Merge branch 'master' into more_fft_utils

Author: Fletterio

3rdparty/CMakeLists.txt

@@ -180,6 +180,8 @@ option(WITH_JAVA "libjpeg-turbo" OFF)
 set(CMAKE_INSTALL_DOCDIR "${CMAKE_CURRENT_BINARY_DIR}/libjpeg-turbo/install/docdir")
 set(CMAKE_INSTALL_MANDIR "${CMAKE_CURRENT_BINARY_DIR}/libjpeg-turbo/install/mandir")
 add_subdirectory(libjpeg-turbo libjpeg-turbo EXCLUDE_FROM_ALL)
+NBL_TARGET_FORCE_ASSEMBLER_EXECUTABLE(simd ASM_NASM libjpeg-turbo/simd/)
+
 unset(CMAKE_INSTALL_DOCDIR)
 unset(CMAKE_INSTALL_MANDIR)
 

3rdparty/git-version-tracking

@@ -13,7 +13,11 @@ include(ExternalProject)
 include(${CMAKE_CURRENT_SOURCE_DIR}/cmake/toolchains/android/build.cmake)
 
 project(Nabla LANGUAGES CXX C)
-enable_language(C CXX ASM)
+enable_language(C CXX ASM ASM_NASM)
+
+if(MSVC)
+	enable_language(ASM_MASM)
+endif()
 
 option(NBL_STATIC_BUILD "" OFF) # ON for static builds, OFF for shared
 option(NBL_DYNAMIC_MSVC_RUNTIME "" ON)
@@ -30,7 +34,7 @@ endif()
 if(NBL_STATIC_BUILD)
 	message(STATUS "Static Nabla build enabled!")
 else()
-	if(WIN32 AND MSVC)
+	if(MSVC)
 		if(NBL_DYNAMIC_MSVC_RUNTIME)
 			message(STATUS "Shared Nabla build enabled!")
 		else()

CMakeLists.txt

@@ -5,7 +5,6 @@
 		"minor": 29,
 		"patch": 2
 	},
-
 	"configurePresets": [
 		{
 			"name": "ci-configure-base",
@@ -109,21 +108,27 @@
 			"inherits": "ci-configure-static-windows-base",
 			"displayName": "[CI]: Static library target, Ninja multi-config generator",
 			"description": "Configure as static library with Ninja multi-config generator",
-			"generator": "Ninja Multi-Config"
+			"generator": "Ninja Multi-Config",
+			"cacheVariables": {
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
+			}
 		},
 		{
 			"name": "ci-configure-dynamic-ninja-multi",
 			"inherits": "ci-configure-dynamic-windows-base",
 			"displayName": "[CI]: Dynamic library target, Ninja multi-config generator",
 			"description": "Configure as dynamic library with Ninja multi-config generator",
-			"generator": "Ninja Multi-Config"
+			"generator": "Ninja Multi-Config",
+			"cacheVariables": {
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
+			}
 		},
 		{
 			"name": "user-configure-base",
 			"hidden": true,
 			"cacheVariables": {
 				"NBL_CI_MODE": "OFF",
-				"NBL_UPDATE_GIT_SUBMODULE": "ON",
+				"NBL_UPDATE_GIT_SUBMODULE": "OFF",
 				"NBL_RUN_TESTS": "OFF",
 				"NBL_CPACK_CI": "OFF"
 			}
@@ -193,14 +198,20 @@
 			"inherits": "user-configure-static-windows-base",
 			"displayName": "[USER]: Static library target, Ninja multi-config generator",
 			"description": "Configure as static library with Ninja multi-config generator",
-			"generator": "Ninja Multi-Config"
+			"generator": "Ninja Multi-Config",
+			"cacheVariables": {
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
+			}
 		},
 		{
 			"name": "user-configure-dynamic-ninja-multi",
 			"inherits": "user-configure-dynamic-windows-base",
 			"displayName": "[USER]: Dynamic library target, Ninja multi-config generator",
 			"description": "Configure as dynamic library with Ninja multi-config generator",
-			"generator": "Ninja Multi-Config"
+			"generator": "Ninja Multi-Config",
+			"cacheVariables": {
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
+			}
 		},
 		{
 			"name": "user-configure-static-ninja-release",
@@ -209,7 +220,8 @@
 			"description": "Configure as static Release library with Ninja single-config generator",
 			"generator": "Ninja",
 			"cacheVariables": {
-				"CMAKE_BUILD_TYPE": "Release"
+				"CMAKE_BUILD_TYPE": "Release",
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
 			}
 		},
 		{
@@ -219,7 +231,8 @@
 			"description": "Configure as dynamic Release library with Ninja single-config generator",
 			"generator": "Ninja",
 			"cacheVariables": {
-				"CMAKE_BUILD_TYPE": "Release"
+				"CMAKE_BUILD_TYPE": "Release",
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
 			}
 		},
 		{
@@ -229,7 +242,8 @@
 			"description": "Configure as static RelWithDebInfo library with Ninja single-config generator",
 			"generator": "Ninja",
 			"cacheVariables": {
-				"CMAKE_BUILD_TYPE": "RelWithDebInfo"
+				"CMAKE_BUILD_TYPE": "RelWithDebInfo",
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
 			}
 		},
 		{
@@ -239,7 +253,8 @@
 			"description": "Configure as dynamic RelWithDebInfo library with Ninja single-config generator",
 			"generator": "Ninja",
 			"cacheVariables": {
-				"CMAKE_BUILD_TYPE": "RelWithDebInfo"
+				"CMAKE_BUILD_TYPE": "RelWithDebInfo",
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
 			}
 		},
 		{
@@ -249,7 +264,8 @@
 			"description": "Configure as static Debug library with Ninja single-config generator",
 			"generator": "Ninja",
 			"cacheVariables": {
-				"CMAKE_BUILD_TYPE": "Debug"
+				"CMAKE_BUILD_TYPE": "Debug",
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
 			}
 		},
 		{
@@ -259,7 +275,8 @@
 			"description": "Configure as dynamic Debug library with Ninja single-config generator",
 			"generator": "Ninja",
 			"cacheVariables": {
-				"CMAKE_BUILD_TYPE": "Debug"
+				"CMAKE_BUILD_TYPE": "Debug",
+				"CMAKE_EXPORT_COMPILE_COMMANDS": "ON"
 			}
 		}
 	],
@@ -413,7 +430,6 @@
 			"displayName": "[USER]: Dynamic Debug library target, Ninja single-config generator",
 			"description": "Build Nabla as dynamic Debug library with Ninja single-config generator"
 		}
-		
 	],
 	"packagePresets": [
 		{

