Implement SIMD matrix multiplications

.clang-format

@@ -28,7 +28,7 @@ BitFieldColonSpacing: Both
 BraceWrapping: 
   AfterCaseLabel: false
   AfterClass: true
-  AfterControlStatement: MultiLine
+  AfterControlStatement: Never
   AfterEnum: true
   AfterFunction: true
   AfterNamespace: false
@@ -69,7 +69,7 @@ IndentCaseBlocks: false
 IndentCaseLabels: true
 IndentExternBlock: NoIndent
 IndentGotoLabels: false
-IndentPPDirectives: AfterHash
+IndentPPDirectives: None
 IndentRequires: true
 IndentWidth: 2
 IndentWrappedFunctionNames: false

.github/workflows/linux-ninja-clang.yml

@@ -52,7 +52,7 @@ jobs:
         shell: bash
         run: |
           if [ ! -x "$(command -v clang++-${{ matrix.ver }})" ]; then
-            sudo apt-get install -qqy --no-install-recommends clang++-${{ matrix.ver }}
+            sudo apt-get install -qqy --no-install-recommends clang++-${{ matrix.ver }} clang-tools-${{ matrix.ver }}
           fi
       - uses: actions/checkout@v4
       - uses: lukka/get-cmake@latest
@@ -78,7 +78,7 @@ jobs:
           annotate: true
           summary: true
           use-suite-name: true
-          file-report: true
+          file-report: false
           failed-report: true
           upload-artifact: false
           integrations-config: |

.github/workflows/linux-ninja-gcc.yml

@@ -19,7 +19,6 @@ jobs:
       fail-fast: false
       matrix:
         include:
-        - ver: "12"
         - ver: "13"
         - ver: "14"
     env:
@@ -66,7 +65,7 @@ jobs:
         run: |
            preset="linux-ninja-gcc-s-vcpkg"
            echo "Executing workflow $preset for gcc-${{ matrix.ver }} with VCPKG_ROOT @ $VCPKG_ROOT"
-           CC=gcc-15 CXX=g++-15 cmake --workflow --preset=$preset
+           cmake --workflow --preset=$preset
       - name: Generate JUnit test results
         uses: ctrf-io/github-test-reporter@v1
         env:
@@ -78,7 +77,7 @@ jobs:
           annotate: true
           summary: true
           use-suite-name: true
-          file-report: true
+          file-report: false
           failed-report: true
           upload-artifact: false
           integrations-config: |

.github/workflows/windows-msbuild-msvc-17.yml

@@ -97,7 +97,7 @@ jobs:
           annotate: true
           summary: true
           use-suite-name: true
-          file-report: true
+          file-report: false
           failed-report: true
           upload-artifact: false
           integrations-config: |

.github/workflows/windows-ninja-clangcl.yml

@@ -51,7 +51,7 @@ jobs:
           annotate: true
           summary: true
           use-suite-name: true
-          file-report: true
+          file-report: false
           failed-report: true
           upload-artifact: false
           integrations-config: |

AGENTS.md

@@ -0,0 +1,156 @@
+# AGENTS.md
+
+This file provides guidance to agents when working with code in this repository.
+
+## Build, Lint, Test, and Run Commands
+
+### Build Commands
+1. **Using CMake directly**:
+   - Configure: 
+     ```bash
+     cmake -S <path to cml-root> -G{generator name} -DBUILD_TESTING=On -DCMAKE_BUILD_TYPE=Release
+     ```
+   - Build:
+     ```bash
+     cmake --build . --config Release
+     ```
+   - Parallel Build (e.g., 4 threads):
+     ```bash
+     cmake --build . --config Release -- -j4
+     ```
+
+2. **Using CMake Presets**:
+   - List available presets:
+     ```bash
+     cmake --list-presets=all
+     ```
+   - Example preset-based build:
+     ```bash
+     cmake --preset=windows-ninja-clangcl-mt-s-vcpkg
+     cmake --build --preset=windows-ninja-clangcl-mt-s-vcpkg-release
+     ```
+
+### Test Commands
+1. **Run all tests**:
+   ```bash
+   ctest -C Release
+   ```
+2. **Parallel Test Execution**:
+   ```bash
+   ctest -C Release -j4
+   ```
+
+### Lint Commands
+No explicit linting commands were found in the analyzed files.
+
+### Run Commands
+The project appears to be a library (header-only), so there are no explicit "run" commands for an application. However, the tests and timing executables can be run as described above.
+
+## Code Style Guidelines
+1. **Formatting**:
+   - **Column Limit**: 80 characters.
+   - **Indentation**: 2 spaces, no tabs.
+   - **Line Endings**: LF (Unix-style).
+   - **Insert Newline at EOF**: Enabled.
+
+2. **Braces**:
+   - **After Class/Struct/Enum/Function**: Always place braces on a new line.
+   - **Control Statements**: No braces for single-line statements.
+
+3. **Alignment**:
+   - **Trailing Comments**: Aligned.
+   - **Operands**: No alignment.
+   - **Pointer Alignment**: Left-aligned (`int* ptr`).
+
+4. **Spaces**:
+   - **Before Parentheses**: None (e.g., `if(condition)`).
+   - **Around Operators**: Space before assignment operators.
+   - **In Empty Blocks/Parentheses**: None.
+
+5. **Includes**:
+   - **Sorting**: Disabled.
+   - **Blocks**: Preserve original order.
+
+6. **Miscellaneous**:
+   - **Short Functions**: Inline only.
+   - **Short If Statements**: Allowed on a single line.
+   - **Namespace Indentation**: None.
+   - **Header Include Guards**: Use `#pragma once` only, never `#ifdef/#endif` guards.
+
+7. **New Files**:
+   - **File Naming**: Use descriptive names that reflect the content of the file.
+   - **File Extension for C++**: `.cpp`, `.h`, `.inl`, `.tpp`, `.hpp`.
+   - **File header**: Use the default file header for all new files.
+
+8. **Comments**:
+   - **C++ Comment Style**: Use C comment delimiters for C++ files.
+   - **Short Comments**: Allowed on a single line if they are short.
+   - **Long Comments**: Aligned at the beginning of the block.
+
+
+### Standard File Headers by Language
+1. C++:
+   ```cpp
+   /*-------------------------------------------------------------------------
+    @@COPYRIGHT@@
+    *-----------------------------------------------------------------------*/
+   ```
+
+2. Python
+   ```python
+   #*-------------------------------------------------------------------------
+   # @@COPYRIGHT@@
+   #*-------------------------------------------------------------------------
+   ```
+
+## Core Architecture and Components
+1. **Vectors (`cml/vector/`)**:
+   - Provides various vector types (e.g., fixed, dynamic, external).
+   - Includes operations like dot product, cross product, and normalization.
+   - Supports expression templates for efficient computation.
+
+2. **Matrices (`cml/matrix/`)**:
+   - Implements matrix types and operations (e.g., inversion, transformation).
+   - Includes support for fixed and dynamic sizes.
+   - Provides utilities for matrix concatenation and basis generation.
+
+3. **Quaternions (`cml/quaternion/`)**:
+   - Encapsulates quaternion operations for rotations and transformations.
+   - Includes scalar and vector operations on quaternions.
+   - Provides utilities for normalization, inversion, and logarithmic operations.
+
+4. **Storage (`cml/storage/`)**:
+   - Defines storage strategies for vectors, matrices, and quaternions.
+   - Includes dynamic, fixed, and external storage types.
+   - Provides utilities for resizing and reshaping storage.
+
+5. **Utilities (`cml/util/`)**:
+   - Contains helper functions for printing and hashing.
+   - Includes mathematical utilities for field-of-view conversions and more.
+
+6. **Testing (`tests/`)**:
+   - Organized into subdirectories for each core component (e.g., `vector`, `matrix`, `quaternion`).
+   - Uses Catch2 for unit testing.
+   - Includes timing tests for performance evaluation.
+
+## Testing Setup
+1. **Framework**:
+   - The project uses **Catch2** for unit testing.
+   - The main entry point for tests is defined in [`tests/main/catch_main.cpp`](tests/main/catch_main.cpp:11), which initializes the Catch2 session.
+
+2. **Test Organization**:
+   - Tests are organized into subdirectories under `tests/`, corresponding to the core components.
+
+3. **Test Execution**:
+   - Build the tests:
+     ```bash
+     cmake --build . --config Release
+     ```
+   - Run the tests:
+     ```bash
+     ctest -C Release
+     ```
+
+4. **Timing Tests**:
+   - Timing tests are conditionally included if `BUILD_TIMING` is enabled in the CMake configuration.
+   - These tests are located in the `tests/timing/` directory.

CMakeLists.txt

@@ -23,6 +23,23 @@ endif()
 # Standard includes:
 include(CMakeDependentOption)
 
+# Allow disabling SIMD:
+if(NOT DEFINED CML_DISABLE_SIMD)
+  option(CML_DISABLE_SIMD "Disable vectorization (also disables timing tests)" OFF)
+endif()
+
+# Build tests if requested:
+if(NOT DEFINED BUILD_TESTING)
+  option(BUILD_TESTING "Build CML tests" OFF)
+endif()
+
+if(NOT DEFINED BUILD_TIMING)
+  cmake_dependent_option(BUILD_TIMING
+    "Build timing tests (requires SIMD)" OFF 
+    "CML_DISABLE_SIMD" OFF
+  )
+endif()
+
 if(NOT DEFINED BUILD_SHARED_LIBS)
   option(BUILD_SHARED_LIBS "Build shared libraries" OFF)
 endif()
@@ -37,10 +54,8 @@ endif()
 if(NOT DEFINED CML_ENABLE_RELATIVE_TEST_PATHS)
   # Force CTest to use relative paths. Primarily for automated testing:
   cmake_dependent_option(CML_ENABLE_RELATIVE_TEST_PATHS
-    "Make test executable paths relative to CMAKE_BINARY_DIR"
-    OFF  # Show and default to OFF if...
-    "BUILD_TESTING"
-    OFF  # Otherwise, default to OFF.
+    "Make test executable paths relative to CMAKE_BINARY_DIR" OFF
+    "BUILD_TESTING" OFF
   )
 endif()
 
