Remove global CUDA graph pool

Signed-off-by: Woosuk Kwon <woosuk@thinkingmachines.ai>

vllm/compilation/cuda_graph.py

@@ -82,7 +82,7 @@ class CUDAGraphWrapper:
         # TODO: in the future, if we want to use multiple
         # streams, it might not be safe to share a global pool.
         # only investigate this when we use multiple streams
-        self.graph_pool = current_platform.get_global_graph_pool()
+        self.graph_pool = current_platform.graph_pool_handle()
 
         if cudagraph_options is None:
             cudagraph_options = CUDAGraphOptions()

vllm/platforms/interface.py

@@ -140,8 +140,6 @@ class Platform:
 
     additional_env_vars: list[str] = []
 
-    _global_graph_pool: Optional[Any] = None
-
     @property
     def supported_dtypes(self) -> list[torch.dtype]:
         """Returns the supported dtypes for the current platform."""
@@ -539,10 +537,7 @@ class Platform:
         """
         Return the global graph pool for this platform.
         """
-        cls = self.__class__
-        if cls._global_graph_pool is None:
-            cls._global_graph_pool = self.graph_pool_handle()
-        return cls._global_graph_pool
+        return self.graph_pool_handle()
 
     @classmethod
     def get_cu_count(cls, device_id: int = 0) -> int:

vllm/v1/worker/gpu_ubatch_wrapper.py

@@ -54,7 +54,7 @@ class UBatchWrapper:
         if runtime_mode is not CUDAGraphMode.NONE:
             self.cudagraph_wrapper = CUDAGraphWrapper(
                 runnable, vllm_config, runtime_mode=runtime_mode)
-            self.graph_pool = current_platform.get_global_graph_pool()
+            self.graph_pool = current_platform.graph_pool_handle()
 
     def __getattr__(self, key: str):
         # allow accessing the attributes of the runnable.