Free CPU pinned memory on environment cleanup

Author: janimo

vllm/distributed/parallel_state.py

@@ -1194,6 +1194,11 @@ def cleanup_dist_env_and_memory(shutdown_ray: bool = False):
     gc.collect()
     if not current_platform.is_cpu():
         torch.cuda.empty_cache()
+    try:
+        torch._C._host_emptyCache()
+    except AttributeError:
+        logger.warning(
+            "torch._C._host_emptyCache() only available in Pytorch >=2.5")
 
 
 def in_the_same_node_as(pg: ProcessGroup, source_rank: int = 0) -> List[bool]: