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

7
tests/ut/worker/test_input_batch.py
View File

@@ -20,19 +20,14 @@ import numpy as np
import pytest
import torch
from vllm.sampling_params import SamplingParams
from vllm.utils.torch_utils import make_tensor_with_pad
from vllm.v1.pool.metadata import PoolingMetadata
from vllm.v1.sample.logits_processor import LogitsProcessors
from vllm.v1.sample.metadata import SamplingMetadata
from vllm_ascend.utils import vllm_version_is
from vllm_ascend.worker.block_table import BlockTable, MultiGroupBlockTable
from vllm_ascend.worker.npu_input_batch import CachedRequestState, InputBatch
if vllm_version_is("0.11.0"):
from vllm.utils import make_tensor_with_pad
else:
from vllm.utils.torch_utils import make_tensor_with_pad
VOCAB_SIZE = 1024
NUM_OUTPUT_TOKENS = 20
MAX_PROMPT_SIZE = 100

33
tests/ut/worker/test_worker_v1.py
View File

@@ -6,10 +6,8 @@ import torch
from vllm.config import CacheConfig, ModelConfig, ParallelConfig, VllmConfig
from tests.ut.base import TestBase
from vllm_ascend.utils import vllm_version_is
init_cached_hf_modules_path = "vllm.utils.init_cached_hf_modules" if vllm_version_is(
"0.11.0") else "vllm.utils.import_utils.init_cached_hf_modules"
init_cached_hf_modules_path = "vllm.utils.import_utils.init_cached_hf_modules"
class TestNPUWorker(TestBase):
@@ -189,26 +187,15 @@ class TestNPUWorker(TestBase):
# Create NPUWorker instance
from vllm_ascend.worker.worker_v1 import NPUWorker
if vllm_version_is("0.11.0"):
with patch("vllm.utils.STR_DTYPE_TO_TORCH_DTYPE",
{"float32": torch.float32}):
worker = NPUWorker(
vllm_config=self.vllm_config_mock,
local_rank=self.local_rank,
rank=self.rank,
distributed_init_method=self.distributed_init_method,
is_driver_worker=self.is_driver_worker,
)
else:
with patch("vllm.utils.torch_utils.STR_DTYPE_TO_TORCH_DTYPE",
{"float32": torch.float32}):
worker = NPUWorker(
vllm_config=self.vllm_config_mock,
local_rank=self.local_rank,
rank=self.rank,
distributed_init_method=self.distributed_init_method,
is_driver_worker=self.is_driver_worker,
)
with patch("vllm.utils.torch_utils.STR_DTYPE_TO_TORCH_DTYPE",
{"float32": torch.float32}):
worker = NPUWorker(
vllm_config=self.vllm_config_mock,
local_rank=self.local_rank,
rank=self.rank,
distributed_init_method=self.distributed_init_method,
is_driver_worker=self.is_driver_worker,
)
# Verify cache_dtype is set to custom value
self.assertEqual(worker.cache_dtype, torch.float32)