chore: bump version to 1.3.9

.github/workflows/release.yml

@@ -0,0 +1,71 @@
+name: Release
+
+on:
+  push:
+    tags:
+      - "v*"
+
+jobs:
+  build-pure:
+    name: Build pure-Python wheel
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Build wheel (no CUDA)
+        run: |
+          pip wheel . --no-deps -w dist/
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: pure-wheel
+          path: dist/*.whl
+
+  build-cuda-linux:
+    name: Build CUDA wheel (Linux)
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+
+      - name: Install torch (CUDA 12.8)
+        run: |
+          pip install torch --index-url https://download.pytorch.org/whl/cu128
+
+      - name: Setup CUDA
+        uses: Jimver/cuda-toolkit@v0.2.35
+        with:
+          cuda: "12.8.0"
+
+      - name: Build wheel (with CUDA kernels)
+        run: |
+          CSRC_KERNELS=true pip wheel . --no-deps --no-build-isolation -w dist/
+
+      - uses: actions/upload-artifact@v4
+        with:
+          name: cuda-wheel-linux
+          path: dist/*.whl
+
+  release:
+    name: Attach wheels to release
+    needs: [build-pure, build-cuda-linux]
+    runs-on: ubuntu-latest
+    permissions:
+      contents: write
+    steps:
+      - uses: actions/download-artifact@v4
+        with:
+          pattern: "*-wheel"
+          merge-multiple: true
+
+      - name: Create release & upload assets
+        uses: softprops/action-gh-release@v2
+        with:
+          files: ./*.whl
+          tag_name: ${{ github.ref_name }}
+          generate_release_notes: true

astrai/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "1.3.8"
+__version__ = "1.3.9"
 __author__ = "ViperEkura"
 
 from astrai.config import (

csrc/build.py

@@ -20,7 +20,7 @@ def _arch_flags() -> list[str]:
 _kernels_dir = Path("csrc/kernels")
 REGISTRY: dict[str, dict] = {}
 
-CXX_FLAGS = ["-O3", "-march=native", "-funroll-loops"]
+CXX_FLAGS = ["-O3", "-funroll-loops"]
 NVCC_FLAGS = [
     "-O3",
     "--expt-relaxed-constexpr",