Add Windows ARM64 wheel build support with NEON optimization

.github/workflows/python-package.yml

@@ -38,6 +38,8 @@ jobs:
             arch: aarch64
           - os: windows-2025
             arch: x86_64
+          - os: windows-11-arm
+            arch: arm64
           - os: macos-15
             arch: arm64
     runs-on: ${{ matrix.os }}
@@ -46,6 +48,8 @@ jobs:
       - name: Setup MSVC
         if: startsWith(matrix.os, 'windows')
         uses: ilammy/msvc-dev-cmd@v1.13.0 # to use cl
+        with:
+          arch: ${{ matrix.arch == 'arm64' && 'arm64' || 'x64' }}
       - name: Build C++
         run: bash .github/scripts/build-cpu.sh
         env:
@@ -188,20 +192,27 @@ jobs:
       - build-xpu
     strategy:
       matrix:
-        os: [ubuntu-22.04, ubuntu-22.04-arm, windows-2025, macos-15]
+        os: [ubuntu-22.04, ubuntu-22.04-arm, windows-2025, windows-11-arm, macos-15]
         include:
           - os: ubuntu-22.04
             arch: x86_64
+            python-version: "3.10"
           - os: ubuntu-22.04-arm
             arch: aarch64
+            python-version: "3.10"
           - os: windows-2025
             arch: x86_64
+            python-version: "3.10"
+          - os: windows-11-arm
+            arch: arm64
+            # Python for Windows ARM64 is only available from 3.12+
+            python-version: "3.12"
           - os: macos-15
             arch: arm64
+            python-version: "3.10"
         # The specific Python version is irrelevant in this context as we are only packaging non-C extension
         # code. This ensures compatibility across Python versions, as compatibility is
         # dictated by the packaged code itself, not the Python version used for packaging.
-        python-version: ["3.10"]
     runs-on: ${{ matrix.os }}
     steps:
       - uses: actions/checkout@v4
@@ -327,6 +338,8 @@ jobs:
               echo "### Linux (aarch64)" >> body.md
             elif [[ "$fname" == *"win_amd64"* ]]; then
               echo "### Windows (x86_64)" >> body.md
+            elif [[ "$fname" == *"win_arm64"* ]]; then
+              echo "### Windows (arm64)" >> body.md
             elif [[ "$fname" == *"macosx"* ]]; then
               echo "### macOS 14+ (arm64)" >> body.md
             else

CMakeLists.txt

@@ -337,7 +337,13 @@ if(WIN32)
 endif()
 
 if(MSVC)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2 /fp:fast")
+    # /arch:AVX2 is only valid for x86/x64 targets, not ARM64
+    string(TOLOWER "${CMAKE_SYSTEM_PROCESSOR}" _msvc_arch)
+    if(_msvc_arch MATCHES "x86|x64|amd64")
+        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2 /fp:fast")
+    else()
+        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /fp:fast")
+    endif()
 endif()
 
 set_source_files_properties(${CPP_FILES} PROPERTIES LANGUAGE CXX)