[Bench] Split serve.py:main into async/async versions (#22405)

Signed-off-by: Linkun <github@lkchen.net>

vllm/benchmarks/serve.py

@@ -948,7 +948,10 @@ def add_cli_args(parser: argparse.ArgumentParser):
     )
 
 
-def main(args: argparse.Namespace):
+def main(args: argparse.Namespace) -> dict[str, Any]:
+    return asyncio.run(main_async(args))
+
+async def main_async(args: argparse.Namespace) -> dict[str, Any]:
     print(args)
     random.seed(args.seed)
     np.random.seed(args.seed)
@@ -1025,8 +1028,7 @@ def main(args: argparse.Namespace):
     gc.collect()
     gc.freeze()
 
-    benchmark_result = asyncio.run(
-        benchmark(
+    benchmark_result = await benchmark(
             endpoint_type=args.endpoint_type,
             api_url=api_url,
             base_url=base_url,
@@ -1052,10 +1054,9 @@ def main(args: argparse.Namespace):
             ramp_up_start_rps=args.ramp_up_start_rps,
             ramp_up_end_rps=args.ramp_up_end_rps,
             ready_check_timeout_sec=args.ready_check_timeout_sec,
-        ))
+        )
 
     # Save config and results to json
-    if args.save_result or args.append_result:
     result_json: dict[str, Any] = {}
 
     # Setup
@@ -1108,6 +1109,7 @@ def main(args: argparse.Namespace):
                 del benchmark_result[field]
 
         # Save to file
+    if args.save_result or args.append_result:
         base_model_id = model_id.split("/")[-1]
         max_concurrency_str = (f"-concurrency{args.max_concurrency}"
                                if args.max_concurrency is not None else "")
@@ -1129,3 +1131,5 @@ def main(args: argparse.Namespace):
                 outfile.write("\n")
             json.dump(result_json, outfile)
         save_to_pytorch_benchmark_format(args, result_json, file_name)
+
+    return result_json
\ No newline at end of file