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

Drop vLLM 0.13.0 support (#6069)

Browse Source 
### What this PR does / why we need it?
Drop vLLM 0.13.0 support, upgrade to 0.14.0

- vLLM version: v0.13.0
- vLLM main:
d68209402d

---------

Signed-off-by: hfadzxy <starmoon_zhang@163.com>
This commit is contained in:
zhangxinyuehfad
2026-01-23 09:45:08 +08:00
committed by GitHub
parent 27a513b672
commit 819a4459ce
39 changed files with 86 additions and 272 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
vllm_ascend/kv_offload/cpu_npu.py
View File

@@ -2,19 +2,11 @@ import numpy as np
import torch
from vllm.logger import init_logger
from vllm.utils.platform_utils import is_pin_memory_available
from vllm.v1.attention.backend import AttentionBackend # type: ignore
from vllm.v1.kv_offload.mediums import CPULoadStoreSpec, GPULoadStoreSpec
from vllm.v1.kv_offload.worker.worker import (OffloadingHandler,
TransferResult, TransferSpec)
from vllm_ascend.utils import vllm_version_is
# isort: off
if vllm_version_is('0.13.0'):
from vllm.attention.backends.abstract import AttentionBackend # type: ignore
else:
from vllm.v1.attention.backend import AttentionBackend # type: ignore
# isort: on
logger = init_logger(__name__)

14
vllm_ascend/kv_offload/npu.py
View File

@@ -3,6 +3,7 @@ from typing import Optional
import torch
from vllm.config import VllmConfig
from vllm.v1.attention.backend import AttentionBackend # type: ignore
from vllm.v1.kv_offload.abstract import LoadStoreSpec, OffloadingManager
from vllm.v1.kv_offload.backends.cpu import CPUBackend
from vllm.v1.kv_offload.lru_manager import LRUOffloadingManager
@@ -12,14 +13,6 @@ from vllm.v1.kv_offload.worker.worker import OffloadingHandler
from vllm.v1.kv_cache_interface import KVCacheConfig
from vllm_ascend.kv_offload.cpu_npu import CpuNpuOffloadingHandler
from vllm_ascend.utils import vllm_version_is
# isort: off
if vllm_version_is('0.13.0'):
from vllm.attention.backends.abstract import AttentionBackend # type: ignore
else:
from vllm.v1.attention.backend import AttentionBackend # type: ignore
# isort: on
class NPUOffloadingSpec(OffloadingSpec):
@@ -27,10 +20,7 @@ class NPUOffloadingSpec(OffloadingSpec):
def __init__(self,
vllm_config: VllmConfig,
kv_cache_config: Optional[KVCacheConfig] = None):
if vllm_version_is('0.13.0'):
super().__init__(vllm_config)
else:
super().__init__(vllm_config, kv_cache_config)
super().__init__(vllm_config, kv_cache_config)
num_cpu_blocks = self.extra_config.get("num_cpu_blocks")
if not num_cpu_blocks: