DirectXMath Versions Save

February 2024 (3.19)

• Fix to address MinGW issue with __cpuid in cpuid.h vs. intrin.h
• Additional updates for clang/LLVM and GNUC
• Minor comment updates

Includes fixes from 3.18b

This version is available on NuGet and via vcpkg.

DirectXMath 3.18b release.

• Hot-fix to address -Wunsafe-buffer-usage warnings from clang v16
• Hot-fix to address MinGW issue with __cpuid in cpuid.h vs. intrin.h
• CMake project updates including pkg-config file generation

This version is available on NuGet and via vcpkg.

The DirectXMath 3.18 release is included in the Windows SDK (10.0.25262) Preview or later.

• C++20 spaceship operators for XMFLOAT2, XMFLOAT3, etc. when building with /std:c++20 /Zc:_cplusplus
• Improved conformance for ARM64 when using /Zc:arm64-aliased-neon-types-
• Minor code review
• CMake project updated to require 3.20 or later
• Added Azure Dev Ops Pipeline YAML files

This version is available on NuGet and via vcpkg.

DirectXMath 3.17 with hot-fixes applied:

• Hot-fix to address -Wreserved-identifier warnings with clang v13
• C++20 spaceship operators for XMFLOAT2, XMFLOAT3, etc. when building with /std:c++20 /Zc:__cplusplus
• Minor CMake project update

This version is available on NuGet and via vcpkg.

The DirectXMath 3.17 release is included in the Windows SDK (10.0.22621) for Windows 11, Version 22H2 .

• Added ColorsLinear namespace to DirectXColors.h with linear versions of .NET colors
• Optimized the XMMatrixRotationRollPitchYaw(FromVector) functions
• Fixed overread problem for 16bpp GPU types Load functions:
  • XMUNIBBLE4, XMU555, XMU565, XMBYTEN2, XMBYTE2, XMUBYTEN2, XMUBYTE2
• XM_CACHE_LINE_SIZE updated for ARM/ARM64 targets to 128 bytes
• A few comments added to improve IntelliSense experience
• Conformance improvements for GNU compiler
• Minor code cleanup

See this post.

This version is available on NuGet and via vcpkg.

This was a CMake only release of DirectXMath 3.16 for vcpkg to resolve case-sensitivity issues in the CMake package implementation.

DirectXMath 3.16 with hot-fixes applied:

• Hot-fixes to resolve build breaks for clang/LLVM and GCC on ARM64
• XM_ALIGNED_DATA and XM_ALIGNED_STRUCT macros updated to use C++17 alignas when available

See this post.

This version is also available on NuGet and via vcpkg.

DirectXMath 3.16 is included in the Windows 10 SDK (10.0.20348.0), version 2104:

• Added XMVectorLog10 / XMVectorExp10
• Added XMColorRGBToYUV_UHD / XMColorYUVToRGB_UHD for Rec. 2020 YUV
• Added optional rhcoords parameter for BoundingFrustum CreateFromMatrix
• Added use of Intel® Short Vector Matrix Library (SVML) supported by VS 2019
  • Opt-in with _XM_SVML_INTRINSICS_; opt-out with _XM_DISABLE_INTEL_SVML_
• Fixed denorm handling for XMConvertFloatToHalf to match F16C/CVT16 instructions.
• Fixed flush (too small for denorm) handling for XMStoreFloat3PK
• Fixed clamping bug in XMStoreByteN4
• Cleaned up ARM-NEON intrinsics type issues for improved portability on GNUC
• Fixed GXMVECTOR for x86 __vectorcall
• Code review
• Plus all changes from 3.15

See this post.

This version is also available on NuGet.

A preview of DirectXMath 3.16 is included in the Windows Insider Preview SDK (20197 and others).

DirectXMath 3.15 includes the following changes:

• Added XMMatrixVectorTensorProduct for creating a matrix from two vectors
• Use of __m256 registers and FMA3 with /arch:AVX2 for stream and some matrix functions
• Optimized load/stores for SSE2 float2 & float3 functions
• Optimized some instruction choices for better AMD CPU support
• Improved conformance for clang/LLVM, GCC, and MinGW compilers
• Code review (constexpr / noexcept usage)
• Retired VS 2015 support

See this post.

This version is also available on NuGet.