Add python script to visualize results, updates on compilation

Author: bertaye

CMakeLists.txt

@@ -26,16 +26,18 @@ endif()
 project(BasicSIMD)
 
 set(CMAKE_CXX_STANDARD 11)
-set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /MT")
-
+if(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
+  set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /MT")
+endif()
 add_library(BasicSIMD INTERFACE)
 target_include_directories(BasicSIMD INTERFACE ${CMAKE_CURRENT_SOURCE_DIR})
 target_sources(BasicSIMD INTERFACE ${CMAKE_CURRENT_SOURCE_DIR}/SIMD.h)
 
-if(MSVC)
+if(CMAKE_CXX_COMPILER_ID STREQUAL "MSVC")
     target_compile_options(BasicSIMD INTERFACE /arch:AVX2)
-else()
-    target_compile_options(BasicSIMD INTERFACE -march=native -fno-strict-aliasing)
+elseif(CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
+    # Save assembly output with very detailed annotations
+    target_compile_options(BasicSIMD INTERFACE -march=native -fno-strict-aliasing -O3 -fno-tree-vectorize)
 endif()
 
 # Tests

SIMD.h

@@ -1,7 +1,7 @@
 #ifdef _MSC_VER
 #define _SIMD_INL_ __forceinline
-#else
-#define _SIMD_INL_ __attribute__((always_inline))
+#elif defined(__GNUC__) || defined(__clang__)
+#define _SIMD_INL_ __attribute__((always_inline)) inline
 #endif
 #include <immintrin.h>
 #include <type_traits>
@@ -1384,31 +1384,31 @@ class Array
     }
 
     //Add + and - operators
-    friend void operator+=(Array& lhs, const Array& rhs)
+    _SIMD_INL_ friend void operator+=(Array& lhs, const Array& rhs)
     {
         for (unsigned int i = 0; i < Length; i++)
         {
             T::AddInplaceRaw(lhs.Data + i*T::ElementCount, rhs.Data + i*T::ElementCount);
         }
     }
 
-    friend void operator-=(Array& lhs, const Array& rhs)
+    _SIMD_INL_ friend void operator-=(Array& lhs, const Array& rhs)
     {
         for (unsigned int i = 0; i < Length; i++)
         {
             T::SubtractInplaceRaw(lhs.Data + i*T::ElementCount, rhs.Data + i*T::ElementCount);            
         }
     }
 
-    friend void operator*=(Array& lhs, const Array& rhs)
+    _SIMD_INL_ friend void operator*=(Array& lhs, const Array& rhs)
     {
         for (unsigned int i = 0; i < Length; i++)
         {
             T::MultiplyInplaceRaw(lhs.Data + i*T::ElementCount, rhs.Data + i*T::ElementCount);
         }
     }
 
-    friend void operator/=(Array& lhs, const Array& rhs)
+    _SIMD_INL_ friend void operator/=(Array& lhs, const Array& rhs)
     {
         for (unsigned int i = 0; i < Length; i++)
         {