attempt to deal with missing Linux-only os.sched_getaffinity

Signed-off-by: Luke Nezda <[email protected]>

intel_extension_for_transformers/llm/runtime/graph/__init__.py

@@ -14,6 +14,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+import multiprocessing
 import os
 
 import torch
@@ -132,7 +133,13 @@ def init_from_bin(self, model_type, model_path, **generate_kwargs):
         if "threads" not in generate_kwargs:
             threads = os.getenv("OMP_NUM_THREADS")
             if threads is None:
-                generate_kwargs["threads"] = len(os.sched_getaffinity(0))
+                # hack from https://stackoverflow.com/questions/74048135/alternatives-to-os-sched-getaffinity-for-macos
+                #  do we need physical core count instead?
+                try:
+                    # NOTE: only available on some Unix platforms
+                    generate_kwargs["threads"] = len(os.sched_getaffinity(0))
+                except AttributeError:
+                    generate_kwargs["threads"] = multiprocessing.cpu_count()
             else:
                 generate_kwargs["threads"] = int(threads)
         self.model.init_model(model_path, **generate_kwargs)