[bugfix][refactor] fix recompute_scheduler break with vllm 0.12.0 & support async scheduling & refactor recompute_scheduler.py

Signed-off-by: linfeng-yuan <[email protected]>

Author: linfeng-yuan

vllm_ascend/core/async_recompute_scheduler.py

@@ -0,0 +1,31 @@
+#
+# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# This file is a part of the vllm-ascend project.
+#
+
+from __future__ import annotations
+
+from vllm.logger import init_logger
+from vllm.v1.core.sched.async_scheduler import AsyncScheduler
+
+from vllm_ascend.core.recompute_scheduler import RecomputeScheduler
+
+
+logger = init_logger(__name__)
+
+
+class AsyncRecomputeScheduler(AsyncScheduler, RecomputeScheduler):
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)

vllm_ascend/core/recompute_schedule_config.py

@@ -18,22 +18,25 @@
 from dataclasses import dataclass, fields
 from typing import Type, Union
 
-from vllm.config import SchedulerConfig
-
-MAX_INT = 2147483647
-
+from vllm.config import SchedulerConfig, VllmConfig
 
 @dataclass
 class RecomputeSchedulerConfig(SchedulerConfig):
     scheduler_cls: Union[str, Type[object]] = (
         "vllm_ascend.core.recompute_scheduler.RecomputeScheduler")
-
     @classmethod
-    def initialize_from_config(cls, vllm_scheduler_config: SchedulerConfig):
+    def initialize_from_config(cls, vllm_config: VllmConfig):
+        vllm_scheduler_config = vllm_config.scheduler_config
         scheduler_config = {
             field.name: getattr(vllm_scheduler_config, field.name)
             for field in fields(vllm_scheduler_config) if field.init
         }
-        scheduler_config["scheduler_cls"] = (
-            "vllm_ascend.core.recompute_scheduler.RecomputeScheduler")
+        if vllm_scheduler_config.async_scheduling:
+            scheduler_config["scheduler_cls"] = (
+                "vllm_ascend.core.async_recompute_scheduler.AsyncRecomputeScheduler")
+        else:
+            scheduler_config["scheduler_cls"] = (
+                "vllm_ascend.core.recompute_scheduler.RecomputeScheduler")
+        scheduler_config["max_model_len"] = vllm_config.model_config.max_model_len
+        scheduler_config["is_encoder_decoder"] = vllm_config.model_config.is_encoder_decoder
         return cls(**scheduler_config)