CMakePresets.json

@@ -1321,4 +1321,15 @@ function(NBL_IMPORT_VS_CONFIG)
 		message(FATAL_ERORR "Cannot request importing VS config, doesn't meet requirements!")
 	endif()
 
-endfunction()
+endfunction()
+
+macro(NBL_TARGET_FORCE_ASSEMBLER_EXECUTABLE _NBL_TARGET_ _NBL_ASM_DIALECT_ _NBL_PREPEND_PATH_TRANSFORM_)
+	get_target_property(_NBL_TARGET_SOURCES_ "${_NBL_TARGET_}" SOURCES)
+	list(FILTER _NBL_TARGET_SOURCES_ INCLUDE REGEX "\\.asm$")
+	list(TRANSFORM _NBL_TARGET_SOURCES_ PREPEND "${_NBL_PREPEND_PATH_TRANSFORM_}")
+
+	set_source_files_properties(${_NBL_TARGET_SOURCES_}
+		TARGET_DIRECTORY "${_NBL_TARGET_}"
+		PROPERTIES LANGUAGE "${_NBL_ASM_DIALECT_}"
+	)
+endmacro()

cmake/common.cmake

@@ -46,18 +46,23 @@ You also have options `BUILD_EXAMPLES` and `BUILD_TOOLS` which do exactly what t
 
 For Windows *MSVC* required, *MinGW* build system maintenance will be delegated to the community.
 
-### Ninja Multi-Config generator
+### Ninja Multi-Config generator & MSVC
 
-The target toolset is MSVC. Before invoking CMake you need to [initialize VS's environment](https://learn.microsoft.com/en-us/cpp/build/cmake-presets-vs?view=msvc-170#sourcing-the-environment-when-building-with-command-line-generators-on-windows) by calling
+#### Command line
+
+Open command line, init MSVC [toolset environment](https://learn.microsoft.com/en-us/cpp/build/cmake-presets-vs?view=msvc-170#sourcing-the-environment-when-building-with-command-line-generators-on-windows) by calling
 
 ```cmd
 vcvarsall.bat <platform>
 ```
 
