EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Drop 0.11.0 support (#4377)

Browse Source 
There is a lot hack code for v0.11.0, which makes the code hard to
upgrade to newer vLLM version. Since v0.11.0 will release soon. Let's
drop v0.11.0 support first. Then we'll upgrade to v0.11.2 soon.


- vLLM version: v0.11.0
- vLLM main:
2918c1b49c

Signed-off-by: wangxiyuan <wangxiyuan1007@gmail.com>
This commit is contained in:
wangxiyuan
2025-11-24 17:08:20 +08:00
committed by GitHub
parent 41ddb06554
commit a1f142b7ad
80 changed files with 467 additions and 1755 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
vllm_ascend/distributed/mooncake/config_data.py
View File

@@ -10,14 +10,7 @@ import torch
from vllm.distributed.kv_transfer.kv_connector.v1.base import \
KVConnectorMetadata
from vllm.utils import logger
from vllm_ascend.utils import vllm_version_is
if vllm_version_is("0.11.0"):
from vllm.utils import cdiv
else:
from vllm.utils.math_utils import cdiv
from vllm.utils.math_utils import cdiv
from vllm.v1.core.sched.output import NewRequestData
DEFAULT_GLOBAL_SEGMENT_SIZE = 3355443200 # 3.125 GiB

7
vllm_ascend/distributed/mooncake/mooncake_engine.py
View File

@@ -8,6 +8,7 @@ from typing import Generator, List, Optional, Union
import torch
from vllm.config import VllmConfig
from vllm.utils import logger
from vllm.utils.torch_utils import get_kv_cache_torch_dtype
from vllm_ascend.distributed.mooncake.config_data import (
ChunkedTokenDatabase, LasyerMultiBlockReqMeta, MooncakeConnectorMetadata,
@@ -16,12 +17,6 @@ from vllm_ascend.distributed.mooncake.kv_transfer import (
KVCacheStoreLayerRecvingThread, KVCacheStoreLayerSendingThread,
KVCacheStoreRecvingThread, KVCacheStoreSendingThread, KVTransferThread)
from vllm_ascend.distributed.mooncake.mooncake_store import Mooncakestore
from vllm_ascend.utils import vllm_version_is
if vllm_version_is("0.11.0"):
from vllm.utils import get_kv_cache_torch_dtype
else:
from vllm.utils.torch_utils import get_kv_cache_torch_dtype
class MooncakeEngine:

7
vllm_ascend/distributed/mooncake/mooncake_store.py
View File

@@ -6,18 +6,13 @@ from mooncake.store import ReplicateConfig # type: ignore
from vllm.config import ParallelConfig
from vllm.distributed.parallel_state import get_tensor_model_parallel_rank
from vllm.utils import logger
from vllm.utils.network_utils import get_ip
from vllm_ascend.distributed.mooncake.config_data import MooncakeEngineKey
from vllm_ascend.distributed.mooncake.transfer_engine import get_global_te
from vllm_ascend.utils import vllm_version_is
from .config_data import MooncakeStoreConfig
if vllm_version_is("0.11.0"):
from vllm.utils import get_ip
else:
from vllm.utils.network_utils import get_ip
METADATA_BYTES_LEN = 24
BASE_PORT = int(os.getenv("VLLM_BASE_PORT", "8790"))

7
vllm_ascend/distributed/mooncake/mooncake_store_connector_v1.py
View File

@@ -10,6 +10,7 @@ from vllm.distributed.kv_transfer.kv_connector.v1.base import (
KVConnectorBase_V1, KVConnectorMetadata, KVConnectorRole)
from vllm.forward_context import ForwardContext
from vllm.utils import logger
from vllm.utils.network_utils import make_zmq_socket
from vllm.v1.core.kv_cache_manager import KVCacheBlocks
from vllm.v1.core.sched.output import SchedulerOutput
from vllm.v1.request import Request
@@ -18,12 +19,6 @@ from vllm.v1.serial_utils import MsgpackDecoder, MsgpackEncoder
from vllm_ascend.distributed.mooncake.config_data import (
LoadSpec, MooncakeConnectorMetadata, ReqMeta, RequestTracker)
from vllm_ascend.distributed.mooncake.mooncake_engine import MooncakeEngine
from vllm_ascend.utils import vllm_version_is
if vllm_version_is("0.11.0"):
from vllm.utils import make_zmq_socket
else:
from vllm.utils.network_utils import make_zmq_socket
class MooncakeConnectorV1(KVConnectorBase_V1):