arm, neon: neon/fma support for glm_mul_rot()

* use glmm_set1() for each for now

.gitattributes

@@ -0,0 +1 @@
+*.h linguist-language=C

.gitignore

@@ -72,3 +72,7 @@ cglm-test-ios*
 /cglm.pc
 test-driver
 Default-568h@2x.png
+build/
+conftest.dir/*
+confdefs.h
+*.xcuserdatad

.travis.yml

@@ -4,6 +4,12 @@ os:
   - linux
   - osx
 
+arch:
+  - amd64
+  - ppc64le
+  - s390x
+  - arm64
+
 sudo: required
 dist: trusty
 

.vscode/settings.json

@@ -0,0 +1,4 @@
+{
+  "C_Cpp.default.configurationProvider": "vector-of-bool.cmake-tools",
+  "restructuredtext.confPath": "${workspaceFolder}/docs/source"
+}

CMakeLists.txt

@@ -0,0 +1,137 @@
+cmake_minimum_required(VERSION 3.8.2)
+project(cglm VERSION 0.8.2 LANGUAGES C)
+
+set(CMAKE_C_STANDARD 11)
+set(CMAKE_C_STANDARD_REQUIRED YES)
+set(DEFAULT_BUILD_TYPE "Release")
+
+set(CGLM_BUILD)
+option(CGLM_SHARED "Shared build" ON)
+option(CGLM_STATIC "Static build" OFF)
+option(CGLM_USE_C99 "" OFF)
+option(CGLM_USE_TEST "Enable Tests" OFF)
+
+if(NOT CGLM_STATIC AND CGLM_SHARED)
+  set(CGLM_BUILD SHARED)
+else(CGLM_STATIC)
+  set(CGLM_BUILD STATIC)
+endif()
+
+if(CGLM_USE_C99)
+  set(CMAKE_C_STANDARD 99)
+endif()
+
+if(MSVC)
+  add_definitions(-DNDEBUG -D_WINDOWS -D_USRDLL)
+  add_compile_options(/W3 /Ox /Gy /Oi /TC)
+  
+  # Ref: https://skia.googlesource.com/third_party/sdl/+/refs/heads/master/CMakeLists.txt#225
+  # Make sure /RTC1 is disabled, otherwise it will use functions from the CRT
+  foreach(flag_var
+      CMAKE_C_FLAGS CMAKE_C_FLAGS_DEBUG CMAKE_C_FLAGS_RELEASE
+      CMAKE_C_FLAGS_MINSIZEREL CMAKE_C_FLAGS_RELWITHDEBINFO)
+    string(REGEX REPLACE "/RTC(su|[1su])" "" ${flag_var} "${${flag_var}}")
+  endforeach(flag_var)
+else()
+  add_compile_options(-Wall -Werror -O3)
+endif()
+
+if(NOT CMAKE_BUILD_TYPE AND NOT CMAKE_CONFIGURATION_TYPES)
+  message(STATUS "Setting build type to '${DEFAULT_BUILD_TYPE}' as none was specified.")
+  set(CMAKE_BUILD_TYPE "${DEFAULT_BUILD_TYPE}" CACHE STRING "Choose the type of build." FORCE)
+  # Set the possible values of build type for cmake-gui
+  set_property(CACHE CMAKE_BUILD_TYPE PROPERTY STRINGS "Debug" "Release" "MinSizeRel" "RelWithDebInfo")
+endif()
+
+include(GNUInstallDirs)
+
+set(CPACK_PROJECT_NAME ${PROJECT_NAME})
+set(CPACK_PROJECT_VERSION ${PROJECT_VERSION})
+include(CPack)
+
+# Target Start
+add_library(${PROJECT_NAME}
+  ${CGLM_BUILD}
+  src/euler.c
+  src/affine.c
+  src/io.c
+  src/quat.c
+  src/cam.c
+  src/vec2.c
+  src/vec3.c
+  src/vec4.c
+  src/mat2.c
+  src/mat3.c
+  src/mat4.c
+  src/plane.c
+  src/frustum.c
+  src/box.c
+  src/project.c
+  src/sphere.c
+  src/ease.c
+  src/curve.c
+  src/bezier.c
+  src/ray.c
+  src/affine2d.c
+  )
+
+if(CGLM_SHARED)
+  add_definitions(-DCGLM_EXPORTS)
+else()
+  target_compile_definitions(${PROJECT_NAME} PUBLIC -DCGLM_STATIC)
+endif()
+
+set_target_properties(${PROJECT_NAME} PROPERTIES
+                              VERSION ${PROJECT_VERSION} 
+                            SOVERSION ${PROJECT_VERSION_MAJOR})
+
+target_include_directories(${PROJECT_NAME}
+    PUBLIC 
+        $<INSTALL_INTERFACE:include>    
+        $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
+    PRIVATE
+        ${CMAKE_CURRENT_SOURCE_DIR}/src
+)
+
+# Target for header-only usage
+add_library(${PROJECT_NAME}_headers INTERFACE)
+target_include_directories(${PROJECT_NAME}_headers INTERFACE
+  ${CMAKE_CURRENT_SOURCE_DIR}/include)
+
+# Test Configuration
+if(CGLM_USE_TEST)
+  include(CTest)
+  enable_testing()
+  add_subdirectory(test)
+endif()
+
+# Install 
+install(TARGETS ${PROJECT_NAME}
+        EXPORT  ${PROJECT_NAME}
+        LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        RUNTIME DESTINATION ${CMAKE_INSTALL_LIBDIR})
+
+install(DIRECTORY include/${PROJECT_NAME} DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
+        PATTERN ".*" EXCLUDE)
+
+# Config
+export(TARGETS ${PROJECT_NAME}
+       NAMESPACE ${PROJECT_NAME}::
+       FILE "${CMAKE_CURRENT_BINARY_DIR}/${PROJECT_NAME}Config.cmake"
+)
+
+install(EXPORT      ${PROJECT_NAME}
+        FILE        "${PROJECT_NAME}Config.cmake"
+        NAMESPACE   ${PROJECT_NAME}::
+        DESTINATION ${CMAKE_INSTALL_LIBDIR}/cmake/${PROJECT_NAME})
+
+set(PACKAGE_NAME ${PROJECT_NAME})
+set(prefix ${CMAKE_INSTALL_PREFIX})
+set(exec_prefix ${CMAKE_INSTALL_PREFIX})
+set(includedir "\${prefix}/${CMAKE_INSTALL_INCLUDEDIR}")
+set(libdir "\${prefix}/${CMAKE_INSTALL_LIBDIR}")
+configure_file(${CMAKE_CURRENT_LIST_DIR}/cglm.pc.in ${CMAKE_BINARY_DIR}/cglm.pc @ONLY)
+
+install(FILES ${CMAKE_BINARY_DIR}/cglm.pc
+  DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)

CREDITS

@@ -65,3 +65,16 @@ https://forums.khronos.org/showthread.php/10651-Animation-TCB-Spline-Interpolati
 12. vec2 cross product
 http://allenchou.net/2013/07/cross-product-of-2d-vectors/
 
+13. Ray triangle intersect
+Möller–Trumbore ray-triangle intersection algorithm, from "Fast, Minimum Storage Ray/Triangle Intersection"
+Authors:
+  Thomas Möller (tompa@clarus.se)
+  Ben Trumbore (wbt@graphics.cornell.edu)
+Link to paper: http://webserver2.tecgraf.puc-rio.br/~mgattass/cg/trbRR/Fast%20MinimumStorage%20RayTriangle%20Intersection.pdf
+
+14. ARM NEON: Matrix Vector Multiplication
+https://stackoverflow.com/a/57793352/2676533
+
+16. ARM NEON Div
+
+http://github.com/microsoft/DirectXMath

Makefile.am

@@ -12,7 +12,6 @@ AM_CFLAGS = -Wall \
             -O3 \
             -Wstrict-aliasing=2 \
             -fstrict-aliasing \
-            -pedantic \
             -Werror=strict-prototypes
 
 lib_LTLIBRARIES = libcglm.la
@@ -22,6 +21,9 @@ checkLDFLAGS = -L./.libs \
                -lm \
                -lcglm
 checkCFLAGS = $(AM_CFLAGS) \
+               -std=gnu11 \
+               -O3 \
+               -DCGLM_DEFINE_PRINTS \
                -I./include
 
 check_PROGRAMS = test/tests
@@ -63,7 +65,9 @@ cglm_HEADERS = include/cglm/version.h \
                include/cglm/ease.h \
                include/cglm/curve.h \
                include/cglm/bezier.h \
-               include/cglm/applesimd.h
+               include/cglm/applesimd.h \
+               include/cglm/ray.h \
+               include/cglm/affine2d.h
 
 cglm_calldir=$(includedir)/cglm/call
 cglm_call_HEADERS = include/cglm/call/mat4.h \
@@ -84,7 +88,9 @@ cglm_call_HEADERS = include/cglm/call/mat4.h \
                     include/cglm/call/sphere.h \
                     include/cglm/call/ease.h \
                     include/cglm/call/curve.h \
-                    include/cglm/call/bezier.h
+                    include/cglm/call/bezier.h \
+                    include/cglm/call/ray.h \
+                    include/cglm/call/affine2d.h
 
 cglm_simddir=$(includedir)/cglm/simd
 cglm_simd_HEADERS = include/cglm/simd/intrin.h \
@@ -95,6 +101,7 @@ cglm_simd_sse2dir=$(includedir)/cglm/simd/sse2
 cglm_simd_sse2_HEADERS = include/cglm/simd/sse2/affine.h \
                          include/cglm/simd/sse2/mat4.h \
                          include/cglm/simd/sse2/mat3.h \
+                         include/cglm/simd/sse2/mat2.h \
                          include/cglm/simd/sse2/quat.h
 
 cglm_simd_avxdir=$(includedir)/cglm/simd/avx
@@ -107,6 +114,7 @@ cglm_simd_neon_HEADERS = include/cglm/simd/neon/mat4.h
 cglm_structdir=$(includedir)/cglm/struct
 cglm_struct_HEADERS = include/cglm/struct/mat4.h \
                       include/cglm/struct/mat3.h \
+                      include/cglm/struct/mat2.h \
                       include/cglm/struct/vec2.h \
                       include/cglm/struct/vec2-ext.h \
                       include/cglm/struct/vec3.h \
@@ -124,7 +132,8 @@ cglm_struct_HEADERS = include/cglm/struct/mat4.h \
                       include/cglm/struct/project.h \
                       include/cglm/struct/sphere.h \
                       include/cglm/struct/color.h \
-                      include/cglm/struct/curve.h
+                      include/cglm/struct/curve.h \
+                      include/cglm/struct/affine2d.h
                       
 libcglm_la_SOURCES=\
     src/euler.c \
@@ -145,7 +154,9 @@ libcglm_la_SOURCES=\
     src/sphere.c \
     src/ease.c \
     src/curve.c \
-    src/bezier.c
+    src/bezier.c \
+    src/ray.c \
+    src/affine2d.c
 
 test_tests_SOURCES=\
     test/runner.c \

Package.swift

@@ -0,0 +1,44 @@
+// swift-tools-version:5.2
+
+import PackageDescription
+
+let package = Package(
+    name: "cglm",
+    products: [
+        .library(name: "cglm", type: .static, targets: ["cglmHeader"]),
+        .library(name: "cglmc", targets: ["cglmCompiled"]),
+    ],
+    dependencies: [],
+    targets: [
+        .target(
+            name: "cglmCompiled",
+            path: "./",
+            exclude: [
+                "./docs",
+                "./src/swift",
+                "./include",
+                "./test",
+                "./win",
+            ],
+            sources: [
+                "./src",
+            ],
+            publicHeadersPath: "./include"
+        ),
+        .target(
+            name: "cglmHeader",
+            path: "./",
+            exclude: [
+                "./docs",
+                "./include",
+                "./test",
+                "./win",
+            ],
+            sources: [
+                "./src/swift",
+            ],
+            publicHeadersPath: "./include"
+        ),
+    ],
+    cLanguageStandard: .c11
+)

README.md

@@ -25,6 +25,7 @@ you have the latest version
 - **[new option]** by starting v0.4.5, you can disable alignment requirement, check options in docs.  
 - **[major change]** by starting v0.5.0, vec3 functions use **glm_vec3_** namespace, it was **glm_vec_** until v0.5.0
 - **[major change]** by starting v0.5.1, built-in alignment is removed from **vec3** and **mat3** types
+- **[major change]** by starting v0.7.3, inline print functions are disabled in release/production mode to eliminate print costs (see options in documentation). Print output also improved. You can disable colors if you need  (see documentation)
 
 #### Note for C++ developers:
 If you are not aware of the original GLM library yet, you may also want to look at:
@@ -89,6 +90,7 @@ Currently *cglm* uses default clip space configuration (-1, 1) for camera functi
 - curves
 - curve interpolation helpers (S*M*C, deCasteljau...)
 - helpers to convert cglm types to Apple's simd library to pass cglm types to Metal GL without packing them on both sides
+- ray intersection helpers
 - and others...
 
 <hr />
@@ -148,6 +150,119 @@ The types used are actually unions that allow access to the same data multiple w
 
 ## Build
 
+### CMake (All platforms)
+```bash
+$ mkdir build
+$ cd build
+$ cmake .. # [Optional] -DCGLM_SHARED=ON
+$ make
+$ sudo make install # [Optional]
+```
+
+##### Cmake options with Defaults:
+
+```CMake
+option(CGLM_SHARED "Shared build" ON)
+option(CGLM_STATIC "Static build" OFF)
+option(CGLM_USE_C99 "" OFF) # C11 
+option(CGLM_USE_TEST "Enable Tests" OFF) # for make check - make test
+```
+
+#### Use as header-only library with your CMake project
+
+This requires no building or installation of cglm.
+
+* Example:
+
+``` cmake
+cmake_minimum_required(VERSION 3.8.2)
+
+project(<Your Project Name>)
+
+add_executable(${PROJECT_NAME} src/main.c)
+target_link_libraries(${LIBRARY_NAME} PRIVATE
+  cglm_headers)
+
+add_subdirectory(external/cglm/ EXCLUDE_FROM_ALL)
+```
+
+#### Use with your CMake project
+* Example:
+```cmake
+cmake_minimum_required(VERSION 3.8.2)
+
+project(<Your Project Name>)
+
+add_executable(${PROJECT_NAME} src/main.c)
+target_link_libraries(${LIBRARY_NAME} PRIVATE
+  cglm)
+
+add_subdirectory(external/cglm/)
+
+# or you can use find_package to configure cglm
+```
+
+### Meson (All platforms)
+
+```bash
+$ meson build # [Optional] --default-library=static
+$ cd build
+$ ninja
+$ sudo ninja install # [Optional]
+```
+
+##### Meson options with Defaults:
+
+```meson
+c_std=c11
+buildtype=release
+default_library=shared
+enable_tests=false # to run tests: ninja test
+```
+#### Use with your Meson project
+* Example:
+```meson
+# Clone cglm or create a cglm.wrap under <source_root>/subprojects
+project('name', 'c')
+
+cglm_dep = dependency('cglm', fallback : 'cglm', 'cglm_dep')
+
+executable('exe', 'src/main.c', dependencies : cglm_dep)
+```
+
+### Swift (Swift Package Manager)
+
+Currently only default build options are supported. Add **cglm** dependency to your project:
+
+```swift
+...
+Package( 
+  ...
+  dependencies: [
+    ...
+    .package(url: "https://github.com/recp/cglm", .branch("master")),
+  ]
+  ...
+)
+```
+
+Now add **cgml** as a dependency to your target. Product choices are:
+- **cglm** for inlined version of the library which can be linked only statically
+- **cglmc** for a compiled version of the library with no linking limitation
+
+```swift
+...
+.target(
+  ...
+  dependencies: [
+    ...
+    .product(name: "cglm", package: "cglm"),
+  ]
+  ...
+)
+...
+```
+
 ### Unix (Autotools)
 
 ```bash
@@ -291,7 +406,7 @@ You can pass matrices the same way to other APIs e.g. Vulkan, DX...
 ## Contributors
 
 This project exists thanks to all the people who contribute. [[Contribute](CONTRIBUTING.md)].
-<a href="graphs/contributors"><img src="https://opencollective.com/cglm/contributors.svg?width=890&button=false" /></a>
+<a href="https://github.com/recp/cglm/graphs/contributors"><img src="https://opencollective.com/cglm/contributors.svg?width=890&button=false" /></a>
 
 
 ## Backers

cglm.podspec

@@ -2,10 +2,10 @@ Pod::Spec.new do |s|
 
   # Description
   s.name         = "cglm"
-  s.version      = "0.6.1"
-  s.summary      = "📽 Optimized OpenGL/Graphics Math (glm) for C"
+  s.version      = "0.8.1"
+  s.summary      = "📽 Highly Optimized Graphics Math (glm) for C"
   s.description  = <<-DESC
-cglm is math library for graphics programming for C. It is similar to original glm but it is written for C instead of C++ (you can use here too). See the documentation or README for all features.
+cglm is math library for graphics programming for C. See the documentation or README for all features.
                    DESC
 
   s.documentation_url = "http://cglm.readthedocs.io"
@@ -25,4 +25,13 @@ cglm is math library for graphics programming for C. It is similar to original g
 
   # Linking
   s.library = "m"
+
+  # Configuration
+  s.pod_target_xcconfig = {
+    'CLANG_ENABLE_MODULES' => 'NO',
+    'CLANG_ALLOW_NON_MODULAR_INCLUDES_IN_FRAMEWORK_MODULES' => 'YES',
+    'CLANG_WARN_DOCUMENTATION_COMMENTS' => 'NO',
+    'GCC_C_LANGUAGE_STANDARD' => 'gnu11',
+    'GCC_PREPROCESSOR_DEFINITIONS' => '$(inherited) GLM_TESTS_NO_COLORFUL_OUTPUT'
+  }
 end

configure.ac

@@ -7,7 +7,7 @@
 #*****************************************************************************
 
 AC_PREREQ([2.69])
-AC_INIT([cglm], [0.7.0], [info@recp.me])
+AC_INIT([cglm], [0.8.2], [info@recp.me])
 AM_INIT_AUTOMAKE([-Wall -Werror foreign subdir-objects serial-tests])
 
 # Don't use the default cflags (-O2 -g), we set ours manually in Makefile.am.

docs/source/affine-mat.rst

@@ -1,6 +1,6 @@
 .. default-domain:: C
 
-affine transform matrix (specialized functions)
+3D Affine Transform Matrix (specialized functions)
 ================================================================================
 
 Header: cglm/affine-mat.h

docs/source/affine.rst

@@ -1,6 +1,6 @@
 .. default-domain:: C
 
-affine transforms
+3D Affine Transforms
 ================================================================================
 
 Header: cglm/affine.h
@@ -45,6 +45,8 @@ The implementation would be:
   glm_rotate(m, angle, axis);
   glm_translate(m, pivotInv); /* pivotInv = -pivot */
 
+.. _TransformsOrder:
+
 Transforms Order
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/affine2d.rst

@@ -0,0 +1,140 @@
+.. default-domain:: C
+
+2D Affine Transforms
+================================================================================
+
+Header: cglm/affine2d.h
+
+2D Transforms uses `2d` suffix for naming. If there is no 2D suffix it is 3D function.
+
+Initialize Transform Matrices
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+Functions with **_make** prefix expect you don't have a matrix and they create
+a matrix for you. You don't need to pass identity matrix.
+
+But other functions expect you have a matrix and you want to transform them. If
+you didn't have any existing matrix you have to initialize matrix to identity
+before sending to transfrom functions.
+
+Transforms Order
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+See :ref:`TransformsOrder` to read similar section.
+
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_translate2d`
+#. :c:func:`glm_translate2d_to`
+#. :c:func:`glm_translate2d_x`
+#. :c:func:`glm_translate2d_y`
+#. :c:func:`glm_translate2d_make`
+#. :c:func:`glm_scale2d_to`
+#. :c:func:`glm_scale2d_make`
+#. :c:func:`glm_scale2d`
+#. :c:func:`glm_scale2d_uni`
+#. :c:func:`glm_rotate2d_make`
+#. :c:func:`glm_rotate2d`
+#. :c:func:`glm_rotate2d_to`
+
+.. c:function:: void glm_translate2d(mat3 m, vec2 v)
+
+    translate existing 2d transform matrix by *v* vector and stores result in same matrix
+
+    Parameters:
+      | *[in, out]*  **m**    2d affine transfrom
+      | *[in]*       **v**    translate vector [x, y]
+
+.. c:function:: void glm_translate2d_to(mat3 m, vec2 v, mat3 dest)
+
+    translate existing 2d transform matrix by *v* vector and store result in dest
+
+    Parameters:
+      | *[in]*  **m**    2d affine transfrom
+      | *[in]*  **v**    translate vector [x, y]
+      | *[out]* **dest** translated matrix
+
+.. c:function:: void glm_translate2d_x(mat3 m, float x)
+
+    translate existing 2d transform matrix by x factor
+
+    Parameters:
+      | *[in, out]*  **m**    2d affine transfrom
+      | *[in]*       **x**    x factor
+
+.. c:function:: void glm_translate2d_y(mat3 m, float y)
+
+    translate existing 2d transform matrix by y factor
+
+    Parameters:
+      | *[in, out]*  **m**    2d affine transfrom
+      | *[in]*       **y**    y factor
+
+.. c:function:: void glm_translate2d_make(mat3 m, vec2 v)
+
+    creates NEW translate 2d transform matrix by *v* vector
+
+    Parameters:
+      | *[in, out]*  **m**    affine transfrom
+      | *[in]*       **v**    translate vector [x, y]
+
+.. c:function:: void glm_scale2d_to(mat3 m, vec2 v, mat3 dest)
+
+    scale existing 2d transform matrix by *v* vector and store result in dest
+
+    Parameters:
+      | *[in]*  **m**    affine transfrom
+      | *[in]*  **v**    scale vector [x, y]
+      | *[out]* **dest** scaled matrix
+
+.. c:function:: void glm_scale2d_make(mat3 m, vec2 v)
+
+    creates NEW 2d scale matrix by *v* vector
+
+    Parameters:
+      | *[in, out]*  **m**    affine transfrom
+      | *[in]*       **v**    scale vector [x, y]
+
+.. c:function:: void glm_scale2d(mat3 m, vec2 v)
+
+    scales existing 2d transform matrix by *v* vector and stores result in same matrix
+
+    Parameters:
+      | *[in, out]*  **m**    affine transfrom
+      | *[in]*       **v**    translate vector [x, y]
+
+.. c:function:: void glm_scale2d_uni(mat3 m, float s)
+
+    applies uniform scale to existing 2d transform matrix v = [s, s] and stores result in same matrix
+
+    Parameters:
+      | *[in, out]*  **m**  affine transfrom
+      | *[in]*       **s**  scale factor
+
+.. c:function:: void glm_rotate2d_make(mat3 m, float angle)
+
+    creates NEW rotation matrix by angle around *Z* axis
+
+    Parameters:
+      | *[in, out]*  **m**      affine transfrom
+      | *[in]*       **angle**  angle (radians)
+
+.. c:function:: void glm_rotate2d(mat3 m, float angle)
+
+    rotate existing 2d transform matrix around *Z* axis by angle and store result in same matrix
+
+    Parameters:
+      | *[in, out]*  **m**      affine transfrom
+      | *[in]*       **angle**  angle (radians)
+
+.. c:function:: void glm_rotate2d_to(mat3 m, float angle, mat3 dest)
+
+    rotate existing 2d transform matrix around *Z* axis by angle and store result in dest
+
+    Parameters:
+      | *[in]*  **m**      affine transfrom
+      | *[in]*  **angle**  angle (radians)
+      | *[out]* **dest**   rotated matrix

docs/source/api.rst

@@ -28,6 +28,7 @@ Follow the :doc:`build` documentation for this
 
    affine
    affine-mat
+   affine2d
    cam
    frustum
    box
@@ -51,3 +52,5 @@ Follow the :doc:`build` documentation for this
    sphere
    curve
    bezier
+   version
+   ray

docs/source/build.rst

@@ -7,6 +7,97 @@ Build cglm
 If you only need to inline versions, you don't need to build **cglm**, you don't need to link it to your program.
 Just import cglm to your project as dependency / external lib by copy-paste then use it as usual
 
+CMake (All platforms):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. code-block:: bash
+  :linenos:
+
+  $ mkdir build
+  $ cd build
+  $ cmake .. # [Optional] -DCGLM_SHARED=ON
+  $ make
+  $ sudo make install # [Optional]
+
+**make** will build cglm to **build** folder.
+If you don't want to install **cglm** to your system's folder you can get static and dynamic libs in this folder.
+
+**CMake Options:**
+
+.. code-block:: CMake
+  :linenos:
+
+  option(CGLM_SHARED "Shared build" ON)
+  option(CGLM_STATIC "Static build" OFF)
+  option(CGLM_USE_C99 "" OFF) # C11 
+  option(CGLM_USE_TEST "Enable Tests" OFF) # for make check - make test
+
+**Use as header-only library with your CMake project example**
+This requires no building or installation of cglm.
+
+.. code-block:: CMake
+  :linenos:
+
+  cmake_minimum_required(VERSION 3.8.2)
+  
+  project(<Your Project Name>)
+  
+  add_executable(${PROJECT_NAME} src/main.c)
+  target_link_libraries(${LIBRARY_NAME} PRIVATE
+    cglm_headers)
+  
+  add_subdirectory(external/cglm/ EXCLUDE_FROM_ALL)
+
+**Use with your CMake project example**
+
+.. code-block:: CMake
+  :linenos:
+
+  cmake_minimum_required(VERSION 3.8.2)
+  
+  project(<Your Project Name>)
+  
+  add_executable(${PROJECT_NAME} src/main.c)
+  target_link_libraries(${LIBRARY_NAME} PRIVATE
+    cglm)
+  
+  add_subdirectory(external/cglm/)
+
+Meson (All platforms):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. code-block::
+  :linenos:
+
+  $ meson build # [Optional] --default-library=static
+  $ cd build
+  $ ninja
+  $ sudo ninja install # [Optional]
+
+**Meson Options:**
+
+.. code-block:: 
+  :linenos:
+
+  c_std=c11
+  buildtype=release
+  default_library=shared
+  enable_tests=false # to run tests: ninja test
+
+
+**Use with your Meson project**
+
+.. code-block::
+  :linenos:
+
+  # Clone cglm or create a cglm.wrap under <source_root>/subprojects
+  project('name', 'c')
+  
+  cglm_dep = dependency('cglm', fallback : 'cglm', 'cglm_dep')
+  
+  executable('exe', 'src/main.c', dependencies : cglm_dep)
+
+
 Unix (Autotools):
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/cam.rst

@@ -198,7 +198,7 @@ Functions documentation
 
     Parameters:
       | *[in]*  **eye**     eye vector
-      | *[in]*  **center**  direction vector
+      | *[in]*  **dir**     direction vector
       | *[in]*  **up**      up vector
       | *[out]* **dest**    result matrix
 
@@ -212,7 +212,7 @@ Functions documentation
 
     Parameters:
       | *[in]*  **eye**     eye vector
-      | *[in]*  **center**  direction vector
+      | *[in]*  **dir**     direction vector
       | *[out]* **dest**    result matrix
 
 .. c:function:: void  glm_persp_decomp(mat4 proj, float *nearVal, float *farVal, float *top, float *bottom, float *left, float *right)

docs/source/conf.py

@@ -25,7 +25,7 @@
 
 # If your documentation needs a minimal Sphinx version, state it here.
 #
-# needs_sphinx = '1.0'
+# needs_sphinx = '3.0'
 
 # Add any Sphinx extension module names here, as strings. They can be
 # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom
@@ -62,9 +62,9 @@ author = u'Recep Aslantas'
 # built documents.
 #
 # The short X.Y version.
-version = u'0.7.0'
+version = u'0.8.2'
 # The full version, including alpha/beta/rc tags.
-release = u'0.7.0'
+release = u'0.8.2'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.
@@ -197,3 +197,7 @@ epub_exclude_files = ['search.html']
 
 # If true, `todo` and `todoList` produce output, else they produce nothing.
 todo_include_todos = True
+
+# -- Options for the C domain ------------------------------------------------
+
+c_id_attributes = ['__restrict']

docs/source/features.rst

@@ -1,10 +1,11 @@
 Features
 ================================================================================
 
+* array api and struct api, you can use arrays or structs.
 * general purpose matrix operations (mat4, mat3)
 * chain matrix multiplication (square only)
 * general purpose vector operations (cross, dot, rotate, proj, angle...)
-* affine transforms
+* affine transformations
 * matrix decomposition (extract rotation, scaling factor)
 * optimized affine transform matrices (mul, rigid-body inverse)
 * camera (lookat)
@@ -20,4 +21,6 @@ Features
 * easing functions
 * curves
 * curve interpolation helpers (SMC, deCasteljau...)
-* and other...
+* helpers to convert cglm types to Apple's simd library to pass cglm types to Metal GL without packing them on both sides
+* ray intersection helpers
+* and others...

docs/source/io.rst

@@ -28,6 +28,23 @@ Example to print mat4 matrix:
 (you probably will in some cases), you can change it temporary.
 cglm may provide precision parameter in the future
 
+Changes since **v0.7.3**:
+* Now mis-alignment of columns are fixed: larger numbers are printed via %g and others are printed via %f. Column withs are calculated before print.
+* Now values are colorful ;)
+* Some print improvements
+* New options with default values:
+
+.. code-block:: c
+
+    #define CGLM_PRINT_PRECISION    5
+    #define CGLM_PRINT_MAX_TO_SHORT 1e5
+    #define CGLM_PRINT_COLOR        "\033[36m"
+    #define CGLM_PRINT_COLOR_RESET  "\033[0m"
+
+* Inline prints are only enabled in DEBUG mode and if **CGLM_DEFINE_PRINTS** is defined.
+
+Check options page.
+
 Table of contents (click to go):
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/mat4.rst

@@ -156,16 +156,22 @@ Functions documentation
     Parameters:
       | *[in]*  **m**     mat4 (left)
       | *[in]*  **v**     vec4 (right, column vector)
+      | *[in]*  **last**  4th item to make it vec4
       | *[out]* **dest**  vec4 (result, column vector)
 
-.. c:function:: void  glm_mat4_mulv3(mat4 m, vec3 v, vec3 dest)
+.. c:function:: void  glm_mat4_mulv3(mat4 m, vec3 v, float last, vec3 dest)
 
-    multiply vector with mat4's mat3 part(rotation)
+    | multiply **vec3** with **mat4** and get **vec3** as result
+    |
+    | actually the result is **vec4**, after multiplication, 
+      the last component is trimmed, if you need the result's last component 
+      then don't use this function and consider to use **glm_mat4_mulv()**
 
     Parameters:
-    | *[in]*  **m**     mat4 (left)
-    | *[in]*  **v**     vec3 (right, column vector)
-    | *[out]* **dest**  vec3 (result, column vector)
+      | *[in]*  **m**     mat4(affine transform)
+      | *[in]*  **v**     vec3
+      | *[in]*  **last**  4th item to make it vec4
+      | *[out]* **dest**  result vector (vec3)
 
 .. c:function:: void  glm_mat4_trace(mat4 m)
 

docs/source/opt.rst

@@ -18,7 +18,7 @@ versor: 16 byte
 By starting **v0.4.5** cglm provides an option to disable alignment requirement.
 To enable this option define **CGLM_ALL_UNALIGNED** macro before all headers.
 You can define it in Xcode, Visual Studio (or other IDEs) or you can also prefer
-to define it in build system. If you use pre-compiled verisons then you
+to define it in build system. If you use pre-compiled versions then you
 have to compile cglm with **CGLM_ALL_UNALIGNED** macro.
 
 **VERY VERY IMPORTANT:** If you use cglm in multiple projects and
@@ -50,3 +50,36 @@ You have to extra options for dot product: **CGLM_SSE4_DOT** and **CGLM_SSE3_DOT
 - If **SSE3** is enabled then you can define **CGLM_SSE3_DOT** to force cglm to use **_mm_hadd_ps** instructions.
 
 otherwise cglm will use custom cglm's hadd functions which are optimized too.
+
+Print Options
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+1. **CGLM_DEFINE_PRINTS**
+2. **CGLM_NO_PRINTS_NOOP** (use CGLM_DEFINE_PRINTS)
+
+Inline prints are only enabled in **DEBUG** mode or if **CGLM_DEFINE_PRINTS** is defined.
+**glmc_** versions will always print too.
+
+Because **cglm** tried to enable print functions in debug mode and disable them in
+release/production mode to eliminate printing costs when we do not need them.
+
+**cglm** checks **DEBUG** or **_DEBUG** macros to test debug mode, if these are not working for you then you can use 
+**CGLM_DEFINE_PRINTS** to force enable, or create a PR to introduce new macro to test against debugging mode.
+
+If DEBUG mode is not enabled then print functions will be emptied to eliminate print function calls.
+You can disable this feature too by defining **CGLM_DEFINE_PRINTS** macro top of cglm header 
+or in project/build settings...
+
+3. **CGLM_PRINT_PRECISION**    5
+
+precision.
+
+4. **CGLM_PRINT_MAX_TO_SHORT** 1e5
+
+if a number is greater than this value then %g will be used, since this is shorten print you won't be able to see high precision.
+
+5. **CGLM_PRINT_COLOR**        "\033[36m"
+6. **CGLM_PRINT_COLOR_RESET**  "\033[0m"
+
+You can disable colorful print output by defining **CGLM_PRINT_COLOR** and **CGLM_PRINT_COLOR_RESET** as empty macro.
+Because some terminals may not support colors.

docs/source/quat.rst

@@ -374,7 +374,7 @@ Functions documentation
       | *[in]*       **q**      quaternion
       | *[in]*       **pivot**  pivot
 
-.. c:function:: void glm_quat_rotate(mat4 m, versor q, mat4 dest)
+.. c:function:: void glm_quat_rotate_atm(mat4 m, versor q, vec3 pivot)
 
     | rotate NEW transform matrix using quaternion at pivot point
     | this creates rotation matrix, it assumes you don't have a matrix

docs/source/ray.rst

@@ -0,0 +1,31 @@
+.. default-domain:: C
+
+ray
+====
+
+Header: cglm/ray.h
+
+This is for collision-checks used by ray-tracers and the like.
+
+Table of contents (click to go):
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Functions:
+
+1. :c:func:`glm_ray_triangle`
+
+Functions documentation
+~~~~~~~~~~~~~~~~~~~~~~~
+
+.. c:function:: bool  glm_ray_triangle(vec3 origin, vec3 direction, vec3 v0, vec3 v1, vec3 v2, float *d)
+
+    Möller–Trumbore ray-triangle intersection algorithm
+
+    Parameters:
+      | *[in]*       **origin**        origin of ray
+      | *[in]*       **direction**     direction of ray
+      | *[in]*       **v0**            first vertex of triangle
+      | *[in]*       **v1**            second vertex of triangle
+      | *[in]*       **v2**            third vertex of triangle
+      | *[in, out]*  **d**             float pointer to save distance to intersection
+      | *[out]*      **intersection**  whether there is intersection

docs/source/troubleshooting.rst

@@ -80,6 +80,19 @@ So be carefull, when your IDE (Xcode, Visual Studio ...) tried to autocomplete f
 
 **Also implementation may be wrong please let us know by creating an issue on Github.**
 
+BAD_ACCESS : Thread 1: EXC_BAD_ACCESS (code=EXC_I386_GPFLT) or Similar Errors/Crashes
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+This is similar issue with alignment. For instance if you compiled **cglm** with 
+AVX (**-mavx**, intentionally or not) and if you use **cglm** in an environment that doesn't 
+support AVX (or if AVX is disabled intentionally) e.g. environment that max support SSE2/3/4, 
+then you probably get **BAD ACCESS** or similar...
+
+Because if you compile **cglm** with AVX it aligns **mat4** with 32 byte boundary, 
+and your project aligns that as 16 byte boundary...
+
+Check alignment, supported vector extension or simd in **cglm** and linked projects...
+
 Other Issues?
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 

docs/source/util.rst

@@ -23,6 +23,7 @@ Functions:
 #. :c:func:`glm_max`
 #. :c:func:`glm_clamp`
 #. :c:func:`glm_lerp`
+#. :c:func:`glm_swapf`
 
 Functions documentation
 ~~~~~~~~~~~~~~~~~~~~~~~
@@ -146,7 +147,7 @@ Functions documentation
       | *[in]*  **b**   b
 
     Returns:
-       true if a and b equals
+       true if a and b are equal
 
 .. c:function:: float glm_percent(float from, float to, float current)
 
@@ -158,7 +159,7 @@ Functions documentation
       | *[in]*  **current**   value between from and to values
 
     Returns:
-       clamped normalized percent (0-100 in 0-1)
+       percentage of current value
 
 .. c:function:: float glm_percentc(float from, float to, float current)
 
@@ -171,3 +172,11 @@ Functions documentation
 
     Returns:
        clamped normalized percent (0-100 in 0-1)
+
+.. c:function:: void glm_swapf(float *a, float *b) 
+
+    swap two float values
+
+    Parameters:
+      | *[in]*  **a**      float 1
+      | *[in]*  **b**      float 2

docs/source/vec2.rst

@@ -55,12 +55,12 @@ Functions:
 Functions documentation
 ~~~~~~~~~~~~~~~~~~~~~~~
 
-.. c:function:: void glm_vec2(vec4 v4, vec2 dest)
+.. c:function:: void glm_vec2(float * v, vec2 dest)
 
-    init vec2 using vec3
+    init vec2 using vec3 or vec4
 
     Parameters:
-      | *[in]*  **v3**    vector3
+      | *[in]*  **v**     vector
       | *[out]* **dest**  destination
 
 .. c:function:: void glm_vec2_copy(vec2 a, vec2 dest)

docs/source/vec3.rst

@@ -392,7 +392,7 @@ Functions documentation
     Parameters:
       | *[in, out]*  **v**      vector
       | *[in]*       **axis**   axis vector (will be normalized)
-      | *[out]*      **angle**  angle (radians)
+      | *[in]*       **angle**  angle (radians)
 
 .. c:function:: void  glm_vec3_rotate_m4(mat4 m, vec3 v, vec3 dest)
 
@@ -435,8 +435,8 @@ Functions documentation
     squared distance between two vectors
 
     Parameters:
-      | *[in]*  **mat**   vector1
-      | *[in]*  **row1**  vector2
+      | *[in]*  **v1**  vector1
+      | *[in]*  **v2**  vector2
 
     Returns:
       | squared distance (distance * distance)
@@ -446,8 +446,8 @@ Functions documentation
     distance between two vectors
 
     Parameters:
-      | *[in]*  **mat**   vector1
-      | *[in]*  **row1**  vector2
+      | *[in]*  **v1**  vector1
+      | *[in]*  **v2**  vector2
 
     Returns:
       | distance
@@ -475,7 +475,7 @@ Functions documentation
     possible orthogonal/perpendicular vector
 
     Parameters:
-      | *[in]*  **mat**   vector
+      | *[in]*  **v**     vector
       | *[out]* **dest**  orthogonal/perpendicular vector
 
 .. c:function:: void  glm_vec3_clamp(vec3 v, float minVal, float maxVal)

docs/source/version.rst

@@ -0,0 +1,15 @@
+.. default-domain:: C
+
+version
+================================================================================
+
+Header: cglm/version.h
+
+**cglm** uses semantic versioning (http://semver.org) which is MAJOR.MINOR.PATCH 
+
+| **CGLM_VERSION_MAJOR** is major number of the version.
+| **CGLM_VERSION_MINOR** is minor number of the version.
+| **CGLM_VERSION_PATCH** is patch number of the version.
+
+every release increases these numbers. You can check existing version by 
+including `cglm/version.h`

include/cglm/affine-mat.h

@@ -26,6 +26,10 @@
 #  include "simd/avx/affine.h"
 #endif
 
+#ifdef CGLM_NEON_FP
+#  include "simd/neon/affine.h"
+#endif
+
 /*!
  * @brief this is similar to glm_mat4_mul but specialized to affine transform
  *
@@ -49,6 +53,8 @@ glm_mul(mat4 m1, mat4 m2, mat4 dest) {
   glm_mul_avx(m1, m2, dest);
 #elif defined( __SSE__ ) || defined( __SSE2__ )
   glm_mul_sse2(m1, m2, dest);
+#elif defined(CGLM_NEON_FP)
+  glm_mul_neon(m1, m2, dest);
 #else
   float a00 = m1[0][0], a01 = m1[0][1], a02 = m1[0][2], a03 = m1[0][3],
         a10 = m1[1][0], a11 = m1[1][1], a12 = m1[1][2], a13 = m1[1][3],
@@ -103,6 +109,8 @@ void
 glm_mul_rot(mat4 m1, mat4 m2, mat4 dest) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mul_rot_sse2(m1, m2, dest);
+#elif defined(CGLM_NEON_FP)
+  glm_mul_rot_neon(m1, m2, dest);
 #else
   float a00 = m1[0][0], a01 = m1[0][1], a02 = m1[0][2], a03 = m1[0][3],
         a10 = m1[1][0], a11 = m1[1][1], a12 = m1[1][2], a13 = m1[1][3],

include/cglm/affine.h

@@ -40,10 +40,6 @@
 #include "mat4.h"
 #include "affine-mat.h"
 
-CGLM_INLINE
-void
-glm_mat4_mul(mat4 m1, mat4 m2, mat4 dest);
-
 /*!
  * @brief translate existing transform matrix by v vector
  *        and stores result in same matrix
@@ -54,26 +50,22 @@ glm_mat4_mul(mat4 m1, mat4 m2, mat4 dest);
 CGLM_INLINE
 void
 glm_translate(mat4 m, vec3 v) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
+#if defined(CGLM_SIMD)
+  glmm_128 m0, m1, m2, m3;
+
+  m0 = glmm_load(m[0]);
+  m1 = glmm_load(m[1]);
+  m2 = glmm_load(m[2]);
+  m3 = glmm_load(m[3]);
+
   glmm_store(m[3],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_load(m[0]),
-                                              _mm_set1_ps(v[0])),
-                                   _mm_mul_ps(glmm_load(m[1]),
-                                              _mm_set1_ps(v[1]))),
-                        _mm_add_ps(_mm_mul_ps(glmm_load(m[2]),
-                                              _mm_set1_ps(v[2])),
-                                   glmm_load(m[3]))))
-  ;
+             glmm_fmadd(m0, glmm_set1(v[0]),
+                        glmm_fmadd(m1, glmm_set1(v[1]),
+                                   glmm_fmadd(m2, glmm_set1(v[2]), m3))));
 #else
-  vec4 v1, v2, v3;
-
-  glm_vec4_scale(m[0], v[0], v1);
-  glm_vec4_scale(m[1], v[1], v2);
-  glm_vec4_scale(m[2], v[2], v3);
-
-  glm_vec4_add(v1, m[3], m[3]);
-  glm_vec4_add(v2, m[3], m[3]);
-  glm_vec4_add(v3, m[3], m[3]);
+  glm_vec4_muladds(m[0], v[0], m[3]);
+  glm_vec4_muladds(m[1], v[1], m[3]);
+  glm_vec4_muladds(m[2], v[2], m[3]);
 #endif
 }
 
@@ -103,12 +95,8 @@ glm_translate_to(mat4 m, vec3 v, mat4 dest) {
 CGLM_INLINE
 void
 glm_translate_x(mat4 m, float x) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
-  glmm_store(m[3],
-             _mm_add_ps(_mm_mul_ps(glmm_load(m[0]),
-                                   _mm_set1_ps(x)),
-                        glmm_load(m[3])))
-  ;
+#if defined(CGLM_SIMD)
+  glmm_store(m[3], glmm_fmadd(glmm_load(m[0]), glmm_set1(x), glmm_load(m[3])));
 #else
   vec4 v1;
   glm_vec4_scale(m[0], x, v1);
@@ -125,12 +113,8 @@ glm_translate_x(mat4 m, float x) {
 CGLM_INLINE
 void
 glm_translate_y(mat4 m, float y) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
-  glmm_store(m[3],
-             _mm_add_ps(_mm_mul_ps(glmm_load(m[1]),
-                                   _mm_set1_ps(y)),
-                        glmm_load(m[3])))
-  ;
+#if defined(CGLM_SIMD)
+  glmm_store(m[3], glmm_fmadd(glmm_load(m[1]), glmm_set1(y), glmm_load(m[3])));
 #else
   vec4 v1;
   glm_vec4_scale(m[1], y, v1);
@@ -147,12 +131,8 @@ glm_translate_y(mat4 m, float y) {
 CGLM_INLINE
 void
 glm_translate_z(mat4 m, float z) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
-  glmm_store(m[3],
-             _mm_add_ps(_mm_mul_ps(glmm_load(m[2]),
-                                   _mm_set1_ps(z)),
-                        glmm_load(m[3])))
-  ;
+#if defined(CGLM_SIMD)
+  glmm_store(m[3], glmm_fmadd(glmm_load(m[2]), glmm_set1(z), glmm_load(m[3])));
 #else
   vec4 v1;
   glm_vec4_scale(m[2], z, v1);
@@ -459,7 +439,7 @@ glm_decompose_rs(mat4 m, mat4 r, vec3 s) {
   glm_vec4_scale(r[1], 1.0f/s[1], r[1]);
   glm_vec4_scale(r[2], 1.0f/s[2], r[2]);
 
-  /* Note from Apple Open Source (asume that the matrix is orthonormal):
+  /* Note from Apple Open Source (assume that the matrix is orthonormal):
      check for a coordinate system flip.  If the determinant
      is -1, then negate the matrix and the scaling factors. */
   glm_vec3_cross(m[0], m[1], v);

include/cglm/affine2d.h

@@ -0,0 +1,268 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+/*
+ Functions:
+   CGLM_INLINE void glm_translate2d(mat3 m, vec2 v)
+   CGLM_INLINE void glm_translate2d_to(mat3 m, vec2 v, mat3 dest)
+   CGLM_INLINE void glm_translate2d_x(mat3 m, float x)
+   CGLM_INLINE void glm_translate2d_y(mat3 m, float y)
+   CGLM_INLINE void glm_translate2d_make(mat3 m, vec2 v)
+   CGLM_INLINE void glm_scale2d_to(mat3 m, vec2 v, mat3 dest)
+   CGLM_INLINE void glm_scale2d_make(mat3 m, vec2 v)
+   CGLM_INLINE void glm_scale2d(mat3 m, vec2 v)
+   CGLM_INLINE void glm_scale2d_uni(mat3 m, float s)
+   CGLM_INLINE void glm_rotate2d_make(mat3 m, float angle)
+   CGLM_INLINE void glm_rotate2d(mat3 m, float angle)
+   CGLM_INLINE void glm_rotate2d_to(mat3 m, float angle, mat3 dest)
+ */
+
+#ifndef cglm_affine2d_h
+#define cglm_affine2d_h
+
+#include "common.h"
+#include "util.h"
+#include "vec2.h"
+#include "mat3.h"
+
+/*!
+ * @brief translate existing 2d transform matrix by v vector
+ *        and stores result in same matrix
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       v  translate vector [x, y]
+ */
+CGLM_INLINE
+void
+glm_translate2d(mat3 m, vec2 v) {
+  m[2][0] = m[0][0] * v[0] + m[1][0] * v[1] + m[2][0];
+  m[2][1] = m[0][1] * v[0] + m[1][1] * v[1] + m[2][1];
+  m[2][2] = m[0][2] * v[0] + m[1][2] * v[1] + m[2][2];
+}
+
+/*!
+ * @brief translate existing 2d transform matrix by v vector
+ *        and store result in dest
+ *
+ * source matrix will remain same
+ *
+ * @param[in]  m    affine transfrom
+ * @param[in]  v    translate vector [x, y]
+ * @param[out] dest translated matrix
+ */
+CGLM_INLINE
+void
+glm_translate2d_to(mat3 m, vec2 v, mat3 dest) {
+  glm_mat3_copy(m, dest);
+  glm_translate2d(dest, v);
+}
+
+/*!
+ * @brief translate existing 2d transform matrix by x factor
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       x  x factor
+ */
+CGLM_INLINE
+void
+glm_translate2d_x(mat3 m, float x) {
+  m[2][0] = m[0][0] * x + m[2][0];
+  m[2][1] = m[0][1] * x + m[2][1];
+  m[2][2] = m[0][2] * x + m[2][2];
+}
+
+/*!
+ * @brief translate existing 2d transform matrix by y factor
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       y  y factor
+ */
+CGLM_INLINE
+void
+glm_translate2d_y(mat3 m, float y) {
+  m[2][0] = m[1][0] * y + m[2][0];
+  m[2][1] = m[1][1] * y + m[2][1];
+  m[2][2] = m[1][2] * y + m[2][2];
+}
+
+/*!
+ * @brief creates NEW translate 2d transform matrix by v vector
+ *
+ * @param[out]  m  affine transfrom
+ * @param[in]   v  translate vector [x, y]
+ */
+CGLM_INLINE
+void
+glm_translate2d_make(mat3 m, vec2 v) {
+  glm_mat3_identity(m);
+  m[2][0] = v[0];
+  m[2][1] = v[1];
+}
+
+/*!
+ * @brief scale existing 2d transform matrix by v vector
+ *        and store result in dest
+ *
+ * @param[in]  m    affine transfrom
+ * @param[in]  v    scale vector [x, y]
+ * @param[out] dest scaled matrix
+ */
+CGLM_INLINE
+void
+glm_scale2d_to(mat3 m, vec2 v, mat3 dest) {
+  dest[0][0] = m[0][0] * v[0];
+  dest[0][1] = m[0][1] * v[0];
+  dest[0][2] = m[0][2] * v[0];
+  
+  dest[1][0] = m[1][0] * v[1];
+  dest[1][1] = m[1][1] * v[1];
+  dest[1][2] = m[1][2] * v[1];
+  
+  dest[2][0] = m[2][0];
+  dest[2][1] = m[2][1];
+  dest[2][2] = m[2][2];
+}
+
+/*!
+ * @brief creates NEW 2d scale matrix by v vector
+ *
+ * @param[out]  m  affine transfrom
+ * @param[in]   v  scale vector [x, y]
+ */
+CGLM_INLINE
+void
+glm_scale2d_make(mat3 m, vec2 v) {
+  glm_mat3_identity(m);
+  m[0][0] = v[0];
+  m[1][1] = v[1];
+}
+
+/*!
+ * @brief scales existing 2d transform matrix by v vector
+ *        and stores result in same matrix
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       v  scale vector [x, y]
+ */
+CGLM_INLINE
+void
+glm_scale2d(mat3 m, vec2 v) {
+  m[0][0] = m[0][0] * v[0];
+  m[0][1] = m[0][1] * v[0];
+  m[0][2] = m[0][2] * v[0];
+
+  m[1][0] = m[1][0] * v[1];
+  m[1][1] = m[1][1] * v[1];
+  m[1][2] = m[1][2] * v[1];
+}
+
+/*!
+ * @brief applies uniform scale to existing 2d transform matrix v = [s, s]
+ *        and stores result in same matrix
+ *
+ * @param[in, out]  m  affine transfrom
+ * @param[in]       s  scale factor
+ */
+CGLM_INLINE
+void
+glm_scale2d_uni(mat3 m, float s) {
+  m[0][0] = m[0][0] * s;
+  m[0][1] = m[0][1] * s;
+  m[0][2] = m[0][2] * s;
+
+  m[1][0] = m[1][0] * s;
+  m[1][1] = m[1][1] * s;
+  m[1][2] = m[1][2] * s;
+}
+
+/*!
+ * @brief creates NEW rotation matrix by angle around Z axis
+ *
+ * @param[out] m     affine transfrom
+ * @param[in]  angle angle (radians)
+ */
+CGLM_INLINE
+void
+glm_rotate2d_make(mat3 m, float angle) {
+  float c, s;
+
+  s = sinf(angle);
+  c = cosf(angle);
+  
+  m[0][0] = c;
+  m[0][1] = s;
+  m[0][2] = 0;
+
+  m[1][0] = -s;
+  m[1][1] = c;
+  m[1][2] = 0;
+  
+  m[2][0] = 0.0f;
+  m[2][1] = 0.0f;
+  m[2][2] = 1.0f;
+}
+
+/*!
+ * @brief rotate existing 2d transform matrix around Z axis by angle
+ *         and store result in same matrix
+ *
+ * @param[in, out]  m      affine transfrom
+ * @param[in]       angle  angle (radians)
+ */
+CGLM_INLINE
+void
+glm_rotate2d(mat3 m, float angle) {
+  float m00 = m[0][0],  m10 = m[1][0],
+        m01 = m[0][1],  m11 = m[1][1],
+        m02 = m[0][2],  m12 = m[1][2];
+  float c, s;
+
+  s = sinf(angle);
+  c = cosf(angle);
+  
+  m[0][0] = m00 * c + m10 * s;
+  m[0][1] = m01 * c + m11 * s;
+  m[0][2] = m02 * c + m12 * s;
+
+  m[1][0] = m00 * -s + m10 * c;
+  m[1][1] = m01 * -s + m11 * c;
+  m[1][2] = m02 * -s + m12 * c;
+}
+
+/*!
+ * @brief rotate existing 2d transform matrix around Z axis by angle
+ *        and store result in dest
+ *
+ * @param[in]  m      affine transfrom
+ * @param[in]  angle  angle (radians)
+ * @param[out] dest   destination
+ */
+CGLM_INLINE
+void
+glm_rotate2d_to(mat3 m, float angle, mat3 dest) {
+  float m00 = m[0][0],  m10 = m[1][0],
+        m01 = m[0][1],  m11 = m[1][1],
+        m02 = m[0][2],  m12 = m[1][2];
+  float c, s;
+
+  s = sinf(angle);
+  c = cosf(angle);
+  
+  dest[0][0] = m00 * c + m10 * s;
+  dest[0][1] = m01 * c + m11 * s;
+  dest[0][2] = m02 * c + m12 * s;
+
+  dest[1][0] = m00 * -s + m10 * c;
+  dest[1][1] = m01 * -s + m11 * c;
+  dest[1][2] = m02 * -s + m12 * c;
+  
+  dest[2][0] = m[2][0];
+  dest[2][1] = m[2][1];
+  dest[2][2] = m[2][2];
+}
+
+#endif /* cglm_affine2d_h */

include/cglm/box.h

@@ -228,6 +228,8 @@ glm_aabb_aabb(vec3 box[2], vec3 other[2]) {
  * https://github.com/erich666/GraphicsGems/blob/master/gems/BoxSphere.c
  * Solid Box - Solid Sphere test.
  *
+ * Sphere Representation in cglm: [center.x, center.y, center.z, radii]
+ *
  * @param[in]   box    solid bounding box
  * @param[in]   s      solid sphere
  */
@@ -237,13 +239,13 @@ glm_aabb_sphere(vec3 box[2], vec4 s) {
   float dmin;
   int   a, b, c;
 
-  a = s[0] >= box[0][0];
-  b = s[1] >= box[0][1];
-  c = s[2] >= box[0][2];
+  a = (s[0] < box[0][0]) + (s[0] > box[1][0]);
+  b = (s[1] < box[0][1]) + (s[1] > box[1][1]);
+  c = (s[2] < box[0][2]) + (s[2] > box[1][2]);
 
-  dmin  = glm_pow2(s[0] - box[a][0])
-        + glm_pow2(s[1] - box[b][1])
-        + glm_pow2(s[2] - box[c][2]);
+  dmin  = glm_pow2((s[0] - box[!(a - 1)][0]) * (a != 0))
+        + glm_pow2((s[1] - box[!(b - 1)][1]) * (b != 0))
+        + glm_pow2((s[2] - box[!(c - 1)][2]) * (c != 0));
 
   return dmin <= glm_pow2(s[3]);
 }

include/cglm/call.h

@@ -31,6 +31,8 @@ extern "C" {
 #include "call/ease.h"
 #include "call/curve.h"
 #include "call/bezier.h"
+#include "call/ray.h"
+#include "call/affine2d.h"
 
 #ifdef __cplusplus
 }

include/cglm/call/affine2d.h

@@ -0,0 +1,67 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_affine2d_h
+#define cglmc_affine2d_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "../cglm.h"
+
+CGLM_EXPORT
+void
+glmc_translate2d_make(mat3 m, vec2 v);
+
+CGLM_EXPORT
+void
+glmc_translate2d_to(mat3 m, vec2 v, mat3 dest);
+
+CGLM_EXPORT
+void
+glmc_translate2d(mat3 m, vec2 v);
+
+CGLM_EXPORT
+void
+glmc_translate2d_x(mat3 m, float to);
+
+CGLM_EXPORT
+void
+glmc_translate2d_y(mat3 m, float to);
+
+CGLM_EXPORT
+void
+glmc_scale2d_to(mat3 m, vec2 v, mat3 dest);
+
+CGLM_EXPORT
+void
+glmc_scale2d_make(mat3 m, vec2 v);
+
+CGLM_EXPORT
+void
+glmc_scale2d(mat3 m, vec2 v);
+
+CGLM_EXPORT
+void
+glmc_scale2d_uni(mat3 m, float s);
+
+CGLM_EXPORT
+void
+glmc_rotate2d_make(mat3 m, float angle);
+
+CGLM_EXPORT
+void
+glmc_rotate2d(mat3 m, float angle);
+
+CGLM_EXPORT
+void
+glmc_rotate2d_to(mat3 m, float angle, mat3 dest);
+
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_affine2d_h */

include/cglm/call/io.h

@@ -7,6 +7,7 @@
 
 #ifndef cglmc_io_h
 #define cglmc_io_h
+
 #ifdef __cplusplus
 extern "C" {
 #endif

include/cglm/call/ray.h

@@ -0,0 +1,27 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglmc_ray_h
+#define cglmc_ray_h
+#ifdef __cplusplus
+extern "C" {
+#endif
+#include "../cglm.h"
+
+CGLM_EXPORT
+bool
+glmc_ray_triangle(vec3   origin,
+                  vec3   direction,
+                  vec3   v0,
+                  vec3   v1,
+                  vec3   v2,
+                  float *d);
+    
+#ifdef __cplusplus
+}
+#endif
+#endif /* cglmc_ray_h */

include/cglm/cglm.h

@@ -30,5 +30,7 @@
 #include "ease.h"
 #include "curve.h"
 #include "bezier.h"
+#include "ray.h"
+#include "affine2d.h"
 
 #endif /* cglm_h */

include/cglm/common.h

@@ -8,7 +8,13 @@
 #ifndef cglm_common_h
 #define cglm_common_h
 
-#define _USE_MATH_DEFINES /* for windows */
+#ifndef _USE_MATH_DEFINES
+#  define _USE_MATH_DEFINES       /* for windows */
+#endif
+
+#ifndef _CRT_SECURE_NO_WARNINGS
+#  define _CRT_SECURE_NO_WARNINGS /* for windows */
+#endif
 
 #include <stdint.h>
 #include <stddef.h>
@@ -17,7 +23,9 @@
