make cuda.core also support CUDA_PYTHON_PARALLEL_LEVEL

Author: leofang

.github/workflows/build-wheel.yml

@@ -78,6 +78,8 @@ jobs:
           CIBW_BUILD_VERBOSITY: 1
           CIBW_BEFORE_BUILD_WINDOWS: "pip install delvewheel"
           CIBW_REPAIR_WHEEL_COMMAND_WINDOWS: "delvewheel repair --namespace-pkg cuda -w {dest_dir} {wheel}"
+          CIBW_ENVIRONMENT: >
+            CUDA_PYTHON_PARALLEL_LEVEL=${{ env.CUDA_PYTHON_PARALLEL_LEVEL }}
         with:
           package-dir: ./cuda_core/
           output-dir: ${{ env.CUDA_CORE_ARTIFACTS_DIR }}

cuda_core/setup.py

@@ -9,6 +9,10 @@
 from setuptools import Extension, setup
 from setuptools.command.build_ext import build_ext as _build_ext
 
+
+nthreads = int(os.environ.get("CUDA_PYTHON_PARALLEL_LEVEL", os.cpu_count() // 2))
+
+
 # It seems setuptools' wildcard support has problems for namespace packages,
 # so we explicitly spell out all Extension instances.
 root_module = "cuda.core.experimental"
@@ -33,7 +37,7 @@ def strip_prefix_suffix(filename):
 
 class build_ext(_build_ext):
     def build_extensions(self):
-        self.parallel = os.cpu_count() // 2
+        self.parallel = nthreads
         super().build_extensions()