[Perf][CLI] Improve overall startup time (#19941)

vllm/engine/arg_utils.py

@@ -3,7 +3,9 @@
 
 # yapf: disable
 import argparse
+import copy
 import dataclasses
+import functools
 import json
 import sys
 import threading
@@ -168,7 +170,8 @@ def get_type_hints(type_hint: TypeHint) -> set[TypeHint]:
     return type_hints
 
 
-def get_kwargs(cls: ConfigType) -> dict[str, Any]:
+@functools.lru_cache(maxsize=30)
+def _compute_kwargs(cls: ConfigType) -> dict[str, Any]:
     cls_docs = get_attr_docs(cls)
     kwargs = {}
     for field in fields(cls):
@@ -269,6 +272,16 @@ def get_kwargs(cls: ConfigType) -> dict[str, Any]:
     return kwargs
 
 
+def get_kwargs(cls: ConfigType) -> dict[str, Any]:
+    """Return argparse kwargs for the given Config dataclass.
+
+    The heavy computation is cached via functools.lru_cache, and a deep copy
+    is returned so callers can mutate the dictionary without affecting the
+    cached version.
+    """
+    return copy.deepcopy(_compute_kwargs(cls))
+
+
 @dataclass
 class EngineArgs:
     """Arguments for vLLM engine."""