 #include <stdbool.h>
 
 #if defined(_MSC_VER)
-#  ifdef CGLM_DLL
+#  ifdef CGLM_STATIC
+#    define CGLM_EXPORT
+#  elif defined(CGLM_EXPORTS)
 #    define CGLM_EXPORT __declspec(dllexport)
 #  else
 #    define CGLM_EXPORT __declspec(dllimport)
@@ -34,4 +42,12 @@
 #include "types.h"
 #include "simd/intrin.h"
 
+#ifndef CGLM_USE_DEFAULT_EPSILON
+#  ifndef GLM_FLT_EPSILON
+#    define GLM_FLT_EPSILON 1e-6
+#  endif
+#else
+#  define GLM_FLT_EPSILON FLT_EPSILON
+#endif
+
 #endif /* cglm_common_h */

include/cglm/euler.h

@@ -55,8 +55,6 @@ typedef enum glm_euler_seq {
   GLM_EULER_ZYX = 2 << 0 | 1 << 2 | 0 << 4
 } glm_euler_seq;
 
-typedef glm_euler_seq glm_euler_sq;
-
 CGLM_INLINE
 glm_euler_seq
 glm_euler_order(int ord[3]) {

include/cglm/io.h

@@ -15,69 +15,130 @@
    CGLM_INLINE void glm_versor_print(versor vec, FILE *ostream);
  */
 
+/*
+ cglm tried to enable print functions in debug mode and disable them in
+ release/production mode to eliminate printing costs.
+ 
+ if you need to force enable then define CGLM_DEFINE_PRINTS macro not DEBUG one
+ 
+ Print functions are enabled if:
+ 
+ - DEBUG or _DEBUG macro is defined (mostly defined automatically in debugging)
+ - CGLM_DEFINE_PRINTS macro is defined including release/production
+   which makes enabled printing always
+ - glmc_ calls for io are always prints
+
+ */
+
+/* DEPRECATED: CGLM_NO_PRINTS_NOOP (use CGLM_DEFINE_PRINTS) */
+
 #ifndef cglm_io_h
 #define cglm_io_h
+#if defined(DEBUG) || defined(_DEBUG) \
+   || defined(CGLM_DEFINE_PRINTS) || defined(CGLM_LIB_SRC) \
+   || defined(CGLM_NO_PRINTS_NOOP)
 
 #include "common.h"
 
 #include <stdio.h>
 #include <stdlib.h>
 
+#ifndef CGLM_PRINT_PRECISION
+#  define CGLM_PRINT_PRECISION    5
+#endif
+
+#ifndef CGLM_PRINT_MAX_TO_SHORT
+#  define CGLM_PRINT_MAX_TO_SHORT 1e5
+#endif
+
+#ifndef CGLM_PRINT_COLOR
+#  define CGLM_PRINT_COLOR        "\033[36m"
+#endif
+
+#ifndef CGLM_PRINT_COLOR_RESET
+#  define CGLM_PRINT_COLOR_RESET  "\033[0m"
+#endif
+
 CGLM_INLINE
 void
 glm_mat4_print(mat4              matrix,
                FILE * __restrict ostream) {
-  int i;
-  int j;
+  char buff[16];
+  int  i, j, cw[4], cwi;
 
 #define m 4
 #define n 4
 
-  fprintf(ostream, "Matrix (float%dx%d):\n", m, n);
+  fprintf(ostream, "Matrix (float%dx%d): " CGLM_PRINT_COLOR "\n" , m, n);
+
+  cw[0] = cw[1] = cw[2] = cw[3] = 0;
 
   for (i = 0; i < m; i++) {
-    fprintf(ostream, "\t|");
     for (j = 0; j < n; j++) {
-      fprintf(ostream, "%0.4f", matrix[j][i]);;
-
-      if (j != n - 1)
-        fprintf(ostream, "\t");
+      if (matrix[i][j] < CGLM_PRINT_MAX_TO_SHORT)
+        cwi = sprintf(buff, "% .*f", CGLM_PRINT_PRECISION, matrix[i][j]);
+      else
+        cwi = sprintf(buff, "% g", matrix[i][j]);
+      cw[i] = GLM_MAX(cw[i], cwi);
+    }
   }
 
-    fprintf(ostream, "|\n");
+  for (i = 0; i < m; i++) {
+    fprintf(ostream, "  |");
+
+    for (j = 0; j < n; j++)
+      if (matrix[i][j] < CGLM_PRINT_MAX_TO_SHORT)
+        fprintf(ostream, " % *.*f", cw[j], CGLM_PRINT_PRECISION, matrix[j][i]);
+      else
+        fprintf(ostream, " % *g", cw[j], matrix[j][i]);
+
+    fprintf(ostream, "  |\n");
   }
 
-  fprintf(ostream, "\n");
+  fprintf(ostream, CGLM_PRINT_COLOR_RESET "\n");
 
 #undef m
 #undef n
 }
 
+
 CGLM_INLINE
 void
 glm_mat3_print(mat3              matrix,
                FILE * __restrict ostream) {
-  int i;
-  int j;
+  char buff[16];
+  int  i, j, cw[4], cwi;
 
 #define m 3
 #define n 3
 
-  fprintf(ostream, "Matrix (float%dx%d):\n", m, n);
+  fprintf(ostream, "Matrix (float%dx%d): " CGLM_PRINT_COLOR "\n", m, n);
+
+  cw[0] = cw[1] = cw[2] = 0;
 
   for (i = 0; i < m; i++) {
-    fprintf(ostream, "\t|");
     for (j = 0; j < n; j++) {
-      fprintf(ostream, "%0.4f", matrix[j][i]);;
-
-      if (j != n - 1)
-        fprintf(ostream, "\t");
+      if (matrix[i][j] < CGLM_PRINT_MAX_TO_SHORT)
+        cwi = sprintf(buff, "% .*f", CGLM_PRINT_PRECISION, matrix[i][j]);
+      else
+        cwi = sprintf(buff, "% g", matrix[i][j]);
+      cw[i] = GLM_MAX(cw[i], cwi);
+    }
   }
 
-    fprintf(ostream, "|\n");
+  for (i = 0; i < m; i++) {
+    fprintf(ostream, "  |");
+    
+    for (j = 0; j < n; j++)
+      if (matrix[i][j] < CGLM_PRINT_MAX_TO_SHORT)
+        fprintf(ostream, " % *.*f", cw[j], CGLM_PRINT_PRECISION, matrix[j][i]);
+      else
+        fprintf(ostream, " % *g", cw[j], matrix[j][i]);
+    
+    fprintf(ostream, "  |\n");
   }
 
-  fprintf(ostream, "\n");
+  fprintf(ostream, CGLM_PRINT_COLOR_RESET "\n");
 
 #undef m
 #undef n
@@ -87,27 +148,39 @@ CGLM_INLINE
 void
 glm_mat2_print(mat2              matrix,
                FILE * __restrict ostream) {
-  int i;
-  int j;
+  char buff[16];
+  int  i, j, cw[4], cwi;
 
 #define m 2
 #define n 2
 
-  fprintf(ostream, "Matrix (float%dx%d):\n", m, n);
+  fprintf(ostream, "Matrix (float%dx%d): " CGLM_PRINT_COLOR "\n", m, n);
+
+  cw[0] = cw[1] = 0;
 
   for (i = 0; i < m; i++) {
-    fprintf(ostream, "\t|");
     for (j = 0; j < n; j++) {
-      fprintf(ostream, "%0.4f", matrix[j][i]);;
-
-      if (j != n - 1)
-        fprintf(ostream, "\t");
+      if (matrix[i][j] < CGLM_PRINT_MAX_TO_SHORT)
+        cwi = sprintf(buff, "% .*f", CGLM_PRINT_PRECISION, matrix[i][j]);
+      else
+        cwi = sprintf(buff, "% g", matrix[i][j]);
+      cw[i] = GLM_MAX(cw[i], cwi);
+    }
   }
 
-    fprintf(ostream, "|\n");
+  for (i = 0; i < m; i++) {
+    fprintf(ostream, "  |");
+    
+    for (j = 0; j < n; j++)
+      if (matrix[i][j] < CGLM_PRINT_MAX_TO_SHORT)
+        fprintf(ostream, " % *.*f", cw[j], CGLM_PRINT_PRECISION, matrix[j][i]);
+      else
+        fprintf(ostream, " % *g", cw[j], matrix[j][i]);
+    
+    fprintf(ostream, "  |\n");
   }
 
-  fprintf(ostream, "\n");
+  fprintf(ostream, CGLM_PRINT_COLOR_RESET "\n");
 
 #undef m
 #undef n
@@ -121,16 +194,16 @@ glm_vec4_print(vec4              vec,
 
 #define m 4
 
-  fprintf(ostream, "Vector (float%d):\n\t|", m);
+  fprintf(ostream, "Vector (float%d): " CGLM_PRINT_COLOR "\n  (", m);
 
   for (i = 0; i < m; i++) {
-    fprintf(ostream, "%0.4f", vec[i]);
-
-    if (i != m - 1)
-      fprintf(ostream, "\t");
+    if (vec[i] < CGLM_PRINT_MAX_TO_SHORT)
+      fprintf(ostream, " % .*f", CGLM_PRINT_PRECISION, vec[i]);
+    else
+      fprintf(ostream, " % g", vec[i]);
   }
 
-  fprintf(ostream, "|\n\n");
+  fprintf(ostream, "  )" CGLM_PRINT_COLOR_RESET "\n\n");
 
 #undef m
 }
@@ -143,16 +216,16 @@ glm_vec3_print(vec3              vec,
 
 #define m 3
 
-  fprintf(ostream, "Vector (float%d):\n\t|", m);
+  fprintf(ostream, "Vector (float%d): " CGLM_PRINT_COLOR "\n  (", m);
 
   for (i = 0; i < m; i++) {
-    fprintf(ostream, "%0.4f", vec[i]);
-
-    if (i != m - 1)
-      fprintf(ostream, "\t");
+    if (vec[i] < CGLM_PRINT_MAX_TO_SHORT)
+      fprintf(ostream, " % .*f", CGLM_PRINT_PRECISION, vec[i]);
+    else
+      fprintf(ostream, " % g", vec[i]);
   }
 
-  fprintf(ostream, "|\n\n");
+  fprintf(ostream, "  )" CGLM_PRINT_COLOR_RESET "\n\n");
 
 #undef m
 }
@@ -165,16 +238,12 @@ glm_ivec3_print(ivec3             vec,
 
 #define m 3
 
-  fprintf(ostream, "Vector (int%d):\n\t|", m);
+  fprintf(ostream, "Vector (int%d): " CGLM_PRINT_COLOR "\n  (", m);
 
-  for (i = 0; i < m; i++) {
-    fprintf(ostream, "%d", vec[i]);
+  for (i = 0; i < m; i++)
+    fprintf(ostream, " % d", vec[i]);
 
-    if (i != m - 1)
-      fprintf(ostream, "\t");
-  }
-
-  fprintf(ostream, "|\n\n");
+  fprintf(ostream, "  )" CGLM_PRINT_COLOR_RESET "\n\n");
   
 #undef m
 }
@@ -187,16 +256,16 @@ glm_vec2_print(vec2              vec,
 
 #define m 2
 
-  fprintf(ostream, "Vector (float%d):\n\t|", m);
+  fprintf(ostream, "Vector (float%d): " CGLM_PRINT_COLOR "\n  (", m);
 
   for (i = 0; i < m; i++) {
-    fprintf(ostream, "%0.4f", vec[i]);
-
-    if (i != m - 1)
-      fprintf(ostream, "\t");
+    if (vec[i] < CGLM_PRINT_MAX_TO_SHORT)
+      fprintf(ostream, " % .*f", CGLM_PRINT_PRECISION, vec[i]);
+    else
+      fprintf(ostream, " % g", vec[i]);
   }
 
-  fprintf(ostream, "|\n\n");
+  fprintf(ostream, "  )" CGLM_PRINT_COLOR_RESET "\n\n");
 
 #undef m
 }
@@ -209,16 +278,17 @@ glm_versor_print(versor            vec,
 
 #define m 4
 
-  fprintf(ostream, "Versor (float%d):\n\t|", m);
+  fprintf(ostream, "Quaternion (float%d): " CGLM_PRINT_COLOR "\n  (", m);
 
   for (i = 0; i < m; i++) {
-    fprintf(ostream, "%0.4f", vec[i]);
-
-    if (i != m - 1)
-      fprintf(ostream, "\t");
+    if (vec[i] < CGLM_PRINT_MAX_TO_SHORT)
+      fprintf(ostream, " % .*f", CGLM_PRINT_PRECISION, vec[i]);
+    else
+      fprintf(ostream, " % g", vec[i]);
   }
 
-  fprintf(ostream, "|\n\n");
+
+  fprintf(ostream, "  )" CGLM_PRINT_COLOR_RESET "\n\n");
 
 #undef m
 }
@@ -232,24 +302,43 @@ glm_aabb_print(vec3                    bbox[2],
 
 #define m 3
 
-  fprintf(ostream, "AABB (%s):\n", tag ? tag: "float");
+  fprintf(ostream, "AABB (%s): " CGLM_PRINT_COLOR "\n", tag ? tag: "float");
 
   for (i = 0; i < 2; i++) {
-    fprintf(ostream, "\t|");
+    fprintf(ostream, "  (");
     
     for (j = 0; j < m; j++) {
-      fprintf(ostream, "%0.4f", bbox[i][j]);
-
-      if (j != m - 1)
-        fprintf(ostream, "\t");
+      if (bbox[i][j] < CGLM_PRINT_MAX_TO_SHORT)
+        fprintf(ostream, " % .*f", CGLM_PRINT_PRECISION, bbox[i][j]);
+      else
+        fprintf(ostream, " % g", bbox[i][j]);
     }
 
-    fprintf(ostream, "|\n");
+    fprintf(ostream, "  )\n");
   }
 
-  fprintf(ostream, "\n");
+  fprintf(ostream, CGLM_PRINT_COLOR_RESET "\n");
 
 #undef m
 }
 
+#else
+
+#include "common.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+
+/* NOOP: Remove print from DEBUG */
+#define glm_mat4_print(v, s) (void)v; (void)s;
+#define glm_mat3_print(v, s) (void)v; (void)s;
+#define glm_mat2_print(v, s) (void)v; (void)s;
+#define glm_vec4_print(v, s) (void)v; (void)s;
+#define glm_vec3_print(v, s) (void)v; (void)s;
+#define glm_ivec3_print(v, s) (void)v; (void)s;
+#define glm_vec2_print(v, s) (void)v; (void)s;
+#define glm_versor_print(v, s) (void)v; (void)s;
+#define glm_aabb_print(v, t, s) (void)v; (void)t; (void)s;
+
+#endif
 #endif /* cglm_io_h */

include/cglm/mat2.h

@@ -56,7 +56,7 @@
 CGLM_INLINE
 void
 glm_mat2_copy(mat2 mat, mat2 dest) {
-  glm_vec4_copy(mat[0], dest[0]);
+  glm_vec4_ucopy(mat[0], dest[0]);
 }
 
 /*!
@@ -216,7 +216,16 @@ glm_mat2_trace(mat2 m) {
 CGLM_INLINE
 void
 glm_mat2_scale(mat2 m, float s) {
-  glm_vec4_scale(m[0], s, m[0]);
+#if defined( __SSE__ ) || defined( __SSE2__ )
+  glmm_store(m[0], _mm_mul_ps(_mm_loadu_ps(m[0]), _mm_set1_ps(s)));
+#elif defined(CGLM_NEON_FP)
+  vst1q_f32(m[0], vmulq_f32(vld1q_f32(m[0]), vdupq_n_f32(s)));
+#else
+  m[0][0] = m[0][0] * s;
+  m[0][1] = m[0][1] * s;
+  m[1][0] = m[1][0] * s;
+  m[1][1] = m[1][1] * s;
+#endif
 }
 
 /*!

include/cglm/mat3.h

@@ -228,9 +228,11 @@ glm_mat3_transpose(mat3 m) {
 CGLM_INLINE
 void
 glm_mat3_mulv(mat3 m, vec3 v, vec3 dest) {
-  dest[0] = m[0][0] * v[0] + m[1][0] * v[1] + m[2][0] * v[2];
-  dest[1] = m[0][1] * v[0] + m[1][1] * v[1] + m[2][1] * v[2];
-  dest[2] = m[0][2] * v[0] + m[1][2] * v[1] + m[2][2] * v[2];
+  vec3 res;
+  res[0] = m[0][0] * v[0] + m[1][0] * v[1] + m[2][0] * v[2];
+  res[1] = m[0][1] * v[0] + m[1][1] * v[1] + m[2][1] * v[2];
+  res[2] = m[0][2] * v[0] + m[1][2] * v[1] + m[2][2] * v[2];
+  glm_vec3_copy(res, dest);
 }
 
 /*!

include/cglm/mat4.h

@@ -358,6 +358,8 @@ void
 glm_mat4_mulv(mat4 m, vec4 v, vec4 dest) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_mulv_sse2(m, v, dest);
+#elif defined(CGLM_NEON_FP)
+  glm_mat4_mulv_neon(m, v, dest);
 #else
   vec4 res;
   res[0] = m[0][0] * v[0] + m[1][0] * v[1] + m[2][0] * v[2] + m[3][0] * v[3];
@@ -476,6 +478,8 @@ void
 glm_mat4_transpose_to(mat4 m, mat4 dest) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_transp_sse2(m, dest);
+#elif defined(CGLM_NEON_FP)
+  glm_mat4_transp_neon(m, dest);
 #else
   dest[0][0] = m[0][0]; dest[1][0] = m[0][1];
   dest[0][1] = m[1][0]; dest[1][1] = m[1][1];
@@ -498,6 +502,8 @@ void
 glm_mat4_transpose(mat4 m) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_transp_sse2(m, m);
+#elif defined(CGLM_NEON_FP)
+  glm_mat4_transp_neon(m, m);
 #else
   mat4 d;
   glm_mat4_transpose_to(m, d);
@@ -533,15 +539,12 @@ glm_mat4_scale_p(mat4 m, float s) {
 CGLM_INLINE
 void
 glm_mat4_scale(mat4 m, float s) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
+#ifdef __AVX__
+  glm_mat4_scale_avx(m, s);
+#elif defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_scale_sse2(m, s);
 #elif defined(CGLM_NEON_FP)
-  float32x4_t v0;
-  v0 = vdupq_n_f32(s);
-  vst1q_f32(m[0], vmulq_f32(vld1q_f32(m[0]), v0));
-  vst1q_f32(m[1], vmulq_f32(vld1q_f32(m[1]), v0));
-  vst1q_f32(m[2], vmulq_f32(vld1q_f32(m[2]), v0));
-  vst1q_f32(m[3], vmulq_f32(vld1q_f32(m[3]), v0));
+  glm_mat4_scale_neon(m, s);
 #else
   glm_mat4_scale_p(m, s);
 #endif
@@ -559,6 +562,8 @@ float
 glm_mat4_det(mat4 mat) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   return glm_mat4_det_sse2(mat);
+#elif defined(CGLM_NEON_FP)
+  return glm_mat4_det_neon(mat);
 #else
   /* [square] det(A) = det(At) */
   float t[6];
@@ -592,6 +597,8 @@ void
 glm_mat4_inv(mat4 mat, mat4 dest) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glm_mat4_inv_sse2(mat, dest);
+#elif defined(CGLM_NEON_FP)
+  glm_mat4_inv_neon(mat, dest);
 #else
   float t[6];
   float det;

include/cglm/quat.h

@@ -63,10 +63,6 @@
 #  include "simd/sse2/quat.h"
 #endif
 
-CGLM_INLINE
-void
-glm_mat4_identity(mat4 mat);
-
 CGLM_INLINE
 void
 glm_mat4_mulv(mat4 m, vec4 v, vec4 dest);

include/cglm/ray.h

@@ -0,0 +1,77 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+/*
+ Functions:
+   CGLM_INLINE bool glm_line_triangle_intersect(vec3   origin,
+                                                vec3   direction,
+                                                vec3   v0,
+                                                vec3   v1,
+                                                vec3   v2,
+                                                float *d);
+*/
+
+#ifndef cglm_ray_h
+#define cglm_ray_h
+
+#include "vec3.h"
+
+/*!
+ * @brief Möller–Trumbore ray-triangle intersection algorithm
+ * 
+ * @param[in] origin         origin of ray
+ * @param[in] direction      direction of ray
+ * @param[in] v0             first vertex of triangle
+ * @param[in] v1             second vertex of triangle
+ * @param[in] v2             third vertex of triangle
+ * @param[in, out] d         distance to intersection
+ * @return whether there is intersection
+ */
+
+CGLM_INLINE
+bool
+glm_ray_triangle(vec3   origin,
+                 vec3   direction,
+                 vec3   v0,
+                 vec3   v1,
+                 vec3   v2,
+                 float *d) {
+  vec3        edge1, edge2, p, t, q;
+  float       det, inv_det, u, v, dist;
+  const float epsilon = 0.000001f;
+
+  glm_vec3_sub(v1, v0, edge1);
+  glm_vec3_sub(v2, v0, edge2);
+  glm_vec3_cross(direction, edge2, p);
+
+  det = glm_vec3_dot(edge1, p);
+  if (det > -epsilon && det < epsilon)
+    return false;
+
+  inv_det = 1.0f / det;
+  
+  glm_vec3_sub(origin, v0, t);
+
+  u = inv_det * glm_vec3_dot(t, p);
+  if (u < 0.0f || u > 1.0f)
+    return false;
+
+  glm_vec3_cross(t, edge1, q);
+
+  v = inv_det * glm_vec3_dot(direction, q);
+  if (v < 0.0f || u + v > 1.0f)
+    return false;
+
+  dist = inv_det * glm_vec3_dot(edge2, q);
+
+  if (d)
+    *d = dist;
+
+  return dist > epsilon;
+}
+
+#endif

include/cglm/simd/arm.h

@@ -10,19 +10,42 @@
 #include "intrin.h"
 #ifdef CGLM_SIMD_ARM
 
+#if defined(_M_ARM64) || defined(_M_HYBRID_X86_ARM64) || defined(_M_ARM64EC) || defined(__aarch64__)
+# define CGLM_ARM64 1
+#endif
+
 #define glmm_load(p)      vld1q_f32(p)
 #define glmm_store(p, a)  vst1q_f32(p, a)
 
+#define glmm_set1(x) vdupq_n_f32(x)
+#define glmm_128     float32x4_t
+
+#define glmm_splat_x(x) vdupq_lane_f32(vget_low_f32(x),  0)
+#define glmm_splat_y(x) vdupq_lane_f32(vget_low_f32(x),  1)
+#define glmm_splat_z(x) vdupq_lane_f32(vget_high_f32(x), 0)
+#define glmm_splat_w(x) vdupq_lane_f32(vget_high_f32(x), 1)
+
+#define glmm_xor(a, b)                                                        \
+  vreinterpretq_f32_s32(veorq_s32(vreinterpretq_s32_f32(a),                   \
+                                  vreinterpretq_s32_f32(b)))
+
 static inline
 float32x4_t
 glmm_abs(float32x4_t v) {
   return vabsq_f32(v);
 }
 
+static inline
+float32x4_t
+glmm_vhadd(float32x4_t v) {
+  v = vaddq_f32(v, vrev64q_f32(v));
+  return vaddq_f32(v, vcombine_f32(vget_high_f32(v), vget_low_f32(v)));
+}
+
 static inline
 float
 glmm_hadd(float32x4_t v) {
-#if defined(__aarch64__)
+#if CGLM_ARM64
   return vaddvq_f32(v);
 #else
   v = vaddq_f32(v, vrev64q_f32(v));
@@ -79,5 +102,58 @@ glmm_norm_inf(float32x4_t a) {
   return glmm_hmax(glmm_abs(a));
 }
 
+static inline
+float32x4_t
+glmm_div(float32x4_t a, float32x4_t b) {
+#if CGLM_ARM64
+  return vdivq_f32(a, b);
+#else
+  /* 2 iterations of Newton-Raphson refinement of reciprocal */
+  float32x4_t r0, r1;
+  r0 = vrecpeq_f32(b);
+  r1 = vrecpsq_f32(r0, b);
+  r0 = vmulq_f32(r1, r0);
+  r1 = vrecpsq_f32(r0, b);
+  r0 = vmulq_f32(r1, r0);
+  return vmulq_f32(a, r0);
+#endif
+}
+
+static inline
+float32x4_t