-in your cmd (eg. `vcvarsall.bat x64` call will initialize environment for amd64, for more info go [here](https://learn.microsoft.com/en-us/cpp/build/building-on-the-command-line?view=msvc-170#vcvarsall-syntax)). Once initialization is done you can use Ninja CMake preset to build Nabla, eg.
+(eg. `vcvarsall.bat x64` call will initialize environment for amd64, for more info see [following](https://learn.microsoft.com/en-us/cpp/build/building-on-the-command-line?view=msvc-170#vcvarsall-syntax)). Once initialized fire a ninja preset.
 
 ```cmd
+cmake --preset user-configure-dynamic-ninja-multi
 cmake --build --preset user-build-dynamic-ninja-multi --config Debug -- --quiet
 ```
 
-The build preset will pick configure preset associated with it and invoke configure then build. Pay attention that if you don't initialize VS's environment the following will fail to compile with MSVC toolset.
+#### With Visual Studio
+
+Follow VS [tutorial](https://learn.microsoft.com/en-us/cpp/build/cmake-projects-in-visual-studio?view=msvc-170).

docs/build/WINDOWS.md

@@ -0,0 +1,97 @@
+#ifndef _NBL_HLSL_FORMAT_OCTAHEDRAL_HLSL_
+#define _NBL_HLSL_FORMAT_OCTAHEDRAL_HLSL_
+
+#include "nbl/builtin/hlsl/cpp_compat.hlsl"
+#include "nbl/builtin/hlsl/type_traits.hlsl"
+#include "nbl/builtin/hlsl/limits.hlsl"
+
+namespace nbl
+{
+namespace hlsl
+{
+namespace format
+{
+
+template<typename UintT, uint16_t Bits=sizeof(UintT)*4>
+struct octahedral// : enable_if_t<Bits*2>sizeof(UintT)||Bits*4<sizeof(UintT)> need a way to static_assert in SPIRV!
+{
+    using this_t = octahedral<UintT,Bits>;
+    using storage_t = UintT;
+
+    NBL_CONSTEXPR_STATIC_INLINE uint16_t BitsUsed = Bits;
+
+    bool operator==(const this_t other)
+    {
+        return storage==other.storage;
+    }
+    bool operator!=(const this_t other)
+    {
+        return storage==other.storage;
+    }
+
+    storage_t storage;
+};
+
+}
+
+// https://www.shadertoy.com/view/Mtfyzl
+namespace impl
+{
+// TODO: remove after the `emulated_float` merge
+template<typename T, typename U>
+struct _static_cast_helper;
+
+// decode
+template<typename float_t, typename UintT, uint16_t Bits>
+struct _static_cast_helper<vector<float_t,3>,format::octahedral<UintT,Bits> >
+{
+    using T = vector<float_t,3>;
+    using U = format::octahedral<UintT,Bits>;
+
+    T operator()(U val)
+    {
+        using storage_t = typename U::storage_t;
+        const storage_t MaxVal = (storage_t(1)<<U::BitsUsed)-1u;
+    
+        // NOTE: We Assume the top unused bits are clean!
+        const vector<float_t,2> v = vector<float_t,2>(val.storage&MaxVal,val.storage>>U::BitsUsed) / (vector<float_t,2>(MaxVal,MaxVal)*0.5) - vector<float_t,2>(1,1);
+
+        // Rune Stubbe's version, much faster than original
+        vector<float_t,3> nor = vector<float_t,3>(v,float_t(1)-abs(v.x)-abs(v.y));
+        const float_t t = max(-nor.z,float_t(0));
+        // TODO: improve the copysign with `^` and a sign mask
+        nor.x += (nor.x>0.0) ? -t:t;
+        nor.y += (nor.y>0.0) ? -t:t;
+
+        return normalize(nor);
+    }
+};
+// encode
+template<typename UintT, uint16_t Bits, typename float_t>
+struct _static_cast_helper<format::octahedral<UintT,Bits>,vector<float_t,3> >
+{
+    using T = format::octahedral<UintT,Bits>;
+    using U = vector<float_t,3>;
+
+    T operator()(U nor)
+    {
+        nor /= (abs(nor.x) + abs(nor.y) + abs(nor.z));
+        if (nor.z<float_t(0)) // TODO: faster sign copy
+            nor.xy = (float_t(1)-abs(nor.yx))*sign(nor.xy);
+
+        vector<float_t,2> v = nor.xy*float_t(0.5)+vector<float_t,2>(0.5,0.5);
+
+        using storage_t = typename T::storage_t;
+        const storage_t MaxVal = (storage_t(1)<<T::BitsUsed)-1u;
+        const vector<storage_t,2> d = vector<storage_t,2>(v*float_t(MaxVal)+vector<float_t,2>(0.5,0.5));
+
+        T retval;
+        retval.storage = (d.y<<T::BitsUsed)|d.x;
+        return retval;
+    }
+};
+}
+
+}
+}
+#endif