.5f, v1.y());
+	EXPECT_FLOAT_EQ(1.5f, v1.z());
+}

test/unit/CMakeLists.txt

@@ -10,3 +10,9 @@ add_executable(${PROJECT_NAME}
 target_link_libraries(${PROJECT_NAME}
 	gtest
 )
+
+add_test(
+	NAME UnitTest
+	WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
+	COMMAND ${PROJECT_NAME}
+)