+glmm_fmadd(float32x4_t a, float32x4_t b, float32x4_t c) {
+#if CGLM_ARM64
+  return vfmaq_f32(c, a, b); /* why vfmaq_f32 is slower than vmlaq_f32 ??? */
+#else
+  return vmlaq_f32(c, a, b);
+#endif
+}
+
+static inline
+float32x4_t
+glmm_fnmadd(float32x4_t a, float32x4_t b, float32x4_t c) {
+#if CGLM_ARM64
+  return vfmsq_f32(c, a, b);
+#else
+  return vmlsq_f32(c, a, b);
+#endif
+}
+
+static inline
+float32x4_t
+glmm_fmsub(float32x4_t a, float32x4_t b, float32x4_t c) {
+#if CGLM_ARM64
+  return vfmsq_f32(c, a, b);
+#else
+  return vmlsq_f32(c, a, b);
+#endif
+}
+
+static inline
+float32x4_t
+glmm_fnmsub(float32x4_t a, float32x4_t b, float32x4_t c) {
+  return vsubq_f32(vdupq_n_f32(0.0f), glmm_fmadd(a, b, c));
+}
+
 #endif
 #endif /* cglm_simd_arm_h */

include/cglm/simd/avx/mat4.h

@@ -14,6 +14,16 @@
 
 #include <immintrin.h>
 
+CGLM_INLINE
+void
+glm_mat4_scale_avx(mat4 m, float s) {
+  __m256 y0;
+  y0 = _mm256_set1_ps(s);
+  
+  glmm_store256(m[0], _mm256_mul_ps(y0, glmm_load256(m[0])));
+  glmm_store256(m[2], _mm256_mul_ps(y0, glmm_load256(m[2])));
+}
+
 CGLM_INLINE
 void
 glm_mat4_mul_avx(mat4 m1, mat4 m2, mat4 dest) {

include/cglm/simd/neon/affine.h

@@ -0,0 +1,80 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#ifndef cglm_affine_neon_h
+#define cglm_affine_neon_h
+#if defined(__ARM_NEON_FP)
+
+#include "../../common.h"
+#include "../intrin.h"
+
+CGLM_INLINE
+void
+glm_mul_neon(mat4 m1, mat4 m2, mat4 dest) {
+  /* D = R * L (Column-Major) */
+
+  glmm_128 l0, l1, l2, l3, r0, r1, r2, r3, v0, v1, v2, v3;
+
+  l0 = glmm_load(m1[0]);    r0 = glmm_load(m2[0]);
+  l1 = glmm_load(m1[1]);    r1 = glmm_load(m2[1]);
+  l2 = glmm_load(m1[2]);    r2 = glmm_load(m2[2]);
+  l3 = glmm_load(m1[3]);    r3 = glmm_load(m2[3]);
+
+  v0 = vmulq_f32(glmm_splat_x(r0), l0);
+  v1 = vmulq_f32(glmm_splat_x(r1), l0);
+  v2 = vmulq_f32(glmm_splat_x(r2), l0);
+  v3 = vmulq_f32(glmm_splat_x(r3), l0);
+
+  v0 = glmm_fmadd(glmm_splat_y(r0), l1, v0);
+  v1 = glmm_fmadd(glmm_splat_y(r1), l1, v1);
+  v2 = glmm_fmadd(glmm_splat_y(r2), l1, v2);
+  v3 = glmm_fmadd(glmm_splat_y(r3), l1, v3);
+
+  v0 = glmm_fmadd(glmm_splat_z(r0), l2, v0);
+  v1 = glmm_fmadd(glmm_splat_z(r1), l2, v1);
+  v2 = glmm_fmadd(glmm_splat_z(r2), l2, v2);
+  v3 = glmm_fmadd(glmm_splat_z(r3), l2, v3);
+
+  v3 = glmm_fmadd(glmm_splat_w(r3), l3, v3);
+
+  glmm_store(dest[0], v0);
+  glmm_store(dest[1], v1);
+  glmm_store(dest[2], v2);
+  glmm_store(dest[3], v3);
+}
+
+CGLM_INLINE
+void
+glm_mul_rot_neon(mat4 m1, mat4 m2, mat4 dest) {
+  /* D = R * L (Column-Major) */
+
+  glmm_128 l0, l1, l2, r0, r1, r2, v0, v1, v2;
+
+  l0 = glmm_load(m1[0]);    r0 = glmm_load(m2[0]);
+  l1 = glmm_load(m1[1]);    r1 = glmm_load(m2[1]);
+  l2 = glmm_load(m1[2]);    r2 = glmm_load(m2[2]);
+
+  v0 = vmulq_f32(glmm_splat_x(r0), l0);
+  v1 = vmulq_f32(glmm_splat_x(r1), l0);
+  v2 = vmulq_f32(glmm_splat_x(r2), l0);
+
+  v0 = glmm_fmadd(glmm_splat_y(r0), l1, v0);
+  v1 = glmm_fmadd(glmm_splat_y(r1), l1, v1);
+  v2 = glmm_fmadd(glmm_splat_y(r2), l1, v2);
+
+  v0 = glmm_fmadd(glmm_splat_z(r0), l2, v0);
+  v1 = glmm_fmadd(glmm_splat_z(r1), l2, v1);
+  v2 = glmm_fmadd(glmm_splat_z(r2), l2, v2);
+
+  glmm_store(dest[0], v0);
+  glmm_store(dest[1], v1);
+  glmm_store(dest[2], v2);
+  glmm_store(dest[3], glmm_load(m1[3]));
+}
+
+#endif
+#endif /* cglm_affine_neon_h */

include/cglm/simd/neon/mat4.h

@@ -12,45 +12,301 @@
 #include "../../common.h"
 #include "../intrin.h"
 
+CGLM_INLINE
+void
+glm_mat4_scale_neon(mat4 m, float s) {
+  float32x4_t v0;
+  
+  v0 = vdupq_n_f32(s);
+
+  vst1q_f32(m[0], vmulq_f32(vld1q_f32(m[0]), v0));
+  vst1q_f32(m[1], vmulq_f32(vld1q_f32(m[1]), v0));
+  vst1q_f32(m[2], vmulq_f32(vld1q_f32(m[2]), v0));
+  vst1q_f32(m[3], vmulq_f32(vld1q_f32(m[3]), v0));
+}
+
+CGLM_INLINE
+void
+glm_mat4_transp_neon(mat4 m, mat4 dest) {
+  float32x4x4_t vmat;
+  
+  vmat = vld4q_f32(m[0]);
+
+  vst1q_f32(dest[0], vmat.val[0]);
+  vst1q_f32(dest[1], vmat.val[1]);
+  vst1q_f32(dest[2], vmat.val[2]);
+  vst1q_f32(dest[3], vmat.val[3]);
+}
+
 CGLM_INLINE
 void
 glm_mat4_mul_neon(mat4 m1, mat4 m2, mat4 dest) {
   /* D = R * L (Column-Major) */
-  float32x4_t l0, l1, l2, l3, r, d0, d1, d2, d3;
 
-  l0 = vld1q_f32(m2[0]);
-  l1 = vld1q_f32(m2[1]);
-  l2 = vld1q_f32(m2[2]);
-  l3 = vld1q_f32(m2[3]);
+  glmm_128 l0, l1, l2, l3, r0, r1, r2, r3, v0, v1, v2, v3;
 
-  r  = vld1q_f32(m1[0]);
-  d0 = vmulq_lane_f32(r, vget_low_f32(l0), 0);
-  d1 = vmulq_lane_f32(r, vget_low_f32(l1), 0);
-  d2 = vmulq_lane_f32(r, vget_low_f32(l2), 0);
-  d3 = vmulq_lane_f32(r, vget_low_f32(l3), 0);
+  l0 = glmm_load(m1[0]);    r0 = glmm_load(m2[0]);
+  l1 = glmm_load(m1[1]);    r1 = glmm_load(m2[1]);
+  l2 = glmm_load(m1[2]);    r2 = glmm_load(m2[2]);
+  l3 = glmm_load(m1[3]);    r3 = glmm_load(m2[3]);
 
-  r  = vld1q_f32(m1[1]);
-  d0 = vmlaq_lane_f32(d0, r, vget_low_f32(l0), 1);
-  d1 = vmlaq_lane_f32(d1, r, vget_low_f32(l1), 1);
-  d2 = vmlaq_lane_f32(d2, r, vget_low_f32(l2), 1);
-  d3 = vmlaq_lane_f32(d3, r, vget_low_f32(l3), 1);
+  v0 = vmulq_f32(glmm_splat_x(r0), l0);
+  v1 = vmulq_f32(glmm_splat_x(r1), l0);
+  v2 = vmulq_f32(glmm_splat_x(r2), l0);
+  v3 = vmulq_f32(glmm_splat_x(r3), l0);
 
-  r  = vld1q_f32(m1[2]);
-  d0 = vmlaq_lane_f32(d0, r, vget_high_f32(l0), 0);
-  d1 = vmlaq_lane_f32(d1, r, vget_high_f32(l1), 0);
-  d2 = vmlaq_lane_f32(d2, r, vget_high_f32(l2), 0);
-  d3 = vmlaq_lane_f32(d3, r, vget_high_f32(l3), 0);
+  v0 = glmm_fmadd(glmm_splat_y(r0), l1, v0);
+  v1 = glmm_fmadd(glmm_splat_y(r1), l1, v1);
+  v2 = glmm_fmadd(glmm_splat_y(r2), l1, v2);
+  v3 = glmm_fmadd(glmm_splat_y(r3), l1, v3);
 
-  r  = vld1q_f32(m1[3]);
-  d0 = vmlaq_lane_f32(d0, r, vget_high_f32(l0), 1);
-  d1 = vmlaq_lane_f32(d1, r, vget_high_f32(l1), 1);
-  d2 = vmlaq_lane_f32(d2, r, vget_high_f32(l2), 1);
-  d3 = vmlaq_lane_f32(d3, r, vget_high_f32(l3), 1);
+  v0 = glmm_fmadd(glmm_splat_z(r0), l2, v0);
+  v1 = glmm_fmadd(glmm_splat_z(r1), l2, v1);
+  v2 = glmm_fmadd(glmm_splat_z(r2), l2, v2);
+  v3 = glmm_fmadd(glmm_splat_z(r3), l2, v3);
 
-  vst1q_f32(dest[0], d0);
-  vst1q_f32(dest[1], d1);
-  vst1q_f32(dest[2], d2);
-  vst1q_f32(dest[3], d3);
+  v0 = glmm_fmadd(glmm_splat_w(r0), l3, v0);
+  v1 = glmm_fmadd(glmm_splat_w(r1), l3, v1);
+  v2 = glmm_fmadd(glmm_splat_w(r2), l3, v2);
+  v3 = glmm_fmadd(glmm_splat_w(r3), l3, v3);
+
+  glmm_store(dest[0], v0);
+  glmm_store(dest[1], v1);
+  glmm_store(dest[2], v2);
+  glmm_store(dest[3], v3);
+}
+
+CGLM_INLINE
+void
+glm_mat4_mulv_neon(mat4 m, vec4 v, vec4 dest) {
+  float32x4_t l0, l1, l2, l3;
+  float32x2_t vlo, vhi;
+  
+  l0  = vld1q_f32(m[0]);
+  l1  = vld1q_f32(m[1]);
+  l2  = vld1q_f32(m[2]);
+  l3  = vld1q_f32(m[3]);
+
+  vlo = vld1_f32(&v[0]);
+  vhi = vld1_f32(&v[2]);
+
+  l0  = vmulq_lane_f32(l0, vlo, 0);
+  l0  = vmlaq_lane_f32(l0, l1, vlo, 1);
+  l0  = vmlaq_lane_f32(l0, l2, vhi, 0);
+  l0  = vmlaq_lane_f32(l0, l3, vhi, 1);
+
+  vst1q_f32(dest, l0);
+}
+
+CGLM_INLINE
+float
+glm_mat4_det_neon(mat4 mat) {
+  float32x4_t   r0, r1, r2, r3, x0, x1, x2;
+  float32x2_t   ij, op, mn, kl, nn, mm, jj, ii, gh, ef, t12, t34;
+  float32x4x2_t a1;
+  float32x4_t   x3 = { 0.f, -0.f, 0.f, -0.f };
+
+  /* 127 <- 0, [square] det(A) = det(At) */
+  r0 = glmm_load(mat[0]);              /* d c b a */
+  r1 = vrev64q_f32(glmm_load(mat[1])); /* g h e f */
+  r2 = vrev64q_f32(glmm_load(mat[2])); /* l k i j */
+  r3 = vrev64q_f32(glmm_load(mat[3])); /* o p m n */
+
+  gh = vget_high_f32(r1);
+  ef = vget_low_f32(r1);
+  kl = vget_high_f32(r2);
+  ij = vget_low_f32(r2);
+  op = vget_high_f32(r3);
+  mn = vget_low_f32(r3);
+  mm = vdup_lane_f32(mn, 1);
+  nn = vdup_lane_f32(mn, 0);
+  ii = vdup_lane_f32(ij, 1);
+  jj = vdup_lane_f32(ij, 0);
+  
+  /*
+   t[1] = j * p - n * l;
+   t[2] = j * o - n * k;
+   t[3] = i * p - m * l;
+   t[4] = i * o - m * k;
+   */
+  x0 = glmm_fnmadd(vcombine_f32(kl, kl), vcombine_f32(nn, mm),
+                   vmulq_f32(vcombine_f32(op, op), vcombine_f32(jj, ii)));
+
+  t12 = vget_low_f32(x0);
+  t34 = vget_high_f32(x0);
+  
+  /* 1 3 1 3 2 4 2 4 */
+  a1 = vuzpq_f32(x0, x0);
+  
+  /*
+   t[0] = k * p - o * l;
+   t[0] = k * p - o * l;
+   t[5] = i * n - m * j;
+   t[5] = i * n - m * j;
+   */
+  x1 = glmm_fnmadd(vcombine_f32(vdup_lane_f32(kl, 0), jj),
+                   vcombine_f32(vdup_lane_f32(op, 1), mm),
+                   vmulq_f32(vcombine_f32(vdup_lane_f32(op, 0), nn),
+                             vcombine_f32(vdup_lane_f32(kl, 1), ii)));
+
+  /*
+     a * (f * t[0] - g * t[1] + h * t[2])
+   - b * (e * t[0] - g * t[3] + h * t[4])
+   + c * (e * t[1] - f * t[3] + h * t[5])
+   - d * (e * t[2] - f * t[4] + g * t[5])
+   */
+  x2 = glmm_fnmadd(vcombine_f32(vdup_lane_f32(gh, 1), vdup_lane_f32(ef, 0)),
+                   vcombine_f32(vget_low_f32(a1.val[0]), t34),
+                   vmulq_f32(vcombine_f32(ef, vdup_lane_f32(ef, 1)),
+                             vcombine_f32(vget_low_f32(x1), t12)));
+
+  x2 = glmm_fmadd(vcombine_f32(vdup_lane_f32(gh, 0), gh),
+                  vcombine_f32(vget_low_f32(a1.val[1]), vget_high_f32(x1)), x2);
+
+  x2 = glmm_xor(x2, x3);
+
+  return glmm_hadd(vmulq_f32(x2, r0));
+}
+
+CGLM_INLINE
+void
+glm_mat4_inv_neon(mat4 mat, mat4 dest) {
+  float32x4_t   r0, r1, r2, r3,
+                v0, v1, v2, v3,
+                t0, t1, t2, t3, t4, t5,
+                x0, x1, x2, x3, x4, x5, x6, x7, x8;
+  float32x4x2_t a1;
+  float32x2_t   lp, ko, hg, jn, im, fe, ae, bf, cg, dh;
+  float32x4_t   x9 = { -0.f, 0.f, -0.f, 0.f };
+
+  x8 = vrev64q_f32(x9);
+
+  /* 127 <- 0 */
+  r0 = glmm_load(mat[0]); /* d c b a */
+  r1 = glmm_load(mat[1]); /* h g f e */
+  r2 = glmm_load(mat[2]); /* l k j i */
+  r3 = glmm_load(mat[3]); /* p o n m */
+  
+  /* l p k o, j n i m */
+  a1  = vzipq_f32(r3, r2);
+  
+  jn  = vget_high_f32(a1.val[0]);
+  im  = vget_low_f32(a1.val[0]);
+  lp  = vget_high_f32(a1.val[1]);
+  ko  = vget_low_f32(a1.val[1]);
+  hg  = vget_high_f32(r1);
+
+  x1  = vcombine_f32(vdup_lane_f32(lp, 0), lp);                   /* l p p p */
+  x2  = vcombine_f32(vdup_lane_f32(ko, 0), ko);                   /* k o o o */
+  x0  = vcombine_f32(vdup_lane_f32(lp, 1), vdup_lane_f32(hg, 1)); /* h h l l */
+  x3  = vcombine_f32(vdup_lane_f32(ko, 1), vdup_lane_f32(hg, 0)); /* g g k k */
+  
+  /* t1[0] = k * p - o * l;
+     t1[0] = k * p - o * l;
+     t2[0] = g * p - o * h;
+     t3[0] = g * l - k * h; */
+  t0 = glmm_fnmadd(x2, x0, vmulq_f32(x3, x1));
+
+  fe = vget_low_f32(r1);
+  x4 = vcombine_f32(vdup_lane_f32(jn, 0), jn);                   /* j n n n */
+  x5 = vcombine_f32(vdup_lane_f32(jn, 1), vdup_lane_f32(fe, 1)); /* f f j j */
+  
+  /* t1[1] = j * p - n * l;
+     t1[1] = j * p - n * l;
+     t2[1] = f * p - n * h;
+     t3[1] = f * l - j * h; */
+   t1 = glmm_fnmadd(x4, x0, vmulq_f32(x5, x1));
+  
+  /* t1[2] = j * o - n * k
+     t1[2] = j * o - n * k;
+     t2[2] = f * o - n * g;
+     t3[2] = f * k - j * g; */
+  t2 = glmm_fnmadd(x4, x3, vmulq_f32(x5, x2));
+  
+  x6 = vcombine_f32(vdup_lane_f32(im, 1), vdup_lane_f32(fe, 0)); /* e e i i */
+  x7 = vcombine_f32(vdup_lane_f32(im, 0), im);                   /* i m m m */
+  
+  /* t1[3] = i * p - m * l;
+     t1[3] = i * p - m * l;
+     t2[3] = e * p - m * h;
+     t3[3] = e * l - i * h; */
+  t3 = glmm_fnmadd(x7, x0, vmulq_f32(x6, x1));
+  
+  /* t1[4] = i * o - m * k;
+     t1[4] = i * o - m * k;
+     t2[4] = e * o - m * g;
+     t3[4] = e * k - i * g; */
+  t4 = glmm_fnmadd(x7, x3, vmulq_f32(x6, x2));
+  
+  /* t1[5] = i * n - m * j;
+     t1[5] = i * n - m * j;
+     t2[5] = e * n - m * f;
+     t3[5] = e * j - i * f; */
+  t5 = glmm_fnmadd(x7, x5, vmulq_f32(x6, x4));
+  
+  /* h d f b, g c e a */
+  a1 = vtrnq_f32(r0, r1);
+  
+  x4 = vrev64q_f32(a1.val[0]); /* c g a e */
+  x5 = vrev64q_f32(a1.val[1]); /* d h b f */
+
+  ae = vget_low_f32(x4);
+  cg = vget_high_f32(x4);
+  bf = vget_low_f32(x5);
+  dh = vget_high_f32(x5);
+  
+  x0 = vcombine_f32(ae, vdup_lane_f32(ae, 1)); /* a a a e */
+  x1 = vcombine_f32(bf, vdup_lane_f32(bf, 1)); /* b b b f */
+  x2 = vcombine_f32(cg, vdup_lane_f32(cg, 1)); /* c c c g */
+  x3 = vcombine_f32(dh, vdup_lane_f32(dh, 1)); /* d d d h */
+  
+  /*
+   dest[0][0] =  f * t1[0] - g * t1[1] + h * t1[2];
+   dest[0][1] =-(b * t1[0] - c * t1[1] + d * t1[2]);
+   dest[0][2] =  b * t2[0] - c * t2[1] + d * t2[2];
+   dest[0][3] =-(b * t3[0] - c * t3[1] + d * t3[2]); */
+  v0 = glmm_xor(glmm_fmadd(x3, t2, glmm_fnmadd(x2, t1, vmulq_f32(x1, t0))), x8);
+  
+  /*
+   dest[2][0] =  e * t1[1] - f * t1[3] + h * t1[5];
+   dest[2][1] =-(a * t1[1] - b * t1[3] + d * t1[5]);
+   dest[2][2] =  a * t2[1] - b * t2[3] + d * t2[5];
+   dest[2][3] =-(a * t3[1] - b * t3[3] + d * t3[5]);*/
+  v2 = glmm_xor(glmm_fmadd(x3, t5, glmm_fnmadd(x1, t3, vmulq_f32(x0, t1))), x8);
+
+  /*
+   dest[1][0] =-(e * t1[0] - g * t1[3] + h * t1[4]);
+   dest[1][1] =  a * t1[0] - c * t1[3] + d * t1[4];
+   dest[1][2] =-(a * t2[0] - c * t2[3] + d * t2[4]);
+   dest[1][3] =  a * t3[0] - c * t3[3] + d * t3[4]; */
+  v1 = glmm_xor(glmm_fmadd(x3, t4, glmm_fnmadd(x2, t3, vmulq_f32(x0, t0))), x9);
+  
+  /*
+   dest[3][0] =-(e * t1[2] - f * t1[4] + g * t1[5]);
+   dest[3][1] =  a * t1[2] - b * t1[4] + c * t1[5];
+   dest[3][2] =-(a * t2[2] - b * t2[4] + c * t2[5]);
+   dest[3][3] =  a * t3[2] - b * t3[4] + c * t3[5]; */
+  v3 = glmm_xor(glmm_fmadd(x2, t5, glmm_fnmadd(x1, t4, vmulq_f32(x0, t2))), x9);
+
+  /* determinant */
+  x0 = vcombine_f32(vget_low_f32(vzipq_f32(v0, v1).val[0]),
+                    vget_low_f32(vzipq_f32(v2, v3).val[0]));
+
+  /*
+  x0 = glmm_div(glmm_set1(1.0f), glmm_vhadd(vmulq_f32(x0, r0)));
+
+  glmm_store(dest[0], vmulq_f32(v0, x0));
+  glmm_store(dest[1], vmulq_f32(v1, x0));
+  glmm_store(dest[2], vmulq_f32(v2, x0));
+  glmm_store(dest[3], vmulq_f32(v3, x0));
+  */
+
+  x0 = glmm_vhadd(vmulq_f32(x0, r0));
+
+  glmm_store(dest[0], glmm_div(v0, x0));
+  glmm_store(dest[1], glmm_div(v1, x0));
+  glmm_store(dest[2], glmm_div(v2, x0));
+  glmm_store(dest[3], glmm_div(v3, x0));
 }
 
 #endif

include/cglm/simd/sse2/affine.h

@@ -25,28 +25,28 @@ glm_mul_sse2(mat4 m1, mat4 m2, mat4 dest) {
   
   r = glmm_load(m2[0]);
   glmm_store(dest[0],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+             glmm_fmadd(glmm_splat(r, 0), l0,
+                        glmm_fmadd(glmm_splat(r, 1), l1,
+                                   _mm_mul_ps(glmm_splat(r, 2), l2))));
   
   r = glmm_load(m2[1]);
   glmm_store(dest[1],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+             glmm_fmadd(glmm_splat(r, 0), l0,
+                        glmm_fmadd(glmm_splat(r, 1), l1,
+                                   _mm_mul_ps(glmm_splat(r, 2), l2))));
 
   r = glmm_load(m2[2]);
   glmm_store(dest[2],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+             glmm_fmadd(glmm_splat(r, 0), l0,
+                        glmm_fmadd(glmm_splat(r, 1), l1,
+                                   _mm_mul_ps(glmm_splat(r, 2), l2))));
 
   r = glmm_load(m2[3]);
   glmm_store(dest[3],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
-                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
+       glmm_fmadd(glmm_splat(r, 0), l0,
+                  glmm_fmadd(glmm_splat(r, 1), l1,
+                             glmm_fmadd(glmm_splat(r, 2), l2,
+                                        _mm_mul_ps(glmm_splat(r, 3), l3)))));
 }
 
 CGLM_INLINE
@@ -62,21 +62,22 @@ glm_mul_rot_sse2(mat4 m1, mat4 m2, mat4 dest) {
 
   r = glmm_load(m2[0]);
   glmm_store(dest[0],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+             glmm_fmadd(glmm_splat(r, 0), l0,
+                        glmm_fmadd(glmm_splat(r, 1), l1,
+                                   _mm_mul_ps(glmm_splat(r, 2), l2))));
   
   r = glmm_load(m2[1]);
   glmm_store(dest[1],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+             glmm_fmadd(glmm_splat(r, 0), l0,
+                        glmm_fmadd(glmm_splat(r, 1), l1,
+                                   _mm_mul_ps(glmm_splat(r, 2), l2))));
+  
   
   r = glmm_load(m2[2]);
   glmm_store(dest[2],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_mul_ps(glmm_shuff1x(r, 2), l2)));
+             glmm_fmadd(glmm_splat(r, 0), l0,
+                        glmm_fmadd(glmm_splat(r, 1), l1,
+                                   _mm_mul_ps(glmm_splat(r, 2), l2))));
 
   glmm_store(dest[3], l3);
 }
@@ -94,9 +95,9 @@ glm_inv_tr_sse2(mat4 mat) {
 
   _MM_TRANSPOSE4_PS(r0, r1, r2, x1);
 
-  x0 = _mm_add_ps(_mm_mul_ps(r0, glmm_shuff1(r3, 0, 0, 0, 0)),
-                  _mm_mul_ps(r1, glmm_shuff1(r3, 1, 1, 1, 1)));
-  x0 = _mm_add_ps(x0, _mm_mul_ps(r2, glmm_shuff1(r3, 2, 2, 2, 2)));
+  x0 = glmm_fmadd(r0, glmm_shuff1(r3, 0, 0, 0, 0),
+             glmm_fmadd(r1, glmm_shuff1(r3, 1, 1, 1, 1),
+                        _mm_mul_ps(r2, glmm_shuff1(r3, 2, 2, 2, 2))));
   x0 = _mm_xor_ps(x0, _mm_set1_ps(-0.f));
 
   x0 = _mm_add_ps(x0, x1);

include/cglm/simd/sse2/mat2.h

@@ -26,11 +26,11 @@ glm_mat2_mul_sse2(mat2 m1, mat2 m2, mat2 dest) {
    dest[1][0] = a * g + c * h;
    dest[1][1] = b * g + d * h;
    */
-  x0 = _mm_mul_ps(_mm_movelh_ps(x1, x1), glmm_shuff1(x2, 2, 2, 0, 0));
