unify init logic for all platforms

vllm/executor/ray_utils.py

@@ -10,7 +10,7 @@
 from vllm.logger import init_logger
 from vllm.platforms import current_platform
 from vllm.sequence import ExecuteModelRequest, IntermediateTensors
-from vllm.utils import get_ip, is_hip, is_xpu
+from vllm.utils import get_ip
 from vllm.worker.worker_base import WorkerWrapperBase
 
 logger = init_logger(__name__)
@@ -230,15 +230,17 @@ def initialize_ray_cluster(
     """
     assert_ray_available()
     # Connect to a ray cluster.
-    if is_hip() or is_xpu():
-        # Try to connect existing ray instance and create a new one if not found
+    if ray_address is None:
+        # Try to connect latest created ray instance and create
+        # a new one if not found
         try:
             ray.init("auto")
         except ConnectionError:
             logger.warning(
-                "No existing RAY instance detected. "
-                "A new instance will be launched with current node resources.")
-            ray.init(address=ray_address,
+                "Neither ray_address specified nor existing RAY instance"
+                " detected. A new instance will be launched with current"
+                " node resources.")
+            ray.init(address=None,
                      ignore_reinit_error=True,
                      num_gpus=parallel_config.world_size)
     else: