ARM runners (faster than cross compilation qemu)

.github/workflows/python-package.yml

@@ -30,13 +30,15 @@ jobs:
   build-shared-libs:
     strategy:
       matrix:
-        os: [ubuntu-latest, macos-latest, windows-latest]
+        os: [ubuntu-latest, macos-latest, windows-latest, ubuntu-22.04-arm]
         arch: [x86_64, aarch64]
         exclude:
-          - os: windows-latest # This probably requires arm64 Windows agents
+          - os: windows-latest
             arch: aarch64
-          - os: ubuntu-latest # Temporary. Takes too long, not ready yet.
+          - os: ubuntu-latest
             arch: aarch64
+          - os: ubuntu-22.04-arm 
+            arch: x86_64
     runs-on: ${{ matrix.os }} # One day, we could run them on native agents. Azure supports this now but it's planned only for Q3 2023 for hosted agents
     steps:
       - uses: actions/checkout@v4
@@ -61,22 +63,20 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        os: [ubuntu-latest, windows-latest]
+        os: [ubuntu-latest, windows-latest, ubuntu-22.04-arm]
         arch: [x86_64, aarch64]
         cuda_version:
           ["11.7.1", "11.8.0", "12.0.1", "12.1.1", "12.2.2", "12.3.2", "12.4.1", "12.5.1", "12.6.3", "12.8.0"]
         exclude:
-          - os: windows-latest # This probably requires arm64 Windows agents
+          - os: windows-latest
             arch: aarch64
-          - os: ubuntu-latest # Temporary. Takes too long, not ready yet.
+          - os: ubuntu-latest
             arch: aarch64
+          - os: ubuntu-22.04-arm 
+            arch: x86_64
     runs-on: ${{ matrix.os }} # One day, we could run them on native agents. Azure supports this now but it's planned only for Q3 2023 for hosted agents
     steps:
       - uses: actions/checkout@v4
-        # Linux: We use Docker to build cross platform Cuda (aarch64 is built in emulation)
-      - name: Set up Docker multiarch
-        if: startsWith(matrix.os, 'ubuntu')
-        uses: docker/setup-qemu-action@v3
         # Windows: We install Cuda on the agent (slow)
       - uses: Jimver/[email protected]
         if: startsWith(matrix.os, 'windows')
@@ -109,17 +109,19 @@ jobs:
       - build-shared-libs-cuda
     strategy:
       matrix:
-        os: [ubuntu-latest, macos-latest, windows-latest]
+        os: [ubuntu-latest, macos-latest, windows-latest, ubuntu-22.04-arm]
         # The specific Python version is irrelevant in this context as we are only packaging non-C extension
         # code. This ensures compatibility across Python versions, including Python 3.8, as compatibility is
         # dictated by the packaged code itself, not the Python version used for packaging.
         python-version: ["3.10"]
         arch: [x86_64, aarch64]
         exclude:
-          - os: windows-latest # This probably requires arm64 Windows agents
+          - os: windows-latest
             arch: aarch64
-          - os: ubuntu-latest # Temporary. Takes too long, not ready yet.
+          - os: ubuntu-latest
             arch: aarch64
+          - os: ubuntu-22.04-arm 
+            arch: x86_64
     runs-on: ${{ matrix.os }}
     steps:
       - uses: actions/checkout@v4
@@ -184,7 +186,7 @@ jobs:
       - name: Inspect wheels directory after renaming files
         run: ls -alFR wheels/
       - name: Create release and upload artifacts
-        uses: softprops/action-gh-release@v2.0.8
+        uses: softprops/action-gh-release@v2.2.1
         with:
           files: wheels/*.whl
           prerelease: true