-  x1 = _mm_mul_ps(_mm_movehl_ps(x1, x1), glmm_shuff1(x2, 3, 3, 1, 1));
-  x1 = _mm_add_ps(x0, x1);
+  x0 = glmm_fmadd(_mm_movelh_ps(x1, x1), glmm_shuff1(x2, 2, 2, 0, 0),
+                  _mm_mul_ps(_mm_movehl_ps(x1, x1),
+                             glmm_shuff1(x2, 3, 3, 1, 1)));
 
-  glmm_store(dest[0], x1);
+  glmm_store(dest[0], x0);
 }
 
 CGLM_INLINE

include/cglm/simd/sse2/mat3.h

@@ -30,23 +30,16 @@ glm_mat3_mul_sse2(mat3 m1, mat3 m2, mat3 dest) {
   x1 = glmm_shuff2(l0, l1, 1, 0, 3, 3, 0, 3, 2, 0);
   x2 = glmm_shuff2(l1, l2, 0, 0, 3, 2, 0, 2, 1, 0);
 
-  x0 = _mm_add_ps(_mm_mul_ps(glmm_shuff1(l0, 0, 2, 1, 0),
-                             glmm_shuff1(r0, 3, 0, 0, 0)),
-                  _mm_mul_ps(x1, glmm_shuff2(r0, r1, 0, 0, 1, 1, 2, 0, 0, 0)));
-
-  x0 = _mm_add_ps(x0,
-                  _mm_mul_ps(x2, glmm_shuff2(r0, r1, 1, 1, 2, 2, 2, 0, 0, 0)));
+  x0 = glmm_fmadd(glmm_shuff1(l0, 0, 2, 1, 0), glmm_shuff1(r0, 3, 0, 0, 0),
+                  glmm_fmadd(x1, glmm_shuff2(r0, r1, 0, 0, 1, 1, 2, 0, 0, 0),
+                             _mm_mul_ps(x2, glmm_shuff2(r0, r1, 1, 1, 2, 2, 2, 0, 0, 0))));
 
   _mm_storeu_ps(dest[0], x0);
 
-  x0 = _mm_add_ps(_mm_mul_ps(glmm_shuff1(l0, 1, 0, 2, 1),
-                             _mm_shuffle_ps(r0, r1, _MM_SHUFFLE(2, 2, 3, 3))),
-                  _mm_mul_ps(glmm_shuff1(x1, 1, 0, 2, 1),
-                             glmm_shuff1(r1, 3, 3, 0, 0)));
-
-  x0 = _mm_add_ps(x0,
+  x0 = glmm_fmadd(glmm_shuff1(l0, 1, 0, 2, 1), _mm_shuffle_ps(r0, r1, _MM_SHUFFLE(2, 2, 3, 3)),
+                  glmm_fmadd(glmm_shuff1(x1, 1, 0, 2, 1), glmm_shuff1(r1, 3, 3, 0, 0),
                              _mm_mul_ps(glmm_shuff1(x2, 1, 0, 2, 1),
-                             _mm_shuffle_ps(r1, r2, _MM_SHUFFLE(0, 0, 1, 1))));
+                                        _mm_shuffle_ps(r1, r2, _MM_SHUFFLE(0, 0, 1, 1)))));
   
   _mm_storeu_ps(&dest[1][1], x0);
 

include/cglm/simd/sse2/mat4.h

@@ -56,46 +56,36 @@ glm_mat4_mul_sse2(mat4 m1, mat4 m2, mat4 dest) {
   l2 = glmm_load(m1[2]);
   l3 = glmm_load(m1[3]);
   
-  r = glmm_load(m2[0]);
-  glmm_store(dest[0],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
-                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
-  r = glmm_load(m2[1]);
-  glmm_store(dest[1],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
-                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
-  r = glmm_load(m2[2]);
-  glmm_store(dest[2],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
-                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
+#define XX(C)                                                                 \
+                                                                              \
+  r = glmm_load(m2[C]);                                                       \
+  glmm_store(dest[C],                                                         \
+       glmm_fmadd(glmm_splat(r, 0), l0,                                       \
+                   glmm_fmadd(glmm_splat(r, 1), l1,                           \
+                             glmm_fmadd(glmm_splat(r, 2), l2,                 \
+                                         _mm_mul_ps(glmm_splat(r, 3), l3)))));
 
-  r = glmm_load(m2[3]);
-  glmm_store(dest[3],
-             _mm_add_ps(_mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 0), l0),
-                                   _mm_mul_ps(glmm_shuff1x(r, 1), l1)),
-                        _mm_add_ps(_mm_mul_ps(glmm_shuff1x(r, 2), l2),
-                                   _mm_mul_ps(glmm_shuff1x(r, 3), l3))));
+  XX(0);
+  XX(1);
+  XX(2);
+  XX(3);
+
+#undef XX
 }
 
 CGLM_INLINE
 void
 glm_mat4_mulv_sse2(mat4 m, vec4 v, vec4 dest) {
-  __m128 x0, x1, x2;
+  __m128 x0, x1;
   
   x0 = glmm_load(v);
-  x1 = _mm_add_ps(_mm_mul_ps(glmm_load(m[0]), glmm_shuff1x(x0, 0)),
-                  _mm_mul_ps(glmm_load(m[1]), glmm_shuff1x(x0, 1)));
+  x1 = glmm_fmadd(glmm_load(m[0]), glmm_splat(x0, 0),
+                  glmm_fmadd(glmm_load(m[1]), glmm_splat(x0, 1),
+                             glmm_fmadd(glmm_load(m[2]), glmm_splat(x0, 2),
+                                        _mm_mul_ps(glmm_load(m[3]),
+                                                   glmm_splat(x0, 3)))));
   
-  x2 = _mm_add_ps(_mm_mul_ps(glmm_load(m[2]), glmm_shuff1x(x0, 2)),
-                  _mm_mul_ps(glmm_load(m[3]), glmm_shuff1x(x0, 3)));
-
-  glmm_store(dest, _mm_add_ps(x1, x2));
+  glmm_store(dest, x1);
 }
 
 CGLM_INLINE
@@ -115,20 +105,18 @@ glm_mat4_det_sse2(mat4 mat) {
    t[3] = i * p - m * l;
    t[4] = i * o - m * k;
    */
-  x0 = _mm_sub_ps(_mm_mul_ps(glmm_shuff1(r2, 0, 0, 1, 1),
-                             glmm_shuff1(r3, 2, 3, 2, 3)),
-                  _mm_mul_ps(glmm_shuff1(r3, 0, 0, 1, 1),
-                             glmm_shuff1(r2, 2, 3, 2, 3)));
+  x0 = glmm_fnmadd(glmm_shuff1(r3, 0, 0, 1, 1), glmm_shuff1(r2, 2, 3, 2, 3),
+                   _mm_mul_ps(glmm_shuff1(r2, 0, 0, 1, 1),
+                              glmm_shuff1(r3, 2, 3, 2, 3)));
   /*
    t[0] = k * p - o * l;
    t[0] = k * p - o * l;
    t[5] = i * n - m * j;
    t[5] = i * n - m * j;
    */
-  x1 = _mm_sub_ps(_mm_mul_ps(glmm_shuff1(r2, 0, 0, 2, 2),
-                             glmm_shuff1(r3, 1, 1, 3, 3)),
-                  _mm_mul_ps(glmm_shuff1(r3, 0, 0, 2, 2),
-                             glmm_shuff1(r2, 1, 1, 3, 3)));
+  x1 = glmm_fnmadd(glmm_shuff1(r3, 0, 0, 2, 2), glmm_shuff1(r2, 1, 1, 3, 3),
+                   _mm_mul_ps(glmm_shuff1(r2, 0, 0, 2, 2),
+                              glmm_shuff1(r3, 1, 1, 3, 3)));
 
   /*
      a * (f * t[0] - g * t[1] + h * t[2])
@@ -136,21 +124,16 @@ glm_mat4_det_sse2(mat4 mat) {
    + c * (e * t[1] - f * t[3] + h * t[5])
    - d * (e * t[2] - f * t[4] + g * t[5])
    */
-  x2 = _mm_sub_ps(_mm_mul_ps(glmm_shuff1(r1, 0, 0, 0, 1),
-                             _mm_shuffle_ps(x1, x0, _MM_SHUFFLE(1, 0, 0, 0))),
-                  _mm_mul_ps(glmm_shuff1(r1, 1, 1, 2, 2),
-                             glmm_shuff1(x0, 3, 2, 2, 0)));
+  x2 = glmm_fnmadd(glmm_shuff1(r1, 1, 1, 2, 2), glmm_shuff1(x0, 3, 2, 2, 0),
+                   _mm_mul_ps(glmm_shuff1(r1, 0, 0, 0, 1),
+                              _mm_shuffle_ps(x1, x0, _MM_SHUFFLE(1, 0, 0, 0))));
+  x2 = glmm_fmadd(glmm_shuff1(r1, 2, 3, 3, 3),
+                  _mm_shuffle_ps(x0, x1, _MM_SHUFFLE(2, 2, 3, 1)),
+                  x2);
   
-  x2 = _mm_add_ps(x2,
-                  _mm_mul_ps(glmm_shuff1(r1, 2, 3, 3, 3),
-                             _mm_shuffle_ps(x0, x1, _MM_SHUFFLE(2, 2, 3, 1))));
   x2 = _mm_xor_ps(x2, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
   
-  x0 = _mm_mul_ps(r0, x2);
-  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 0, 1, 2, 3));
-  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 3, 3, 1));
-
-  return _mm_cvtss_f32(x0);
+  return glmm_hadd(_mm_mul_ps(x2, r0));
 }
 
 CGLM_INLINE
@@ -159,7 +142,10 @@ glm_mat4_inv_fast_sse2(mat4 mat, mat4 dest) {
   __m128 r0, r1, r2, r3,
          v0, v1, v2, v3,
          t0, t1, t2, t3, t4, t5,
-         x0, x1, x2, x3, x4, x5, x6, x7;
+         x0, x1, x2, x3, x4, x5, x6, x7, x8, x9;
+  
+  x8 = _mm_set_ps(-0.f, 0.f, -0.f, 0.f);
+  x9 = glmm_shuff1(x8, 2, 1, 2, 1);
   
   /* 127 <- 0 */
   r0 = glmm_load(mat[0]); /* d c b a */
@@ -177,7 +163,7 @@ glm_mat4_inv_fast_sse2(mat4 mat, mat4 dest) {
      t1[0] = k * p - o * l;
      t2[0] = g * p - o * h;
      t3[0] = g * l - k * h; */
-  t0 = _mm_sub_ps(_mm_mul_ps(x3, x1), _mm_mul_ps(x2, x0));
+  t0 = glmm_fnmadd(x2, x0, _mm_mul_ps(x3, x1));
   
   x4 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(2, 1, 2, 1)); /* o n k j */
   x4 = glmm_shuff1(x4, 0, 2, 2, 2);                     /* j n n n */
@@ -187,13 +173,13 @@ glm_mat4_inv_fast_sse2(mat4 mat, mat4 dest) {
      t1[1] = j * p - n * l;
      t2[1] = f * p - n * h;
      t3[1] = f * l - j * h; */
-  t1 = _mm_sub_ps(_mm_mul_ps(x5, x1), _mm_mul_ps(x4, x0));
+   t1 = glmm_fnmadd(x4, x0, _mm_mul_ps(x5, x1));
   
   /* t1[2] = j * o - n * k
      t1[2] = j * o - n * k;
      t2[2] = f * o - n * g;
      t3[2] = f * k - j * g; */
-  t2 = _mm_sub_ps(_mm_mul_ps(x5, x2), _mm_mul_ps(x4, x3));
+  t2 = glmm_fnmadd(x4, x3, _mm_mul_ps(x5, x2));
   
   x6 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(0, 0, 0, 0)); /* e e i i */
   x7 = glmm_shuff2(r3, r2, 0, 0, 0, 0, 2, 0, 0, 0);     /* i m m m */
@@ -202,19 +188,19 @@ glm_mat4_inv_fast_sse2(mat4 mat, mat4 dest) {
      t1[3] = i * p - m * l;
      t2[3] = e * p - m * h;
      t3[3] = e * l - i * h; */
-  t3 = _mm_sub_ps(_mm_mul_ps(x6, x1), _mm_mul_ps(x7, x0));
+  t3 = glmm_fnmadd(x7, x0, _mm_mul_ps(x6, x1));
   
   /* t1[4] = i * o - m * k;
      t1[4] = i * o - m * k;
      t2[4] = e * o - m * g;
      t3[4] = e * k - i * g; */
-  t4 = _mm_sub_ps(_mm_mul_ps(x6, x2), _mm_mul_ps(x7, x3));
+  t4 = glmm_fnmadd(x7, x3, _mm_mul_ps(x6, x2));
   
   /* t1[5] = i * n - m * j;
      t1[5] = i * n - m * j;
      t2[5] = e * n - m * f;
      t3[5] = e * j - i * f; */
-  t5 = _mm_sub_ps(_mm_mul_ps(x6, x4), _mm_mul_ps(x7, x5));
+  t5 = glmm_fnmadd(x7, x5, _mm_mul_ps(x6, x4));
   
   x0 = glmm_shuff2(r1, r0, 0, 0, 0, 0, 2, 2, 2, 0); /* a a a e */
   x1 = glmm_shuff2(r1, r0, 1, 1, 1, 1, 2, 2, 2, 0); /* b b b f */
@@ -226,50 +212,35 @@ glm_mat4_inv_fast_sse2(mat4 mat, mat4 dest) {
    dest[0][1] =-(b * t1[0] - c * t1[1] + d * t1[2]);
    dest[0][2] =  b * t2[0] - c * t2[1] + d * t2[2];
    dest[0][3] =-(b * t3[0] - c * t3[1] + d * t3[2]); */
-  v0 = _mm_add_ps(_mm_mul_ps(x3, t2),
-                  _mm_sub_ps(_mm_mul_ps(x1, t0),
-                             _mm_mul_ps(x2, t1)));
-  v0 = _mm_xor_ps(v0, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
-
-  /*
-   dest[1][0] =-(e * t1[0] - g * t1[3] + h * t1[4]);
-   dest[1][1] =  a * t1[0] - c * t1[3] + d * t1[4];
-   dest[1][2] =-(a * t2[0] - c * t2[3] + d * t2[4]);
-   dest[1][3] =  a * t3[0] - c * t3[3] + d * t3[4]; */
-  v1 = _mm_add_ps(_mm_mul_ps(x3, t4),
-                  _mm_sub_ps(_mm_mul_ps(x0, t0),
-                             _mm_mul_ps(x2, t3)));
-  v1 = _mm_xor_ps(v1, _mm_set_ps(0.f, -0.f, 0.f, -0.f));
+  v0 = _mm_xor_ps(glmm_fmadd(x3, t2, glmm_fnmadd(x2, t1, _mm_mul_ps(x1, t0))), x8);
   
   /*
    dest[2][0] =  e * t1[1] - f * t1[3] + h * t1[5];
    dest[2][1] =-(a * t1[1] - b * t1[3] + d * t1[5]);
    dest[2][2] =  a * t2[1] - b * t2[3] + d * t2[5];
    dest[2][3] =-(a * t3[1] - b * t3[3] + d * t3[5]);*/
-  v2 = _mm_add_ps(_mm_mul_ps(x3, t5),
-                  _mm_sub_ps(_mm_mul_ps(x0, t1),
-                             _mm_mul_ps(x1, t3)));
-  v2 = _mm_xor_ps(v2, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
+  v2 = _mm_xor_ps(glmm_fmadd(x3, t5, glmm_fnmadd(x1, t3, _mm_mul_ps(x0, t1))), x8);
+
+  /*
+   dest[1][0] =-(e * t1[0] - g * t1[3] + h * t1[4]);
+   dest[1][1] =  a * t1[0] - c * t1[3] + d * t1[4];
+   dest[1][2] =-(a * t2[0] - c * t2[3] + d * t2[4]);
+   dest[1][3] =  a * t3[0] - c * t3[3] + d * t3[4]; */
+  v1 = _mm_xor_ps(glmm_fmadd(x3, t4, glmm_fnmadd(x2, t3, _mm_mul_ps(x0, t0))), x9);
 
   /*
    dest[3][0] =-(e * t1[2] - f * t1[4] + g * t1[5]);
    dest[3][1] =  a * t1[2] - b * t1[4] + c * t1[5];
    dest[3][2] =-(a * t2[2] - b * t2[4] + c * t2[5]);
    dest[3][3] =  a * t3[2] - b * t3[4] + c * t3[5]; */
-  v3 = _mm_add_ps(_mm_mul_ps(x2, t5),
-                  _mm_sub_ps(_mm_mul_ps(x0, t2),
-                             _mm_mul_ps(x1, t4)));
-  v3 = _mm_xor_ps(v3, _mm_set_ps(0.f, -0.f, 0.f, -0.f));
+  v3 = _mm_xor_ps(glmm_fmadd(x2, t5, glmm_fnmadd(x1, t4, _mm_mul_ps(x0, t2))), x9);
 
   /* determinant */
   x0 = _mm_shuffle_ps(v0, v1, _MM_SHUFFLE(0, 0, 0, 0));
   x1 = _mm_shuffle_ps(v2, v3, _MM_SHUFFLE(0, 0, 0, 0));
   x0 = _mm_shuffle_ps(x0, x1, _MM_SHUFFLE(2, 0, 2, 0));
 
-  x0 = _mm_mul_ps(x0, r0);
-  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 0, 1, 2, 3));
-  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 0, 1));
-  x0 = _mm_rcp_ps(x0);
+  x0 = _mm_rcp_ps(glmm_vhadd(_mm_mul_ps(x0, r0)));
 
   glmm_store(dest[0], _mm_mul_ps(v0, x0));
   glmm_store(dest[1], _mm_mul_ps(v1, x0));
@@ -283,7 +254,10 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
   __m128 r0, r1, r2, r3,
          v0, v1, v2, v3,
          t0, t1, t2, t3, t4, t5,
-         x0, x1, x2, x3, x4, x5, x6, x7;
+         x0, x1, x2, x3, x4, x5, x6, x7, x8, x9;
+  
+  x8 = _mm_set_ps(-0.f, 0.f, -0.f, 0.f);
+  x9 = glmm_shuff1(x8, 2, 1, 2, 1);
   
   /* 127 <- 0 */
   r0 = glmm_load(mat[0]); /* d c b a */
@@ -301,7 +275,7 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
      t1[0] = k * p - o * l;
      t2[0] = g * p - o * h;
      t3[0] = g * l - k * h; */
-  t0 = _mm_sub_ps(_mm_mul_ps(x3, x1), _mm_mul_ps(x2, x0));
+  t0 = glmm_fnmadd(x2, x0, _mm_mul_ps(x3, x1));
   
   x4 = _mm_shuffle_ps(r2, r3, _MM_SHUFFLE(2, 1, 2, 1)); /* o n k j */
   x4 = glmm_shuff1(x4, 0, 2, 2, 2);                     /* j n n n */
@@ -311,13 +285,13 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
      t1[1] = j * p - n * l;
      t2[1] = f * p - n * h;
      t3[1] = f * l - j * h; */
-  t1 = _mm_sub_ps(_mm_mul_ps(x5, x1), _mm_mul_ps(x4, x0));
+   t1 = glmm_fnmadd(x4, x0, _mm_mul_ps(x5, x1));
   
   /* t1[2] = j * o - n * k
      t1[2] = j * o - n * k;
      t2[2] = f * o - n * g;
      t3[2] = f * k - j * g; */
-  t2 = _mm_sub_ps(_mm_mul_ps(x5, x2), _mm_mul_ps(x4, x3));
+  t2 = glmm_fnmadd(x4, x3, _mm_mul_ps(x5, x2));
   
   x6 = _mm_shuffle_ps(r2, r1, _MM_SHUFFLE(0, 0, 0, 0)); /* e e i i */
   x7 = glmm_shuff2(r3, r2, 0, 0, 0, 0, 2, 0, 0, 0);     /* i m m m */
@@ -326,19 +300,19 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
      t1[3] = i * p - m * l;
      t2[3] = e * p - m * h;
      t3[3] = e * l - i * h; */
-  t3 = _mm_sub_ps(_mm_mul_ps(x6, x1), _mm_mul_ps(x7, x0));
+  t3 = glmm_fnmadd(x7, x0, _mm_mul_ps(x6, x1));
   
   /* t1[4] = i * o - m * k;
      t1[4] = i * o - m * k;
      t2[4] = e * o - m * g;
      t3[4] = e * k - i * g; */
-  t4 = _mm_sub_ps(_mm_mul_ps(x6, x2), _mm_mul_ps(x7, x3));
+  t4 = glmm_fnmadd(x7, x3, _mm_mul_ps(x6, x2));
   
   /* t1[5] = i * n - m * j;
      t1[5] = i * n - m * j;
      t2[5] = e * n - m * f;
      t3[5] = e * j - i * f; */
-  t5 = _mm_sub_ps(_mm_mul_ps(x6, x4), _mm_mul_ps(x7, x5));
+  t5 = glmm_fnmadd(x7, x5, _mm_mul_ps(x6, x4));
   
   x0 = glmm_shuff2(r1, r0, 0, 0, 0, 0, 2, 2, 2, 0); /* a a a e */
   x1 = glmm_shuff2(r1, r0, 1, 1, 1, 1, 2, 2, 2, 0); /* b b b f */
@@ -350,50 +324,35 @@ glm_mat4_inv_sse2(mat4 mat, mat4 dest) {
    dest[0][1] =-(b * t1[0] - c * t1[1] + d * t1[2]);
    dest[0][2] =  b * t2[0] - c * t2[1] + d * t2[2];
    dest[0][3] =-(b * t3[0] - c * t3[1] + d * t3[2]); */
-  v0 = _mm_add_ps(_mm_mul_ps(x3, t2),
-                  _mm_sub_ps(_mm_mul_ps(x1, t0),
-                             _mm_mul_ps(x2, t1)));
-  v0 = _mm_xor_ps(v0, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
-
-  /*
-   dest[1][0] =-(e * t1[0] - g * t1[3] + h * t1[4]);
-   dest[1][1] =  a * t1[0] - c * t1[3] + d * t1[4];
-   dest[1][2] =-(a * t2[0] - c * t2[3] + d * t2[4]);
-   dest[1][3] =  a * t3[0] - c * t3[3] + d * t3[4]; */
-  v1 = _mm_add_ps(_mm_mul_ps(x3, t4),
-                  _mm_sub_ps(_mm_mul_ps(x0, t0),
-                             _mm_mul_ps(x2, t3)));
-  v1 = _mm_xor_ps(v1, _mm_set_ps(0.f, -0.f, 0.f, -0.f));
+  v0 = _mm_xor_ps(glmm_fmadd(x3, t2, glmm_fnmadd(x2, t1, _mm_mul_ps(x1, t0))), x8);
   
   /*
    dest[2][0] =  e * t1[1] - f * t1[3] + h * t1[5];
    dest[2][1] =-(a * t1[1] - b * t1[3] + d * t1[5]);
    dest[2][2] =  a * t2[1] - b * t2[3] + d * t2[5];
    dest[2][3] =-(a * t3[1] - b * t3[3] + d * t3[5]);*/
-  v2 = _mm_add_ps(_mm_mul_ps(x3, t5),
-                  _mm_sub_ps(_mm_mul_ps(x0, t1),
-                             _mm_mul_ps(x1, t3)));
-  v2 = _mm_xor_ps(v2, _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
+  v2 = _mm_xor_ps(glmm_fmadd(x3, t5, glmm_fnmadd(x1, t3, _mm_mul_ps(x0, t1))), x8);
+
+  /*
+   dest[1][0] =-(e * t1[0] - g * t1[3] + h * t1[4]);
+   dest[1][1] =  a * t1[0] - c * t1[3] + d * t1[4];
+   dest[1][2] =-(a * t2[0] - c * t2[3] + d * t2[4]);
+   dest[1][3] =  a * t3[0] - c * t3[3] + d * t3[4]; */
+  v1 = _mm_xor_ps(glmm_fmadd(x3, t4, glmm_fnmadd(x2, t3, _mm_mul_ps(x0, t0))), x9);
 
   /*
    dest[3][0] =-(e * t1[2] - f * t1[4] + g * t1[5]);
    dest[3][1] =  a * t1[2] - b * t1[4] + c * t1[5];
    dest[3][2] =-(a * t2[2] - b * t2[4] + c * t2[5]);
    dest[3][3] =  a * t3[2] - b * t3[4] + c * t3[5]; */
-  v3 = _mm_add_ps(_mm_mul_ps(x2, t5),
-                  _mm_sub_ps(_mm_mul_ps(x0, t2),
-                             _mm_mul_ps(x1, t4)));
-  v3 = _mm_xor_ps(v3, _mm_set_ps(0.f, -0.f, 0.f, -0.f));
+  v3 = _mm_xor_ps(glmm_fmadd(x2, t5, glmm_fnmadd(x1, t4, _mm_mul_ps(x0, t2))), x9);
 
   /* determinant */
   x0 = _mm_shuffle_ps(v0, v1, _MM_SHUFFLE(0, 0, 0, 0));
   x1 = _mm_shuffle_ps(v2, v3, _MM_SHUFFLE(0, 0, 0, 0));
   x0 = _mm_shuffle_ps(x0, x1, _MM_SHUFFLE(2, 0, 2, 0));
 
-  x0 = _mm_mul_ps(x0, r0);
-  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 0, 1, 2, 3));
-  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 0, 1));
-  x0 = _mm_div_ps(_mm_set1_ps(1.0f), x0);
