[CPU] Fix torch.compile crash from torch.accelerator.synchronize on CPU-only hosts (#49960)

Signed-off-by: R <Ganesh.R@amd.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Li, Jiang <jiang1.li@intel.com>

vllm/v1/worker/cpu/shm.py

@@ -17,6 +17,12 @@ def noop(*args: Any, **kwargs: Any) -> None:
     pass
 
 
+# Distinct no-op so empty_cache does not alias synchronize: Dynamo's
+# handle_synchronize is keyed on that object and asserts on CPU-only hosts.
+def empty_cache_noop(*args: Any, **kwargs: Any) -> None:
+    pass
+
+
 def fake_pin_memory(self: torch.Tensor, *args: Any, **kwargs: Any) -> torch.Tensor:
     return self
 
@@ -58,7 +64,7 @@ torch.cuda.Stream = _StreamPlaceholder
 torch.cuda.set_stream = noop
 torch.cuda.current_stream = lambda *args, **kwargs: _StreamPlaceholder()
 torch.accelerator.synchronize = noop
-torch.accelerator.empty_cache = noop
+torch.accelerator.empty_cache = empty_cache_noop
 torch.Tensor.pin_memory = fake_pin_memory
 torch.Tensor.record_stream = noop
 torch.accelerator.get_memory_info = get_memory_info

vllm/v1/worker/cpu_model_runner.py

@@ -240,5 +240,10 @@ def _set_torch_accelerator_to_noop() -> None:
     def noop(*args: Any, **kwargs: Any) -> None:
         pass
 
+    # Distinct no-op so empty_cache does not alias synchronize: Dynamo's
+    # handle_synchronize is keyed on that object and asserts on CPU-only hosts.
+    def empty_cache_noop(*args: Any, **kwargs: Any) -> None:
+        pass
+
     torch.accelerator.synchronize = noop
-    torch.accelerator.empty_cache = noop
+    torch.accelerator.empty_cache = empty_cache_noop