@@ -50,7 +65,16 @@ endif()
 
 list(INSERT CMAKE_MODULE_PATH 0 "${CMAKE_CURRENT_LIST_DIR}/cmake")
 include(default-paths)
-include(${CMAKE_CXX_COMPILER_ID}-compiler)
+include(configure-compiler)
+
+# SIMDe is required when SIMD is enabled:
+if(NOT CML_DISABLE_SIMD)
+  set(_hints)
+  if(DEFINED VCPKG_INSTALLED_DIR AND DEFINED VCPKG_TARGET_TRIPLET)
+    list(APPEND _hints HINTS "${VCPKG_INSTALLED_DIR}/${VCPKG_TARGET_TRIPLET}/include")
+  endif()
+  find_path(SIMDE_INCLUDE_DIR NAMES "simde/simde-common.h" ${_hints})
+endif()
 
 # Create the CML interface library:
 add_subdirectory(cml)
@@ -100,4 +124,4 @@ install(
 export(
   EXPORT Targets
   FILE "${CMAKE_BINARY_DIR}/${CML_INSTALL_CMAKEDIR}/${_TARGET_FILE}"
-  )
+  )

CMakePresets.json

@@ -9,6 +9,18 @@
     "cmake/presets/vcpkg.json"
   ],
   "configurePresets": [
+    {
+      "name": ".timing",
+      "hidden": true,
+      "description": "Build timing executables",
+      "cacheVariables": {
+        "BUILD_TIMING": {
+          "type": "BOOL",
+          "value": true
+        }
+      }
+    },
+
     {
       "name": "linux-ninja-clang-s",
       "description": "Linux Ninja configuration using clang++-${ENV_COMPILER_VERSION}",
@@ -18,13 +30,13 @@
     {
       "name": "linux-ninja-clang-s-vcpkg",
       "description": "Linux Ninja configuration using clang++-${ENV_COMPILER_VERSION} and vcpkg",
-      "inherits": [ ".linux-ninja-clang", ".static", ".vcpkg-linux-any-static", ".testing" ]
+      "inherits": [ ".linux-ninja-clang", ".static", ".vcpkg-linux-any-static", ".testing", ".timing" ]
     },
 
     {
       "name": "linux-ninja-gcc-s-vcpkg",
       "description": "Linux Ninja configuration using g++-${ENV_COMPILER_VERSION} and vcpkg",
-      "inherits": [ ".linux-ninja-gcc", ".static", ".vcpkg-linux-any-static", ".testing" ]
+      "inherits": [ ".linux-ninja-gcc", ".static", ".vcpkg-linux-any-static", ".testing", ".timing" ]
     },
 
     {
@@ -36,19 +48,19 @@
     {
       "name": "windows-ninja-clangcl-mt-s-vcpkg",
       "description": "Windows Ninja configuration using clang-cl.ex and vcpkg",
-      "inherits": [ ".windows-ninja-clangcl", ".static", ".vcpkg-windows-clangcl-static", ".testing" ]
+      "inherits": [ ".windows-ninja-clangcl", ".static", ".vcpkg-windows-clangcl-static", ".testing", ".timing" ]
     },
 
     {
       "name": "windows-msbuild-msvc-17-mt-s-vcpkg",
       "description": "Windows MSBuild configuration using ${MSVC_TOOLSET_VERSION} and vcpkg",
-      "inherits": [ ".windows-msbuild-msvc-17", ".static", ".vcpkg-windows-msvc-static", ".testing" ]
+      "inherits": [ ".windows-msbuild-msvc-17", ".static", ".vcpkg-windows-msvc-static", ".testing", ".timing" ]
     },
 
     {
       "name": "windows-msbuild-clangcl-17-mt-s-vcpkg",
       "description": "Windows MSBuild configuration using ClangCL and vcpkg",
-      "inherits": [ ".windows-msbuild-clangcl-17", ".static", ".vcpkg-windows-clangcl-static", ".testing" ]
+      "inherits": [ ".windows-msbuild-clangcl-17", ".static", ".vcpkg-windows-clangcl-static", ".testing", ".timing" ]
     }
   ],
   "buildPresets": [