+  x0 = _mm_div_ps(_mm_set1_ps(1.0f), glmm_vhadd(_mm_mul_ps(x0, r0)));
 
   glmm_store(dest[0], _mm_mul_ps(v0, x0));
   glmm_store(dest[1], _mm_mul_ps(v1, x0));

include/cglm/simd/sse2/quat.h

@@ -27,15 +27,15 @@ glm_quat_mul_sse2(versor p, versor q, versor dest) {
   xp = glmm_load(p); /* 3 2 1 0 */
   xq = glmm_load(q);
 
-  r  = _mm_mul_ps(glmm_shuff1x(xp, 3), xq);
+  r  = _mm_mul_ps(glmm_splat(xp, 3), xq);
 
-  x0 = _mm_xor_ps(glmm_shuff1x(xp, 0), _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
+  x0 = _mm_xor_ps(glmm_splat(xp, 0), _mm_set_ps(-0.f, 0.f, -0.f, 0.f));
   r  = _mm_add_ps(r, _mm_mul_ps(x0, glmm_shuff1(xq, 0, 1, 2, 3)));
 
-  x0 = _mm_xor_ps(glmm_shuff1x(xp, 1), _mm_set_ps(-0.f, -0.f, 0.f, 0.f));
+  x0 = _mm_xor_ps(glmm_splat(xp, 1), _mm_set_ps(-0.f, -0.f, 0.f, 0.f));
   r  = _mm_add_ps(r, _mm_mul_ps(x0, glmm_shuff1(xq, 1, 0, 3, 2)));
 
-  x0 = _mm_xor_ps(glmm_shuff1x(xp, 2), _mm_set_ps(-0.f, 0.f, 0.f, -0.f));
+  x0 = _mm_xor_ps(glmm_splat(xp, 2), _mm_set_ps(-0.f, 0.f, 0.f, -0.f));
   r  = _mm_add_ps(r, _mm_mul_ps(x0, glmm_shuff1(xq, 2, 3, 0, 1)));
 
   glmm_store(dest, r);

include/cglm/simd/x86.h

@@ -18,6 +18,9 @@
 #  define glmm_store(p, a)  _mm_store_ps(p, a)
 #endif
 
+#define glmm_set1(x) _mm_set1_ps(x)
+#define glmm_128     __m128
+
 #ifdef CGLM_USE_INT_DOMAIN
 #  define glmm_shuff1(xmm, z, y, x, w)                                        \
      _mm_castsi128_ps(_mm_shuffle_epi32(_mm_castps_si128(xmm),                \
@@ -27,7 +30,16 @@
        _mm_shuffle_ps(xmm, xmm, _MM_SHUFFLE(z, y, x, w))
 #endif
 
+#define glmm_splat(x, lane) glmm_shuff1(x, lane, lane, lane, lane)
+
+#define glmm_splat_x(x) glmm_splat(x, 0)
+#define glmm_splat_y(x) glmm_splat(x, 1)
+#define glmm_splat_z(x) glmm_splat(x, 2)
+#define glmm_splat_w(x) glmm_splat(x, 3)
+
+/* glmm_shuff1x() is DEPRECATED!, use glmm_splat() */
 #define glmm_shuff1x(xmm, x) glmm_shuff1(xmm, x, x, x, x)
+
 #define glmm_shuff2(a, b, z0, y0, x0, w0, z1, y1, x1, w1)                     \
      glmm_shuff1(_mm_shuffle_ps(a, b, _MM_SHUFFLE(z0, y0, x0, w0)),           \
                  z1, y1, x1, w1)
@@ -48,6 +60,15 @@ glmm_abs(__m128 x) {
   return _mm_andnot_ps(_mm_set1_ps(-0.0f), x);
 }
 
+static inline
+__m128
+glmm_vhadd(__m128 v) {
+  __m128 x0;
+  x0 = _mm_add_ps(v,  glmm_shuff1(v, 0, 1, 2, 3));
+  x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 0, 1));
+  return x0;
+}
+
 static inline
 __m128
 glmm_vhadds(__m128 v) {
@@ -80,7 +101,7 @@ glmm_vhmin(__m128 v) {
   __m128 x0, x1, x2;
   x0 = _mm_movehl_ps(v, v);     /* [2, 3, 2, 3] */
   x1 = _mm_min_ps(x0, v);       /* [0|2, 1|3, 2|2, 3|3] */
-  x2 = glmm_shuff1x(x1, 1);     /* [1|3, 1|3, 1|3, 1|3] */
+  x2 = glmm_splat(x1, 1);       /* [1|3, 1|3, 1|3, 1|3] */
   return _mm_min_ss(x1, x2);
 }
 
@@ -96,7 +117,7 @@ glmm_vhmax(__m128 v) {
   __m128 x0, x1, x2;
   x0 = _mm_movehl_ps(v, v);     /* [2, 3, 2, 3] */
   x1 = _mm_max_ps(x0, v);       /* [0|2, 1|3, 2|2, 3|3] */
-  x2 = glmm_shuff1x(x1, 1);     /* [1|3, 1|3, 1|3, 1|3] */
+  x2 = glmm_splat(x1, 1);       /* [1|3, 1|3, 1|3, 1|3] */
   return _mm_max_ss(x1, x2);
 }
 
@@ -175,7 +196,7 @@ glmm_load3(float v[3]) {
   __m128i xy;
   __m128  z;
 
-  xy = _mm_loadl_epi64((const __m128i *)v);
+  xy = _mm_loadl_epi64(CGLM_CASTPTR_ASSUME_ALIGNED(v, const __m128i));
   z  = _mm_load_ss(&v[2]);
 
   return _mm_movelh_ps(_mm_castsi128_ps(xy), z);
@@ -184,9 +205,103 @@ glmm_load3(float v[3]) {
 static inline
 void
 glmm_store3(float v[3], __m128 vx) {
-  _mm_storel_pi((__m64 *)&v[0], vx);
+  _mm_storel_pi(CGLM_CASTPTR_ASSUME_ALIGNED(v, __m64), vx);
   _mm_store_ss(&v[2], glmm_shuff1(vx, 2, 2, 2, 2));
 }
 
+static inline
+__m128
+glmm_div(__m128 a, __m128 b) {
+  return _mm_div_ps(a, b);
+}
+
+/* enable FMA macro for MSVC? */
+#if defined(_MSC_VER) && !defined(__FMA__) && defined(__AVX2__)
+#  define __FMA__ 1
+#endif
+
+static inline
+__m128
+glmm_fmadd(__m128 a, __m128 b, __m128 c) {
+#ifdef __FMA__
+  return _mm_fmadd_ps(a, b, c);
+#else
+  return _mm_add_ps(c, _mm_mul_ps(a, b));
+#endif
+}
+
+static inline
+__m128
+glmm_fnmadd(__m128 a, __m128 b, __m128 c) {
+#ifdef __FMA__
+  return _mm_fnmadd_ps(a, b, c);
+#else
+  return _mm_sub_ps(c, _mm_mul_ps(a, b));
+#endif
+}
+
+static inline
+__m128
+glmm_fmsub(__m128 a, __m128 b, __m128 c) {
+#ifdef __FMA__
+  return _mm_fmsub_ps(a, b, c);
+#else
+  return _mm_sub_ps(_mm_mul_ps(a, b), c);
+#endif
+}
+
+static inline
+__m128
+glmm_fnmsub(__m128 a, __m128 b, __m128 c) {
+#ifdef __FMA__
+  return _mm_fnmsub_ps(a, b, c);
+#else
+  return _mm_xor_ps(_mm_add_ps(_mm_mul_ps(a, b), c), _mm_set1_ps(-0.0f));
+#endif
+}
+
+#if defined(__AVX__)
+static inline
+__m256
+glmm256_fmadd(__m256 a, __m256 b, __m256 c) {
+#ifdef __FMA__
+  return _mm256_fmadd_ps(a, b, c);
+#else
+  return _mm256_add_ps(c, _mm256_mul_ps(a, b));
+#endif
+}
+
+static inline
+__m256
+glmm256_fnmadd(__m256 a, __m256 b, __m256 c) {
+#ifdef __FMA__
+  return _mm256_fnmadd_ps(a, b, c);
+#else
+  return _mm256_sub_ps(c, _mm256_mul_ps(a, b));
+#endif
+}
+
+static inline
+__m256
+glmm256_fmsub(__m256 a, __m256 b, __m256 c) {
+#ifdef __FMA__
+  return _mm256_fmsub_ps(a, b, c);
+#else
+  return _mm256_sub_ps(_mm256_mul_ps(a, b), c);
+#endif
+}
+
+static inline
+__m256
+glmm256_fnmsub(__m256 a, __m256 b, __m256 c) {
+#ifdef __FMA__
+  return _mm256_fmsub_ps(a, b, c);
+#else
+  return _mm256_xor_ps(_mm256_sub_ps(_mm256_mul_ps(a, b), c),
+                       _mm256_set1_ps(-0.0f));
+#endif
+}
+#endif
+
 #endif
 #endif /* cglm_simd_x86_h */

include/cglm/struct.h

@@ -31,6 +31,7 @@ extern "C" {
 #include "struct/project.h"
 #include "struct/sphere.h"
 #include "struct/curve.h"
+#include "struct/affine2d.h"
 
 #ifdef __cplusplus
 }

include/cglm/struct/affine.h

@@ -39,10 +39,6 @@
 #include "vec4.h"
 #include "mat4.h"
 
-CGLM_INLINE
-mat4s
-glms_mat4_mul(mat4s m1, mat4s m2);
-
 /*!
  * @brief translate existing transform matrix by v vector
  *        and stores result in same matrix

include/cglm/struct/affine2d.h

@@ -0,0 +1,177 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+/*
+ Functions:
+   CGLM_INLINE mat3s glms_translate2d(mat3 m, vec2 v)
+   CGLM_INLINE mat3s glms_translate2d_x(mat3s m, float x)
+   CGLM_INLINE mat3s glms_translate2d_y(mat3s m, float y)
+   CGLM_INLINE mat3s glms_translate2d_make(vec2s v)
+   CGLM_INLINE mat3s glms_scale2d_make(vec2s v)
+   CGLM_INLINE mat3s glms_scale2d(mat3s m, vec2s v)
+   CGLM_INLINE mat3s glms_scale2d_uni(mat3s m, float s)
+   CGLM_INLINE mat3s glms_rotate2d_make(float angle)
+   CGLM_INLINE mat3s glms_rotate2d(mat3s m, float angle)
+   CGLM_INLINE mat3s glms_rotate2d_to(mat3s m, float angle)
+ */
+
+#ifndef cglms_affine2ds_h
+#define cglms_affine2ds_h
+
+#include "../common.h"
+#include "../types-struct.h"
+#include "../affine2d.h"
+#include "vec3.h"
+#include "mat3.h"
+
+/*!
+ * @brief translate existing 2d transform matrix by v vector
+ *        and stores result in same matrix
+ *
+ * @param[in] m  affine transfrom
+ * @param[in] v  translate vector [x, y]
+ * @returns      affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_translate2d(mat3s m, vec2s v) {
+  glm_translate2d(m.raw, v.raw);
+  return m;
+}
+
+/*!
+ * @brief translate existing 2d transform matrix by x factor
+ *
+ * @param[in] m  affine transfrom
+ * @param[in] x  x factor
+ * @returns      affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_translate2d_x(mat3s m, float x) {
+  glm_translate2d_x(m.raw, x);
+  return m;
+}
+
+/*!
+ * @brief translate existing 2d transform matrix by y factor
+ *
+ * @param[in] m  affine transfrom
+ * @param[in] y  y factor
+ * @returns      affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_translate2d_y(mat3s m, float y) {
+  glm_translate2d_y(m.raw, y);
+  return m;
+}
+
+/*!
+ * @brief creates NEW translate 2d transform matrix by v vector
+ *
+ * @param[in] v  translate vector [x, y]
+ * @returns      affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_translate2d_make(vec2s v) {
+  mat3s m;
+  glm_translate2d_make(m.raw, v.raw);
+  return m;
+}
+
+/*!
+ * @brief creates NEW 2d scale matrix by v vector
+ *
+ * @param[in]   v  scale vector [x, y]
+ * @returns affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_scale2d_make(vec2s v) {
+  mat3s m;
+  glm_scale2d_make(m.raw, v.raw);
+  return m;
+}
+
+/*!
+ * @brief scales existing 2d transform matrix by v vector
+ *        and stores result in same matrix
+ *
+ * @param[in]  m  affine transfrom
+ * @param[in]  v  scale vector [x, y, z]
+ * @returns       affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_scale2d(mat3s m, vec2s v) {
+  mat3s r;
+  glm_scale2d_to(m.raw, v.raw, r.raw);
+  return r;
+}
+
+/*!
+ * @brief applies uniform scale to existing 2d transform matrix v = [s, s, s]
+ *        and stores result in same matrix
+ *
+ * @param[in] m  affine transfrom
+ * @param[in] s  scale factor
+ * @returns      affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_scale2d_uni(mat3s m, float s) {
+  glm_scale2d_uni(m.raw, s);
+  return m;
+}
+
+/*!
+ * @brief creates NEW 2d rotation matrix by angle and axis
+ *
+ * axis will be normalized so you don't need to normalize it
+ *
+ * @param[in]  angle  angle (radians)
+ * @returns           affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_rotate2d_make(float angle) {
+  mat3s m;
+  glm_rotate2d_make(m.raw, angle);
+  return m;
+}
+
+/*!
+ * @brief rotate existing 2d transform matrix around given axis by angle
+ *
+ * @param[in] m      affine transfrom
+ * @param[in] angle  angle (radians)
+ * @returns          affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_rotate2d(mat3s m, float angle) {
+  glm_rotate2d(m.raw, angle);
+  return m;
+}
+
+/*!
+ * @brief rotate existing 2d transform matrix around given axis by angle
+ *
+ * @param[in] m      affine transfrom
+ * @param[in] angle  angle (radians)
+ * @returns          affine transfrom
+ */
+CGLM_INLINE
+mat3s
+glms_rotate2d_to(mat3s m, float angle) {
+  glm_rotate2d(m.raw, angle);
+  return m;
+}
+
+#endif /* cglms_affine2ds_h */

include/cglm/struct/mat2.h

@@ -36,8 +36,8 @@
 #include "../types-struct.h"
 #include "../mat2.h"
 
-#define GLMS_MAT2_IDENTITY_INIT {1.0f, 0.0f, 0.0f, 1.0f}
-#define GLMS_MAT2_ZERO_INIT     {0.0f, 0.0f, 0.0f, 0.0f}
+#define GLMS_MAT2_IDENTITY_INIT {GLM_MAT2_IDENTITY_INIT}
+#define GLMS_MAT2_ZERO_INIT     {GLM_MAT2_ZERO_INIT}
 
 /* for C only */
 #define GLMS_MAT2_IDENTITY ((mat3s)GLMS_MAT2_IDENTITY_INIT)

include/cglm/struct/sphere.h

@@ -42,9 +42,9 @@ glms_sphere_radii(vec4s s) {
  */
 CGLM_INLINE
 vec4s
-glms_sphere_transform(vec4s s, mat4 m) {
+glms_sphere_transform(vec4s s, mat4s m) {
   vec4s r;
-  glm_sphere_transform(s.raw, m, r.raw);
+  glm_sphere_transform(s.raw, m.raw, r.raw);
   return r;
 }
 

include/cglm/struct/vec2.h

@@ -66,7 +66,7 @@
 /*!
  * @brief init vec2 using vec2
  *
- * @param[in]  v4   vector3
+ * @param[in]  v3   vector3
  * @returns         destination
  */
 CGLM_INLINE
@@ -455,18 +455,18 @@ glms_vec2_normalize(vec2s v) {
 }
 
 /*!
- * @brief rotate vec2 around axis by angle using Rodrigues' rotation formula
+ * @brief rotate vec2 by angle using Rodrigues' rotation formula
  *
  * @param[in]     v     vector
- * @param[in]     axis  axis vector (must be unit vector)
  * @param[in]     angle angle by radians
  * @returns             rotated vector
  */
 CGLM_INLINE
 vec2s
-glms_vec2_rotate(vec2s v, float angle, vec2s axis) {
-  glm_vec2_rotate(v.raw, angle, axis.raw);
-  return v;
+glms_vec2_rotate(vec2s v, float angle) {
+  vec2s r;
+  glm_vec2_rotate(v.raw, angle, r.raw);
+  return r;
 }
 
 /**

include/cglm/types.h

@@ -32,6 +32,16 @@
 #  define CGLM_ALIGN_MAT CGLM_ALIGN(16)
 #endif
 
+#ifdef __GNUC__
+#  define CGLM_ASSUME_ALIGNED(expr, alignment) \
+  __builtin_assume_aligned((expr), (alignment))
+#else
+#  define CGLM_ASSUME_ALIGNED(expr, alignment) (expr)
+#endif
+
+#define CGLM_CASTPTR_ASSUME_ALIGNED(expr, type) \
+  ((type*)CGLM_ASSUME_ALIGNED((expr), __alignof__(type)))
+
 typedef float                   vec2[2];
 typedef float                   vec3[3];
 typedef int                    ivec3[3];

include/cglm/util.h

@@ -289,7 +289,7 @@ glm_smoothinterpc(float from, float to, float t) {
 CGLM_INLINE
 bool
 glm_eq(float a, float b) {
-  return fabsf(a - b) <= FLT_EPSILON;
+  return fabsf(a - b) <= GLM_FLT_EPSILON;
 }
 
 /*!
@@ -322,7 +322,22 @@ glm_percent(float from, float to, float current) {
 CGLM_INLINE
 float
 glm_percentc(float from, float to, float current) {
-  return glm_clamp(glm_percent(from, to, current), 0.0f, 1.0f);
+  return glm_clamp_zo(glm_percent(from, to, current));
+}
+
+/*!
+* @brief swap two float values
+*
+* @param[in]   a float value 1 (pointer)
+* @param[in]   b float value 2 (pointer)
+*/
+CGLM_INLINE
+void
+glm_swapf(float * __restrict a, float * __restrict b) {
+  float t;
+  t  = *a;
+  *a = *b;
+  *b = t;
 }
 
 #endif /* cglm_util_h */

include/cglm/vec2-ext.h

@@ -61,8 +61,8 @@ glm_vec2_eq(vec2 v, float val) {
 CGLM_INLINE
 bool
 glm_vec2_eq_eps(vec2 v, float val) {
-  return fabsf(v[0] - val) <= FLT_EPSILON
-         && fabsf(v[1] - val) <= FLT_EPSILON;
+  return fabsf(v[0] - val) <= GLM_FLT_EPSILON
+         && fabsf(v[1] - val) <= GLM_FLT_EPSILON;
 }
 
 /*!
@@ -73,7 +73,7 @@ glm_vec2_eq_eps(vec2 v, float val) {
 CGLM_INLINE
 bool
 glm_vec2_eq_all(vec2 v) {
-  return v[0] == v[1];
+  return glm_vec2_eq_eps(v, v[0]);
 }
 
 /*!
@@ -97,8 +97,8 @@ glm_vec2_eqv(vec2 a, vec2 b) {
 CGLM_INLINE
 bool
 glm_vec2_eqv_eps(vec2 a, vec2 b) {
-  return fabsf(a[0] - b[0]) <= FLT_EPSILON
-         && fabsf(a[1] - b[1]) <= FLT_EPSILON;
+  return fabsf(a[0] - b[0]) <= GLM_FLT_EPSILON
+         && fabsf(a[1] - b[1]) <= GLM_FLT_EPSILON;
 }
 
 /*!

include/cglm/vec3-ext.h

@@ -81,9 +81,9 @@ glm_vec3_eq(vec3 v, float val) {
 CGLM_INLINE
 bool
 glm_vec3_eq_eps(vec3 v, float val) {
-  return fabsf(v[0] - val) <= FLT_EPSILON
-         && fabsf(v[1] - val) <= FLT_EPSILON
-         && fabsf(v[2] - val) <= FLT_EPSILON;
+  return fabsf(v[0] - val) <= GLM_FLT_EPSILON
+         && fabsf(v[1] - val) <= GLM_FLT_EPSILON
+         && fabsf(v[2] - val) <= GLM_FLT_EPSILON;
 }
 
 /*!
@@ -94,7 +94,7 @@ glm_vec3_eq_eps(vec3 v, float val) {
 CGLM_INLINE
 bool
 glm_vec3_eq_all(vec3 v) {
-  return v[0] == v[1] && v[0] == v[2];
+  return glm_vec3_eq_eps(v, v[0]);
 }
 
 /*!
@@ -120,9 +120,9 @@ glm_vec3_eqv(vec3 a, vec3 b) {
 CGLM_INLINE
 bool
 glm_vec3_eqv_eps(vec3 a, vec3 b) {
-  return fabsf(a[0] - b[0]) <= FLT_EPSILON
-         && fabsf(a[1] - b[1]) <= FLT_EPSILON
-         && fabsf(a[2] - b[2]) <= FLT_EPSILON;
+  return fabsf(a[0] - b[0]) <= GLM_FLT_EPSILON
+         && fabsf(a[1] - b[1]) <= GLM_FLT_EPSILON
+         && fabsf(a[2] - b[2]) <= GLM_FLT_EPSILON;
 }
 
 /*!
@@ -237,9 +237,9 @@ glm_vec3_abs(vec3 v, vec3 dest) {
 CGLM_INLINE
 void
 glm_vec3_fract(vec3 v, vec3 dest) {
-  dest[0] = fminf(v[0] - floorf(v[0]), 0x1.fffffep-1f);
-  dest[1] = fminf(v[1] - floorf(v[1]), 0x1.fffffep-1f);
-  dest[2] = fminf(v[2] - floorf(v[2]), 0x1.fffffep-1f);
+  dest[0] = fminf(v[0] - floorf(v[0]), 0.999999940395355224609375f);
+  dest[1] = fminf(v[1] - floorf(v[1]), 0.999999940395355224609375f);
+  dest[2] = fminf(v[2] - floorf(v[2]), 0.999999940395355224609375f);
 }
 
 /*!

include/cglm/vec4-ext.h

@@ -92,10 +92,10 @@ glm_vec4_eq(vec4 v, float val) {
 CGLM_INLINE
 bool
 glm_vec4_eq_eps(vec4 v, float val) {
-  return fabsf(v[0] - val) <= FLT_EPSILON
-         && fabsf(v[1] - val) <= FLT_EPSILON
-         && fabsf(v[2] - val) <= FLT_EPSILON
-         && fabsf(v[3] - val) <= FLT_EPSILON;
+  return fabsf(v[0] - val) <= GLM_FLT_EPSILON
+         && fabsf(v[1] - val) <= GLM_FLT_EPSILON
+         && fabsf(v[2] - val) <= GLM_FLT_EPSILON
+         && fabsf(v[3] - val) <= GLM_FLT_EPSILON;
 }
 
 /*!
@@ -106,9 +106,7 @@ glm_vec4_eq_eps(vec4 v, float val) {
 CGLM_INLINE
 bool
 glm_vec4_eq_all(vec4 v) {
-  return v[0] == v[1]
-         && v[0] == v[2]
-         && v[0] == v[3];
+  return glm_vec4_eq_eps(v, v[0]);
 }
 
 /*!
@@ -135,10 +133,10 @@ glm_vec4_eqv(vec4 a, vec4 b) {
 CGLM_INLINE
 bool
 glm_vec4_eqv_eps(vec4 a, vec4 b) {
-  return fabsf(a[0] - b[0]) <= FLT_EPSILON
-         && fabsf(a[1] - b[1]) <= FLT_EPSILON
-         && fabsf(a[2] - b[2]) <= FLT_EPSILON
-         && fabsf(a[3] - b[3]) <= FLT_EPSILON;
+  return fabsf(a[0] - b[0]) <= GLM_FLT_EPSILON
+         && fabsf(a[1] - b[1]) <= GLM_FLT_EPSILON
+         && fabsf(a[2] - b[2]) <= GLM_FLT_EPSILON
+         && fabsf(a[3] - b[3]) <= GLM_FLT_EPSILON;
 }
 
 /*!
@@ -226,10 +224,10 @@ glm_vec4_sign(vec4 v, vec4 dest) {
 
   x0 = glmm_load(v);
   x1 = _mm_set_ps(0.0f, 0.0f, 1.0f, -1.0f);
-  x2 = glmm_shuff1x(x1, 2);
+  x2 = glmm_splat(x1, 2);
 
-  x3 = _mm_and_ps(_mm_cmpgt_ps(x0, x2), glmm_shuff1x(x1, 1));
-  x4 = _mm_and_ps(_mm_cmplt_ps(x0, x2), glmm_shuff1x(x1, 0));
+  x3 = _mm_and_ps(_mm_cmpgt_ps(x0, x2), glmm_splat(x1, 1));
+  x4 = _mm_and_ps(_mm_cmplt_ps(x0, x2), glmm_splat(x1, 0));
 
   glmm_store(dest, _mm_or_ps(x3, x4));
 #else
@@ -270,10 +268,10 @@ glm_vec4_abs(vec4 v, vec4 dest) {
 CGLM_INLINE
 void
 glm_vec4_fract(vec4 v, vec4 dest) {
-  dest[0] = fminf(v[0] - floorf(v[0]), 0x1.fffffep-1f);
-  dest[1] = fminf(v[1] - floorf(v[1]), 0x1.fffffep-1f);
-  dest[2] = fminf(v[2] - floorf(v[2]), 0x1.fffffep-1f);
-  dest[3] = fminf(v[3] - floorf(v[3]), 0x1.fffffep-1f);
+  dest[0] = fminf(v[0] - floorf(v[0]), 0.999999940395355224609375f);
+  dest[1] = fminf(v[1] - floorf(v[1]), 0.999999940395355224609375f);
+  dest[2] = fminf(v[2] - floorf(v[2]), 0.999999940395355224609375f);
+  dest[3] = fminf(v[3] - floorf(v[3]), 0.999999940395355224609375f);
 }
 
 /*!

include/cglm/vec4.h

@@ -473,8 +473,8 @@ glm_vec4_scale_as(vec4 v, float s, vec4 dest) {
 CGLM_INLINE
 void
 glm_vec4_div(vec4 a, vec4 b, vec4 dest) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
-  glmm_store(dest, _mm_div_ps(glmm_load(a), glmm_load(b)));
+#if defined(CGLM_SIMD)
+  glmm_store(dest, glmm_div(glmm_load(a), glmm_load(b)));
 #else
   dest[0] = a[0] / b[0];
   dest[1] = a[1] / b[1];
@@ -568,14 +568,8 @@ glm_vec4_subadd(vec4 a, vec4 b, vec4 dest) {
 CGLM_INLINE
 void
 glm_vec4_muladd(vec4 a, vec4 b, vec4 dest) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
-  glmm_store(dest, _mm_add_ps(glmm_load(dest),
-                              _mm_mul_ps(glmm_load(a),
-                                         glmm_load(b))));
-#elif defined(CGLM_NEON_FP)
-  vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
-                            vmulq_f32(vld1q_f32(a),
-                                      vld1q_f32(b))));
+#if defined(CGLM_SIMD)
+  glmm_store(dest, glmm_fmadd(glmm_load(a), glmm_load(b), glmm_load(dest)));
 #else
   dest[0] += a[0] * b[0];
   dest[1] += a[1] * b[1];
@@ -596,14 +590,8 @@ glm_vec4_muladd(vec4 a, vec4 b, vec4 dest) {
 CGLM_INLINE
 void
 glm_vec4_muladds(vec4 a, float s, vec4 dest) {
-#if defined( __SSE__ ) || defined( __SSE2__ )
-  glmm_store(dest, _mm_add_ps(glmm_load(dest),
-                              _mm_mul_ps(glmm_load(a),
-                                         _mm_set1_ps(s))));
-#elif defined(CGLM_NEON_FP)
-  vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
-                            vsubq_f32(vld1q_f32(a),
-                                      vdupq_n_f32(s))));
+#if defined(CGLM_SIMD)
+  glmm_store(dest, glmm_fmadd(glmm_load(a), glmm_set1(s), glmm_load(dest)));
 #else
   dest[0] += a[0] * s;
   dest[1] += a[1] * s;
@@ -680,7 +668,7 @@ glm_vec4_negate_to(vec4 v, vec4 dest) {
 #if defined( __SSE__ ) || defined( __SSE2__ )
   glmm_store(dest, _mm_xor_ps(glmm_load(v), _mm_set1_ps(-0.0f)));
 #elif defined(CGLM_NEON_FP)
-  vst1q_f32(dest, veorq_s32(vld1q_f32(v), vdupq_n_f32(-0.0f)));
+  vst1q_f32(dest, vnegq_f32(vld1q_f32(v)));
 #else
   dest[0] = -v[0];
   dest[1] = -v[1];

include/cglm/version.h

@@ -9,7 +9,7 @@
 #define cglm_version_h
 
 #define CGLM_VERSION_MAJOR 0
-#define CGLM_VERSION_MINOR 7
-#define CGLM_VERSION_PATCH 0
+#define CGLM_VERSION_MINOR 8
+#define CGLM_VERSION_PATCH 2
 
 #endif /* cglm_version_h */

include/module.modulemap

@@ -0,0 +1,14 @@
+module cglm {
+    header "cglm/cglm.h"
+    header "cglm/struct.h"
+    
+    export *
+}
+
+module cglmc {
+    header "cglm/cglm.h"
+    header "cglm/struct.h"
+    header "cglm/call.h"
+    
+    export *
+}

meson.build

@@ -0,0 +1,108 @@
+project('cglm', 'c',
+    version : '0.8.2', 
+    license : 'mit',
+    default_options : [
+        'c_std=c11',
+        'werror=true',
+        'warning_level=2',
+        'buildtype=release'
+    ]
+)
+
+cc = meson.get_compiler('c')
+
+cglm_install = get_option('install')
+cglm_deps = cc.find_library('m', required : false)
+
+cglm_args = []
+build_args = []
+
+if get_option('default_library') == 'static'
+    cglm_args = '-DCGLM_STATIC'
+endif
+
+if host_machine.system() == 'windows'
+    build_args = '-DCGLM_EXPORTS'
+endif
+
+cglm_inc = include_directories('include')
+
+cglm_src = files(
+    'src/affine.c',
+    'src/affine2d.c',
+    'src/bezier.c',
+    'src/box.c',
+    'src/cam.c',
+    'src/curve.c',
+    'src/ease.c',
+    'src/euler.c',
+    'src/frustum.c',
+    'src/io.c',
+    'src/mat2.c',
+    'src/mat3.c',
+    'src/mat4.c',
+    'src/plane.c',
+    'src/project.c',
+    'src/quat.c',
+    'src/ray.c',
+    'src/sphere.c',
+    'src/vec2.c',
+    'src/vec3.c',
+    'src/vec4.c'
+)
+
+cglm_lib = library('cglm',
+    cglm_src,
+    install : cglm_install,
+    dependencies : cglm_deps,
+    c_args : [ build_args, cglm_args ]
+)
+
+cglm_dep = declare_dependency(
+    link_with : cglm_lib,
+    dependencies : cglm_deps,
+    compile_args : cglm_args,
+    include_directories : cglm_inc,
+    version : meson.project_version()
+)
+
+if meson.version().version_compare('>= 0.54.0')
+    meson.override_dependency('cglm', cglm_dep)
+endif
+
+if cglm_install
+    install_subdir('include/cglm', install_dir : get_option('includedir'))
+
+    pkg = import('pkgconfig')
+    pkg.generate(
+        name : 'cglm',
+        libraries : cglm_lib,
+        extra_cflags : cglm_args,
+        version : meson.project_version(),
+        url : 'https://github.com/recp/cglm',
+        description : 'OpenGL Mathematics (glm) for C'
+    )
+endif
+
+if get_option('build_tests') == true
+
+test_src = files(
+    'test/runner.c',
+    'test/src/test_euler.c',
+    'test/src/test_bezier.c',
+    'test/src/test_cam.c',
+    'test/src/test_struct.c',
+    'test/src/test_clamp.c',
+    'test/src/test_common.c',
+    'test/src/tests.c'
+)
+
+test_exe = executable('tests', 
+    test_src, 
+    dependencies : cglm_dep,
+    c_args : '-DGLM_TESTS_NO_COLORFUL_OUTPUT'
+)
+
+test('cglm.tests', test_exe)
+
+endif

meson_options.txt

@@ -0,0 +1,2 @@
+option('build_tests', type : 'boolean', value : false, description : 'Build tests')
+option('install', type : 'boolean', value : true, description : 'Include the library, headers, and pkg-config file in the install target')

src/affine2d.c

@@ -0,0 +1,81 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#include "../include/cglm/cglm.h"
+#include "../include/cglm/call.h"
+
+CGLM_EXPORT
+void
+glmc_translate2d_make(mat3 m, vec2 v) {
+  glm_translate2d_make(m, v);
+}
+
+CGLM_EXPORT
+void
+glmc_translate2d_to(mat3 m, vec2 v, mat3 dest) {
+  glm_translate2d_to(m, v, dest);
+}
+
+CGLM_EXPORT
+void
+glmc_translate2d(mat3 m, vec2 v) {
+  glm_translate2d(m, v);
+}
+
+CGLM_EXPORT
+void
+glmc_translate2d_x(mat3 m, float to) {
+  glm_translate2d_x(m, to);
+}
+
+CGLM_EXPORT
+void
+glmc_translate2d_y(mat3 m, float to) {
+  glm_translate2d_y(m, to);
+}
+
+CGLM_EXPORT
+void
+glmc_scale2d_to(mat3 m, vec2 v, mat3 dest) {
+  glm_scale2d_to(m, v, dest);
+}
+
+CGLM_EXPORT
+void
+glmc_scale2d_make(mat3 m, vec2 v) {
+  glm_scale2d_make(m, v);
+}
+
+CGLM_EXPORT
+void
+glmc_scale2d(mat3 m, vec2 v) {
+  glm_scale2d(m, v);
+}
+
+CGLM_EXPORT
+void
+glmc_scale2d_uni(mat3 m, float s) {
+  glm_scale2d_uni(m, s);
+}
+
+CGLM_EXPORT
+void
+glmc_rotate2d_make(mat3 m, float angle) {
+  glm_rotate2d_make(m, angle);
+}
+
+CGLM_EXPORT
+void
+glmc_rotate2d(mat3 m, float angle) {
+  glm_rotate2d(m, angle);
+}
+
+CGLM_EXPORT
+void
+glmc_rotate2d_to(mat3 m, float angle, mat3 dest) {
+  glm_rotate2d_to(m, angle, dest);
+}

src/config.h

@@ -1,9 +1,9 @@
 /*
-* Copyright (c), Recep Aslantas.
-*
-* MIT License (MIT), http://opensource.org/licenses/MIT
-* Full license can be found in the LICENSE file
-*/
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
 
 #ifndef cglm__config__h_
 #define cglm__config__h_

src/io.c

@@ -5,6 +5,8 @@
  * Full license can be found in the LICENSE file
  */
 
+#define CGLM_LIB_SRC
+
 #include "../include/cglm/cglm.h"
 #include "../include/cglm/call.h"
 

src/ray.c

@@ -0,0 +1,13 @@
+#include "../include/cglm/cglm.h"
+#include "../include/cglm/call.h"
+
+CGLM_EXPORT
+bool
+glmc_ray_triangle(vec3   origin,
+                  vec3   direction,
+                  vec3   v0,
+                  vec3   v1,
+                  vec3   v2,
+                  float *d) {
+    return glm_ray_triangle(origin, direction, v0, v1, v2, d);
+}

src/swift/empty.c

@@ -0,0 +1 @@
+// This empty file is needed to trick swiftpm to build the header-only version of cglm as swiftpm itself does not support C targets that have no source code files

test/CMakeLists.txt

@@ -0,0 +1,44 @@
+cmake_minimum_required(VERSION 3.8.2)
+
+# List all files containing tests. (Change as needed)
+set(TESTFILES
+  runner.c
+  src/test_euler.c
+  src/test_bezier.c
+  src/test_cam.c
+  src/test_struct.c
+  src/test_clamp.c
+  src/test_common.c
+  src/tests.c
+  )
+
+set(TEST_MAIN tests)
+set(TEST_RUNNER_PARAMS "")
+
+add_executable(${TEST_MAIN} ${TESTFILES})
+target_compile_definitions(${TEST_MAIN} PRIVATE CGLM_DEFINE_PRINTS=1)
+
+if(NOT MSVC)
+  target_link_libraries(${TEST_MAIN} PRIVATE m)
+endif()
+
+target_link_libraries(${TEST_MAIN} PRIVATE cglm)
+target_include_directories(${TEST_MAIN} PRIVATE
+  ${CMAKE_CURRENT_LIST_DIR}/include
+  ${CMAKE_CURRENT_LIST_DIR}/src
+  )
+
+set_target_properties(${TEST_MAIN} PROPERTIES RUNTIME_OUTPUT_DIRECTORY ${PROJECT_BINARY_DIR})
+
+if(LDFLAGS)
+  target_compile_options(${TEST_MAIN} PRIVATE ${LDFLAGS})
+endif()
+
+add_test(
+    NAME cglm.${TEST_MAIN}
+    COMMAND ${TEST_MAIN} ${TEST_RUNNER_PARAMS})
+
+add_custom_target(check
+  make
+  COMMAND ${CMAKE_CTEST_COMMAND} -V
+  DEPENDS cglm)

test/include/common.h

@@ -8,6 +8,18 @@
 #ifndef tests_common_h
 #define tests_common_h
 
+#ifndef _USE_MATH_DEFINES
+#  define _USE_MATH_DEFINES       /* for windows */
+#endif
+
+#ifndef _CRT_SECURE_NO_WARNINGS
+#  define _CRT_SECURE_NO_WARNINGS /* for windows */
+#endif
+
+#ifndef _GNU_SOURCE
+#  define _GNU_SOURCE /* for drand48() */
+#endif
+
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
@@ -30,6 +42,8 @@ typedef struct test_entry_t {
   int    show_output;
 } test_entry_t;
 
+#ifndef GLM_TESTS_NO_COLORFUL_OUTPUT
+
 #define RESET       "\033[0m"
 #define BLACK       "\033[30m"             /* Black */
 #define RED         "\033[31m"             /* Red */
@@ -48,6 +62,28 @@ typedef struct test_entry_t {
 #define BOLDCYAN    "\033[1m\033[36m"      /* Bold Cyan */
 #define BOLDWHITE   "\033[1m\033[37m"      /* Bold White */
 
+#else
+
+#define RESET
+#define BLACK
+#define RED
+#define GREEN
+#define YELLOW
+#define BLUE
+#define MAGENTA
+#define CYAN
+#define WHITE
+#define BOLDBLACK
+#define BOLDRED
+#define BOLDGREEN
+#define BOLDYELLOW
+#define BOLDBLUE
+#define BOLDMAGENTA
+#define BOLDCYAN
+#define BOLDWHITE
+
+#endif
+
 #define TEST_DECLARE(FUN) test_status_t test_ ## FUN(void);
 #define TEST_ENTRY(FUN)   { #FUN, test_ ## FUN, 0, 0 },
 #define TEST_LIST         static test_entry_t tests[] = 

test/runner.c

@@ -18,6 +18,8 @@ main(int argc, const char * argv[]) {
   test_status_t st;
   int32_t       i, count, passed, failed, maxlen;
   double        start, end, elapsed, total;
+  (void)argc;
+  (void)argv;
 
   passed = failed = maxlen  = 0;
   total  = 0.0;

test/src/test_affine2d.h

@@ -0,0 +1,310 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#include "test_common.h"
+
+TEST_IMPL(GLM_PREFIX, translate2d) {
+  mat3 m1;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  glm_mat3_identity(m1);
+  GLM(translate2d)(m1, (vec2){13.0f, 11.0f});
+  glm_mat3_mulv(m1, v1, v2);
+
+  ASSERT(test_eq(v2[0], 15.0f))
+  ASSERT(test_eq(v2[1], 14.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  glm_mat3_identity(m1);
+  GLM(translate2d)(m1, (vec2){1.0f, -1.0f});
+  glm_mat3_mulv(m1, v2, v2);
+
+  ASSERT(test_eq(v2[0], 16.0f))
+  ASSERT(test_eq(v2[1], 13.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, translate2d_to) {
+  mat3 m1, m2;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  glm_mat3_identity(m1);
+  GLM(translate2d_to)(m1, (vec2){13.0f, 11.0f}, m2);
+  glm_mat3_mulv(m2, v1, v2);
+
+  ASSERT(test_eq(v2[0], 15.0f))
+  ASSERT(test_eq(v2[1], 14.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  glm_mat3_identity(m1);
+  GLM(translate2d_to)(m1, (vec2){1.0f, -1.0f}, m2);
+  glm_mat3_mulv(m2, v2, v2);
+
+  ASSERT(test_eq(v2[0], 16.0f))
+  ASSERT(test_eq(v2[1], 13.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, translate2d_x) {
+  mat3 m1;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  glm_mat3_identity(m1);
+  GLM(translate2d_x)(m1, 13.0f);
+  glm_mat3_mulv(m1, v1, v2);
+
+  ASSERT(test_eq(v2[0], 15.0f))
+  ASSERT(test_eq(v2[1], 3.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  glm_mat3_identity(m1);
+  GLM(translate2d_x)(m1, -1.0f);
+  glm_mat3_mulv(m1, v2, v2);
+
+  ASSERT(test_eq(v2[0], 14.0f))
+  ASSERT(test_eq(v2[1], 3.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, translate2d_y) {
+  mat3 m1;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  glm_mat3_identity(m1);
+  GLM(translate2d_y)(m1, 11.0f);
+  glm_mat3_mulv(m1, v1, v2);
+  
+
+  ASSERT(test_eq(v2[0], 2.0f))
+  ASSERT(test_eq(v2[1], 14.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  glm_mat3_identity(m1);
+  GLM(translate2d_y)(m1, -1.0f);
+  glm_mat3_mulv(m1, v2, v2);
+
+  ASSERT(test_eq(v2[0], 2.0f))
+  ASSERT(test_eq(v2[1], 13.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+  
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, translate2d_make) {
+  mat3 m1;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  glm_mat3_identity(m1);
+  GLM(translate2d_make)(m1, (vec2){13.0f, 11.0f});
+  glm_mat3_mulv(m1, v1, v2);
+
+  ASSERT(test_eq(v2[0], 15.0f))
+  ASSERT(test_eq(v2[1], 14.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  glm_mat3_identity(m1);
+  GLM(translate2d_make)(m1, (vec2){-1.0f, -5.0f});
+  glm_mat3_mulv(m1, v2, v2);
+
+  ASSERT(test_eq(v2[0], 14.0f))
+  ASSERT(test_eq(v2[1], 9.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, scale2d_to) {
+  mat3 m1, m2;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  glm_mat3_identity(m1);
+  GLM(scale2d_to)(m1, (vec2){13.0f, 11.0f}, m2);
+  glm_mat3_mulv(m2, v1, v2);
+
+  ASSERT(test_eq(v2[0], 26.0f))
+  ASSERT(test_eq(v2[1], 33.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  glm_mat3_identity(m1);
+  GLM(scale2d_to)(m1, (vec2){-1.0f, -5.0f}, m2);
+  glm_mat3_mulv(m2, v2, v2);
+
+  ASSERT(test_eq(v2[0], -26.0f))
+  ASSERT(test_eq(v2[1], -165.0f))
+  ASSERT(test_eq(v2[2],  1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, scale2d_make) {
+  mat3 m1;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  GLM(scale2d_make)(m1, (vec2){13.0f, 11.0f});
+  glm_mat3_mulv(m1, v1, v2);
+
+  ASSERT(test_eq(v2[0], 26.0f))
+  ASSERT(test_eq(v2[1], 33.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  GLM(scale2d_make)(m1, (vec3){-1.0f, -5.0f});
+  glm_mat3_mulv(m1, v2, v2);
+
+  ASSERT(test_eq(v2[0], -26.0f))
+  ASSERT(test_eq(v2[1], -165.0f))
+  ASSERT(test_eq(v2[2],  1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, scale2d) {
+  mat3 m1;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  glm_mat3_identity(m1);
+  GLM(scale2d)(m1, (vec2){13.0f, 11.0f});
+  glm_mat3_mulv(m1, v1, v2);
+
+  ASSERT(test_eq(v2[0], 26.0f))
+  ASSERT(test_eq(v2[1], 33.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  glm_mat3_identity(m1);
+  GLM(scale2d)(m1, (vec2){-1.0f, -5.0f});
+  glm_mat3_mulv(m1, v2, v2);
+
+  ASSERT(test_eq(v2[0], -26.0f))
+  ASSERT(test_eq(v2[1], -165.0f))
+  ASSERT(test_eq(v2[2],  1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, scale2d_uni) {
+  mat3 m1;
+  vec3 v1 = {2.0f, 3.0f, 1.0f}, v2;
+
+  glm_mat3_identity(m1);
+  GLM(scale2d_uni)(m1, 13.0f);
+  glm_mat3_mulv(m1, v1, v2);
+
+  ASSERT(test_eq(v2[0], 26.0f))
+  ASSERT(test_eq(v2[1], 39.0f))
+  ASSERT(test_eq(v2[2], 1.0f))
+
+  glm_mat3_identity(m1);
+  GLM(scale2d_uni)(m1, -5.0f);
+  glm_mat3_mulv(m1, v2, v2);
+
+  ASSERT(test_eq(v2[0], -130.0f))
+  ASSERT(test_eq(v2[1], -195.0f))
+  ASSERT(test_eq(v2[2],  1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, rotate2d_make) {
+  mat3 m1 = GLM_MAT3_IDENTITY_INIT;
+  vec3 v1 = {0.0f, 1.0f, 1.0f}, v2 = {0.0f, 1.0f, 1.0f};
+
+  GLM(rotate2d_make)(m1, GLM_PI_2f);
+  glm_mat3_mulv(m1, v1, v1);
+
+  ASSERT(test_eq(v1[0], -1.0f))
+  ASSERT(test_eq(v1[1],  0.0f))
+  ASSERT(test_eq(v1[2],  1.0f))
+  
+  glm_vec3_copy(v2, v1);
+
+  GLM(rotate2d_make)(m1, -GLM_PI_2f);
+  glm_mat3_mulv(m1, v1, v1);
+
+  ASSERT(test_eq(v1[0],  1.0f))
+  ASSERT(test_eq(v1[1],  0.0f))
+  ASSERT(test_eq(v1[2],  1.0f))
+
+  glm_vec3_copy(v2, v1);
+
+  GLM(rotate2d_make)(m1, GLM_PIf);
+  glm_mat3_mulv(m1, v1, v1);
+
+  ASSERT(test_eq(v1[0],  0.0f))
+  ASSERT(test_eq(v1[1], -1.0f))
+  ASSERT(test_eq(v1[2],  1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, rotate2d) {
+  mat3 m1 = GLM_MAT3_IDENTITY_INIT;
+  vec3 v1 = {0.0f, 1.0f, 1.0f}, v2 = {0.0f, 1.0f, 1.0f};
+
+  GLM(rotate2d)(m1, GLM_PI_2f);
+  glm_mat3_mulv(m1, v1, v1);
+
+  ASSERT(test_eq(v1[0], -1.0f))
+  ASSERT(test_eq(v1[1],  0.0f))
+  ASSERT(test_eq(v1[2],  1.0f))
+  
+  glm_vec3_copy(v2, v1);
+
+  GLM(rotate2d)(m1, GLM_PI_2f);
+  glm_mat3_mulv(m1, v1, v1);
+
+  ASSERT(test_eq(v1[0],  0.0f))
+  ASSERT(test_eq(v1[1], -1.0f))
+  ASSERT(test_eq(v1[2],  1.0f))
+
+  glm_vec3_copy(v2, v1);
+
+  GLM(rotate2d)(m1, GLM_PI_2f);
+  glm_mat3_mulv(m1, v1, v1);
+
+  ASSERT(test_eq(v1[0], 1.0f))
+  ASSERT(test_eq(v1[1], 0.0f))
+  ASSERT(test_eq(v1[2], 1.0f))
+
+  TEST_SUCCESS
+}
+
+TEST_IMPL(GLM_PREFIX, rotate2d_to) {
+  mat3 m1 = GLM_MAT3_IDENTITY_INIT, m2;
+  vec3 v1 = {0.0f, 1.0f, 1.0f}, v2 = {0.0f, 1.0f, 1.0f};
+
+  GLM(rotate2d_to)(m1, GLM_PI_2f, m1);
+  glm_mat3_mulv(m1, v1, v1);
+
+  ASSERT(test_eq(v1[0], -1.0f))
+  ASSERT(test_eq(v1[1],  0.0f))
+  ASSERT(test_eq(v1[2],  1.0f))
+  
+  glm_vec3_copy(v2, v1);
+
+  GLM(rotate2d_to)(m1, GLM_PI_2f, m2);
+  glm_mat3_mulv(m2, v1, v1);
+
+  ASSERT(test_eq(v1[0],  0.0f))
+  ASSERT(test_eq(v1[1], -1.0f))
+  ASSERT(test_eq(v1[2],  1.0f))
+
+  glm_vec3_copy(v2, v1);
+
+  GLM(rotate2d_to)(m2, GLM_PI_2f, m1);
+  glm_mat3_mulv(m1, v1, v1);
+
+  ASSERT(test_eq(v1[0], 1.0f))
+  ASSERT(test_eq(v1[1], 0.0f))
+  ASSERT(test_eq(v1[2], 1.0f))
+
+  TEST_SUCCESS
+}

test/src/test_cam.c

@@ -35,9 +35,9 @@ TEST_IMPL(camera_decomp) {
   farVal  = 100.0f;
 
   glm_perspective(fovy, aspect, nearVal, farVal, proj);
-  ASSERT(fabsf(aspect  - glm_persp_aspect(proj)) < FLT_EPSILON)
-  ASSERT(fabsf(fovy    - glm_persp_fovy(proj))   < FLT_EPSILON)
-  ASSERT(fabsf(49.984f - glm_deg(glm_persp_fovy(proj))) < FLT_EPSILON)
+  ASSERT(fabsf(aspect  - glm_persp_aspect(proj)) < GLM_FLT_EPSILON)
+  ASSERT(fabsf(fovy    - glm_persp_fovy(proj))   < GLM_FLT_EPSILON)
+  ASSERT(fabsf(49.984f - glm_deg(glm_persp_fovy(proj))) < GLM_FLT_EPSILON)
 
   glm_persp_sizes(proj, fovy, sizes);
 

test/src/test_camera.h

@@ -0,0 +1,68 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#include "test_common.h"
+
+TEST_IMPL(GLM_PREFIX, frustum) {
+  mat4  proj;
+  vec4  vp = {0.0f, 0.0f, 800.0f, 600.0f};
+  float left, right, top, bottom, znear, zfar;
+  
+  znear  =  0.1f;
+  zfar   =  100.0f;
+  left   = -100.0f;
+  right  =  100.0f;
+  bottom = -100.0f;
+  top    =  100.0f;
+  
+  GLM(frustum)(left, right, bottom, top, znear, zfar, proj);
+  
+  ASSERT(test_eq(proj[0][1], 0.0f))
+  ASSERT(test_eq(proj[0][2], 0.0f))
+  ASSERT(test_eq(proj[0][3], 0.0f))
+
+  ASSERT(test_eq(proj[1][0], 0.0f))
+  ASSERT(test_eq(proj[1][2], 0.0f))
+  ASSERT(test_eq(proj[1][3], 0.0f))
+
+  ASSERT(test_eq(proj[2][3], -1.0f))
+
+  ASSERT(test_eq(proj[3][0], 0.0f))
+  ASSERT(test_eq(proj[3][1], 0.0f))
+  ASSERT(test_eq(proj[3][3], 0.0f))
+
+  vec4  v1 = {1.0f, 20.0f, znear};
+  vec4  v2 = {1.0f, 20.0f, zfar};
+  vec4  v3, v4;
+
+  /* perspective test */
+  GLM(mat4_mulv)(proj, v1, v3);
+  GLM(project)(v3, proj, vp, v3);
+  
+  ASSERT(v3[0] > v1[0])
+  ASSERT(v3[1] > v1[1])
+  
+  GLM(mat4_mulv)(proj, v2, v4);
+  GLM(project)(v4, proj, vp, v4);
+  
+  ASSERT(v4[0] < v3[0])
+  ASSERT(v4[1] < v3[1])
+  
+  /* not infinity */
+  ASSERT(!GLM(vec4_isinf)(proj[0]))
+  ASSERT(!GLM(vec4_isinf)(proj[1]))
+  ASSERT(!GLM(vec4_isinf)(proj[2]))
+  ASSERT(!GLM(vec4_isinf)(proj[3]))
+  
+  /* not NaN */
+  ASSERT(!GLM(vec4_isnan)(proj[0]))
+  ASSERT(!GLM(vec4_isnan)(proj[1]))
+  ASSERT(!GLM(vec4_isnan)(proj[2]))
+  ASSERT(!GLM(vec4_isnan)(proj[3]))
+
+  TEST_SUCCESS
+}

test/src/test_common.h

@@ -106,7 +106,7 @@ test_rand_quat(versor q);
 CGLM_INLINE
 bool
 test_eq(float a, float b) {
-  return fabsf(a - b) <= 1e-6;
+  return fabsf(a - b) <= GLM_FLT_EPSILON * 10;
 }
 
 CGLM_INLINE

test/src/test_ray.h

@@ -0,0 +1,34 @@
+/*
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
+
+#include "test_common.h"
+
+TEST_IMPL(GLM_PREFIX, ray_triangle) {
+  /* Check whether a simple hit is recognized with the right distance */
+  vec3  origin    = { 0.0f,  0.0f,  0.0f};
+  vec3  direction = { 1.0f,  0.0f,  0.0f};
+  vec3  opposite  = {-1.0f,  0.0f,  0.0f};
+  vec3  v0        = { 5.0f, -1.0f,  1.0f};
+  vec3  v1        = { 5.0f, -1.0f, -1.0f};
+  vec3  v2        = { 5.0f,  1.0f,  0.0f};
+  float d;
+  bool  hit;
+
+  hit = GLM(ray_triangle)(origin, direction, v0, v1, v2, &d);
+  ASSERT(hit);
+  ASSERT(fabsf(d - 5.0f) <= 0.0000009);
+
+  /* Check whether a simple miss works */
+  hit = GLM(ray_triangle)(origin, opposite, v0, v1, v2, &d);
+  ASSERT(!hit);
+
+  /* Check that we can disregard distance and pass NULL pointer instead */
+  hit = GLM(ray_triangle)(origin, direction, v0, v1, v2, NULL);
+  ASSERT(hit);
+
+  TEST_SUCCESS
+}

test/src/test_struct.c

@@ -1,9 +1,9 @@
 /*
-* Copyright (c), Recep Aslantas.
-*
-* MIT License (MIT), http://opensource.org/licenses/MIT
-* Full license can be found in the LICENSE file
-*/
+ * Copyright (c), Recep Aslantas.
+ *
+ * MIT License (MIT), http://opensource.org/licenses/MIT
+ * Full license can be found in the LICENSE file
+ */
 
 #include "test_common.h"
 

test/src/tests.c

@@ -22,7 +22,10 @@
 #include "test_project.h"
 #include "test_plane.h"
 #include "test_affine.h"
+#include "test_affine2d.h"
 #include "test_affine_mat.h"
+#include "test_ray.h"
+#include "test_camera.h"
 
 #undef GLM
 #undef GLM_PREFIX
@@ -43,7 +46,10 @@
 #include "test_project.h"
 #include "test_plane.h"
 #include "test_affine.h"
+#include "test_affine2d.h"
 #include "test_affine_mat.h"
+#include "test_ray.h"
+#include "test_camera.h"
 
 #undef GLM
 #undef GLM_PREFIX

test/tests.h

@@ -70,6 +70,33 @@ TEST_DECLARE(glmc_uniscaled)
 TEST_DECLARE(glmc_decompose_rs)
 TEST_DECLARE(glmc_decompose)
 
+/* affine 2d */
+TEST_DECLARE(glm_translate2d)
+TEST_DECLARE(glm_translate2d_to)
+TEST_DECLARE(glm_translate2d_x)
+TEST_DECLARE(glm_translate2d_y)
+TEST_DECLARE(glm_translate2d_make)
+TEST_DECLARE(glm_scale2d_to)
+TEST_DECLARE(glm_scale2d_make)
+TEST_DECLARE(glm_scale2d)
+TEST_DECLARE(glm_scale2d_uni)
+TEST_DECLARE(glm_rotate2d_make)
+TEST_DECLARE(glm_rotate2d)
+TEST_DECLARE(glm_rotate2d_to)
+
+TEST_DECLARE(glmc_translate2d)
+TEST_DECLARE(glmc_translate2d_to)
+TEST_DECLARE(glmc_translate2d_x)
+TEST_DECLARE(glmc_translate2d_y)
+TEST_DECLARE(glmc_translate2d_make)
+TEST_DECLARE(glmc_scale2d_to)
+TEST_DECLARE(glmc_scale2d_make)
+TEST_DECLARE(glmc_scale2d)
+TEST_DECLARE(glmc_scale2d_uni)
+TEST_DECLARE(glmc_rotate2d_make)
+TEST_DECLARE(glmc_rotate2d)
+TEST_DECLARE(glmc_rotate2d_to)
+
 /* mat4 */
 TEST_DECLARE(glm_mat4_ucopy)
 TEST_DECLARE(glm_mat4_copy)
@@ -199,6 +226,10 @@ TEST_DECLARE(glmc_mat2_rmc)
 TEST_DECLARE(camera_lookat)
 TEST_DECLARE(camera_decomp)
 
+TEST_DECLARE(glm_frustum)
+
+TEST_DECLARE(glmc_frustum)
+
 /* project */
 TEST_DECLARE(glm_unprojecti)
 TEST_DECLARE(glm_unproject)
@@ -218,6 +249,10 @@ TEST_DECLARE(clamp)
 /* euler */
 TEST_DECLARE(euler)
 
+/* ray */
+TEST_DECLARE(glm_ray_triangle)
+TEST_DECLARE(glmc_ray_triangle)
+
 /* quat */
 TEST_DECLARE(MACRO_GLM_QUAT_IDENTITY_INIT)
 TEST_DECLARE(MACRO_GLM_QUAT_IDENTITY)
@@ -757,6 +792,33 @@ TEST_LIST {
   TEST_ENTRY(glmc_decompose_rs)
   TEST_ENTRY(glmc_decompose)
   
+  /* affine 2d */
+  TEST_ENTRY(glm_translate2d)
+  TEST_ENTRY(glm_translate2d_to)
+  TEST_ENTRY(glm_translate2d_x)
+  TEST_ENTRY(glm_translate2d_y)
+  TEST_ENTRY(glm_translate2d_make)
+  TEST_ENTRY(glm_scale2d_to)
+  TEST_ENTRY(glm_scale2d_make)
+  TEST_ENTRY(glm_scale2d)
+  TEST_ENTRY(glm_scale2d_uni)
+  TEST_ENTRY(glm_rotate2d_make)
+  TEST_ENTRY(glm_rotate2d)
+  TEST_ENTRY(glm_rotate2d_to)
+
+  TEST_ENTRY(glmc_translate2d)
+  TEST_ENTRY(glmc_translate2d_to)
+  TEST_ENTRY(glmc_translate2d_x)
+  TEST_ENTRY(glmc_translate2d_y)
+  TEST_ENTRY(glmc_translate2d_make)
+  TEST_ENTRY(glmc_scale2d_to)
+  TEST_ENTRY(glmc_scale2d_make)
+  TEST_ENTRY(glmc_scale2d)
+  TEST_ENTRY(glmc_scale2d_uni)
+  TEST_ENTRY(glmc_rotate2d_make)
+  TEST_ENTRY(glmc_rotate2d)
+  TEST_ENTRY(glmc_rotate2d_to)
+  
   /* mat4 */
   TEST_ENTRY(glm_mat4_ucopy)
   TEST_ENTRY(glm_mat4_copy)
@@ -886,6 +948,10 @@ TEST_LIST {
   TEST_ENTRY(camera_lookat)
   TEST_ENTRY(camera_decomp)
   
+  TEST_ENTRY(glm_frustum)
+  
+  TEST_ENTRY(glmc_frustum)
+  
   /* project */
   TEST_ENTRY(glm_unprojecti)
   TEST_ENTRY(glm_unproject)
@@ -905,6 +971,10 @@ TEST_LIST {
   /* euler */
   TEST_ENTRY(euler)
   
+  /* ray */
+  TEST_ENTRY(glm_ray_triangle)
+  TEST_ENTRY(glmc_ray_triangle)
+
   /* quat */
   TEST_ENTRY(MACRO_GLM_QUAT_IDENTITY_INIT)
   TEST_ENTRY(MACRO_GLM_QUAT_IDENTITY)
@@ -984,6 +1054,7 @@ TEST_LIST {
   /* bezier */
   TEST_ENTRY(bezier)
 
+  /* vec2 */
   /* Macros */
 
   TEST_ENTRY(MACRO_GLM_VEC2_ONE_INIT)
@@ -1062,7 +1133,6 @@ TEST_LIST {
   TEST_ENTRY(glmc_vec2_lerp)
 
   /* vec3 */
-
   /* Macros */
 
   TEST_ENTRY(MACRO_GLM_VEC3_ONE_INIT)
@@ -1228,6 +1298,7 @@ TEST_LIST {
   TEST_ENTRY(glmc_vec3_sqrt)
 
   /* vec4 */
+  /* Macros */
 
   TEST_ENTRY(MACRO_GLM_VEC4_ONE_INIT)
   TEST_ENTRY(MACRO_GLM_VEC4_ZERO_INIT)

win/cglm-test.vcxproj

@@ -31,7 +31,9 @@
   <ItemGroup>
     <ClInclude Include="..\test\include\common.h" />
     <ClInclude Include="..\test\src\test_affine.h" />
+    <ClInclude Include="..\test\src\test_affine2d.h" />
     <ClInclude Include="..\test\src\test_affine_mat.h" />
+    <ClInclude Include="..\test\src\test_camera.h" />
     <ClInclude Include="..\test\src\test_common.h" />
     <ClInclude Include="..\test\src\test_mat2.h" />
     <ClInclude Include="..\test\src\test_mat3.h" />
@@ -42,6 +44,7 @@
     <ClInclude Include="..\test\src\test_vec2.h" />
     <ClInclude Include="..\test\src\test_vec3.h" />
     <ClInclude Include="..\test\src\test_vec4.h" />
+    <ClInclude Include="..\test\src\test_ray.h" />
     <ClInclude Include="..\test\tests.h" />
   </ItemGroup>
   <ItemGroup>
@@ -144,6 +147,7 @@
       <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <ConformanceMode>true</ConformanceMode>
       <AdditionalIncludeDirectories>../include;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <BasicRuntimeChecks>Default</BasicRuntimeChecks>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>
@@ -161,6 +165,7 @@
       <PreprocessorDefinitions>_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <ConformanceMode>true</ConformanceMode>
       <AdditionalIncludeDirectories>../include;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <BasicRuntimeChecks>Default</BasicRuntimeChecks>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>

win/cglm-test.vcxproj.filters

@@ -79,5 +79,14 @@
     <ClInclude Include="..\test\src\test_vec2.h">
       <Filter>src</Filter>
     </ClInclude>
+    <ClInclude Include="..\test\src\test_ray.h">
+      <Filter>src</Filter>
+    </ClInclude>
+    <ClInclude Include="..\test\src\test_affine2d.h">
+      <Filter>src</Filter>
+    </ClInclude>
+    <ClInclude Include="..\test\src\test_camera.h">
+      <Filter>src</Filter>
+    </ClInclude>
   </ItemGroup>
 </Project>

win/cglm.vcxproj

@@ -20,6 +20,7 @@
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="..\src\affine.c" />
+    <ClCompile Include="..\src\affine2d.c" />
     <ClCompile Include="..\src\bezier.c" />
     <ClCompile Include="..\src\box.c" />
     <ClCompile Include="..\src\cam.c" />
@@ -34,6 +35,7 @@
     <ClCompile Include="..\src\plane.c" />
     <ClCompile Include="..\src\project.c" />
     <ClCompile Include="..\src\quat.c" />
+    <ClCompile Include="..\src\ray.c" />
     <ClCompile Include="..\src\sphere.c" />
     <ClCompile Include="..\src\vec2.c" />
     <ClCompile Include="..\src\vec3.c" />
@@ -42,11 +44,13 @@
   <ItemGroup>
     <ClInclude Include="..\include\cglm\affine-mat.h" />
     <ClInclude Include="..\include\cglm\affine.h" />
+    <ClInclude Include="..\include\cglm\affine2d.h" />
     <ClInclude Include="..\include\cglm\applesimd.h" />
     <ClInclude Include="..\include\cglm\bezier.h" />
     <ClInclude Include="..\include\cglm\box.h" />
     <ClInclude Include="..\include\cglm\call.h" />
     <ClInclude Include="..\include\cglm\call\affine.h" />
+    <ClInclude Include="..\include\cglm\call\affine2d.h" />
     <ClInclude Include="..\include\cglm\call\bezier.h" />
     <ClInclude Include="..\include\cglm\call\box.h" />
     <ClInclude Include="..\include\cglm\call\cam.h" />
@@ -61,6 +65,7 @@
     <ClInclude Include="..\include\cglm\call\plane.h" />
     <ClInclude Include="..\include\cglm\call\project.h" />
     <ClInclude Include="..\include\cglm\call\quat.h" />
+    <ClInclude Include="..\include\cglm\call\ray.h" />
     <ClInclude Include="..\include\cglm\call\sphere.h" />
     <ClInclude Include="..\include\cglm\call\vec2.h" />
     <ClInclude Include="..\include\cglm\call\vec3.h" />
@@ -80,12 +85,14 @@
     <ClInclude Include="..\include\cglm\plane.h" />
     <ClInclude Include="..\include\cglm\project.h" />
     <ClInclude Include="..\include\cglm\quat.h" />
+    <ClInclude Include="..\include\cglm\ray.h" />
     <ClInclude Include="..\include\cglm\simd\arm.h" />
     <ClInclude Include="..\include\cglm\simd\avx\affine.h" />
     <ClInclude Include="..\include\cglm\simd\avx\mat4.h" />
     <ClInclude Include="..\include\cglm\simd\intrin.h" />
     <ClInclude Include="..\include\cglm\simd\neon\mat4.h" />
     <ClInclude Include="..\include\cglm\simd\sse2\affine.h" />
+    <ClInclude Include="..\include\cglm\simd\sse2\mat2.h" />
     <ClInclude Include="..\include\cglm\simd\sse2\mat3.h" />
     <ClInclude Include="..\include\cglm\simd\sse2\mat4.h" />
     <ClInclude Include="..\include\cglm\simd\sse2\quat.h" />
@@ -93,6 +100,7 @@
     <ClInclude Include="..\include\cglm\sphere.h" />
     <ClInclude Include="..\include\cglm\struct.h" />
     <ClInclude Include="..\include\cglm\struct\affine.h" />
+    <ClInclude Include="..\include\cglm\struct\affine2d.h" />
     <ClInclude Include="..\include\cglm\struct\box.h" />
     <ClInclude Include="..\include\cglm\struct\cam.h" />
     <ClInclude Include="..\include\cglm\struct\color.h" />
@@ -201,7 +209,7 @@
       <PrecompiledHeader>NotUsing</PrecompiledHeader>
       <WarningLevel>Level3</WarningLevel>
       <Optimization>MaxSpeed</Optimization>
-      <PreprocessorDefinitions>WIN32;_DEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;CGLM_DLL;%(PreprocessorDefinitions);CGLM_DLL</PreprocessorDefinitions>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <DebugInformationFormat>None</DebugInformationFormat>
       <BasicRuntimeChecks>Default</BasicRuntimeChecks>
       <PrecompiledHeaderFile />
@@ -217,7 +225,7 @@
     <ClCompile>
       <WarningLevel>Level3</WarningLevel>
       <Optimization>MaxSpeed</Optimization>
-      <PreprocessorDefinitions>_DEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;CGLM_DLL;%(PreprocessorDefinitions);CGLM_DLL</PreprocessorDefinitions>
+      <PreprocessorDefinitions>_DEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <InlineFunctionExpansion>AnySuitable</InlineFunctionExpansion>
       <IntrinsicFunctions>true</IntrinsicFunctions>
       <WholeProgramOptimization>true</WholeProgramOptimization>
@@ -238,11 +246,12 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>WIN32;NDEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;CGLM_DLL;%(PreprocessorDefinitions);CGLM_DLL</PreprocessorDefinitions>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <DebugInformationFormat>None</DebugInformationFormat>
       <PrecompiledHeaderFile />
       <CompileAs>CompileAsC</CompileAs>
       <EnablePREfast>true</EnablePREfast>
+      <BasicRuntimeChecks>Default</BasicRuntimeChecks>
     </ClCompile>
     <Link>
       <SubSystem>Windows</SubSystem>
@@ -257,11 +266,12 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>NDEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;CGLM_DLL;%(PreprocessorDefinitions);CGLM_DLL</PreprocessorDefinitions>
+      <PreprocessorDefinitions>NDEBUG;_WINDOWS;_USRDLL;CGLM_EXPORTS;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <DebugInformationFormat>None</DebugInformationFormat>
       <PrecompiledHeaderFile />
       <CompileAs>CompileAsC</CompileAs>
       <EnablePREfast>true</EnablePREfast>
+      <BasicRuntimeChecks>Default</BasicRuntimeChecks>
     </ClCompile>
     <Link>
       <SubSystem>Windows</SubSystem>

win/cglm.vcxproj.filters

@@ -92,6 +92,12 @@
     <ClCompile Include="..\src\vec2.c">
       <Filter>src</Filter>
     </ClCompile>
+    <ClCompile Include="..\src\ray.c">
+      <Filter>src</Filter>
+    </ClCompile>
+    <ClCompile Include="..\src\affine2d.c">
+      <Filter>src</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <ClInclude Include="..\src\config.h">
@@ -124,6 +130,9 @@
     <ClInclude Include="..\include\cglm\call\vec4.h">
       <Filter>include\cglm\call</Filter>
     </ClInclude>
+    <ClInclude Include="..\include\cglm\call\ray.h">
+      <Filter>include\cglm\call</Filter>
+    </ClInclude>
     <ClInclude Include="..\include\cglm\simd\avx\affine.h">
       <Filter>include\cglm\simd\avx</Filter>
     </ClInclude>
@@ -241,6 +250,9 @@
     <ClInclude Include="..\include\cglm\ease.h">
       <Filter>include\cglm</Filter>
     </ClInclude>
+    <ClInclude Include="..\include\cglm\ray.h">
+      <Filter>include\cglm</Filter>
+    </ClInclude>
     <ClInclude Include="..\include\cglm\simd\arm.h">
       <Filter>include\cglm\simd</Filter>
     </ClInclude>
@@ -346,5 +358,17 @@
     <ClInclude Include="..\include\cglm\struct\vec2-ext.h">
       <Filter>include\cglm\struct</Filter>
     </ClInclude>
+    <ClInclude Include="..\include\cglm\simd\sse2\mat2.h">
+      <Filter>include\cglm\simd\sse2</Filter>
+    </ClInclude>
+    <ClInclude Include="..\include\cglm\call\affine2d.h">
+      <Filter>include\cglm\call</Filter>
+    </ClInclude>
+    <ClInclude Include="..\include\cglm\affine2d.h">
+      <Filter>include\cglm</Filter>
+    </ClInclude>
+    <ClInclude Include="..\include\cglm\struct\affine2d.h">
+      <Filter>include\cglm\struct</Filter>
+    </ClInclude>
   </ItemGroup>
 </Project>