[Chore] Separate out system utilities from vllm.utils (#27201)

Signed-off-by: dongbo910220 <1275604947@qq.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
This commit is contained in:
dongbo910220
2025-10-22 20:25:25 +00:00
committed by GitHub
co-authored by Cyrus Leung
parent 5beacce2ea
commit a0003b56b0
25 changed files with 199 additions and 155 deletions
+1 -1
View File
@@ -25,7 +25,7 @@ from vllm.distributed.parallel_state import (
initialize_model_parallel,
)
from vllm.platforms import current_platform
from vllm.utils import update_environment_variables
from vllm.utils.system_utils import update_environment_variables
from ..models.registry import HF_EXAMPLE_MODELS
from ..utils import (
+1 -1
View File
@@ -31,7 +31,7 @@ from vllm.model_executor.layers.quantization.utils.w8a8_utils import (
GroupShape,
)
from vllm.platforms import current_platform
from vllm.utils import update_environment_variables
from vllm.utils.system_utils import update_environment_variables
from ..utils import has_module_attribute, multi_gpu_test
from .backend import TestBackend
+1 -1
View File
@@ -29,7 +29,7 @@ from vllm.distributed.parallel_state import (
from vllm.model_executor.layers.layernorm import RMSNorm
from vllm.model_executor.layers.quantization.utils.w8a8_utils import Fp8LinearOp
from vllm.platforms import current_platform
from vllm.utils import update_environment_variables
from vllm.utils.system_utils import update_environment_variables
from ..utils import multi_gpu_test
from .backend import TestBackend
+1 -1
View File
@@ -15,7 +15,7 @@ from vllm.distributed.parallel_state import (
get_tp_group,
init_distributed_environment,
)
from vllm.utils import update_environment_variables
from vllm.utils.system_utils import update_environment_variables
def distributed_run(fn, world_size):
@@ -23,7 +23,7 @@ from vllm.distributed.parallel_state import (
initialize_model_parallel,
)
from vllm.platforms import current_platform
from vllm.utils import update_environment_variables
from vllm.utils.system_utils import update_environment_variables
torch.manual_seed(42)
random.seed(44)
+1 -1
View File
@@ -18,7 +18,7 @@ from vllm.distributed.parallel_state import (
graph_capture,
init_distributed_environment,
)
from vllm.utils import update_environment_variables
from vllm.utils.system_utils import update_environment_variables
def distributed_run(fn, world_size):
+1 -1
View File
@@ -10,8 +10,8 @@ import torch.distributed as dist
from vllm.distributed.device_communicators.shm_broadcast import MessageQueue
from vllm.distributed.utils import StatelessProcessGroup
from vllm.utils import update_environment_variables
from vllm.utils.network_utils import get_open_port
from vllm.utils.system_utils import update_environment_variables
def get_arrays(n: int, seed: int = 0) -> list[np.ndarray]:
+1 -1
View File
@@ -23,7 +23,7 @@ from vllm.distributed.parallel_state import (
from vllm.engine.arg_utils import EngineArgs
from vllm.engine.llm_engine import LLMEngine
from vllm.platforms import current_platform
from vllm.utils import update_environment_variables
from vllm.utils.system_utils import update_environment_variables
torch.manual_seed(42)
random.seed(44)
+1 -1
View File
@@ -10,8 +10,8 @@ import torch
import vllm.envs as envs
from vllm.distributed.device_communicators.pynccl import PyNcclCommunicator
from vllm.distributed.utils import StatelessProcessGroup
from vllm.utils import update_environment_variables
from vllm.utils.network_utils import get_open_port
from vllm.utils.system_utils import update_environment_variables
from vllm.utils.torch_utils import cuda_device_count_stateless
from ..utils import multi_gpu_test
+1 -1
View File
@@ -13,7 +13,7 @@ from vllm.distributed.parallel_state import (
)
from vllm.model_executor.layers.mamba.mamba_mixer2 import Mixer2RMSNormGated
from vllm.platforms import current_platform
from vllm.utils import update_environment_variables
from vllm.utils.system_utils import update_environment_variables
@multi_gpu_test(num_gpus=2)
+1 -1
View File
@@ -19,8 +19,8 @@ from vllm.model_executor.models.vision import (
run_dp_sharded_vision_model,
)
from vllm.platforms import current_platform
from vllm.utils import update_environment_variables
from vllm.utils.network_utils import get_open_port
from vllm.utils.system_utils import update_environment_variables
pytestmark = pytest.mark.cpu_test
+19
View File
@@ -0,0 +1,19 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
import tempfile
from pathlib import Path
from vllm.utils.system_utils import unique_filepath
def test_unique_filepath():
temp_dir = tempfile.mkdtemp()
path_fn = lambda i: Path(temp_dir) / f"file_{i}.txt"
paths = set()
for i in range(10):
path = unique_filepath(path_fn)
path.write_text("test")
paths.add(path)
assert len(paths) == 10
assert len(list(Path(temp_dir).glob("*.txt"))) == 10
-13
View File
@@ -19,7 +19,6 @@ from vllm.transformers_utils.detokenizer_utils import convert_ids_list_to_tokens
from vllm.utils import (
FlexibleArgumentParser,
bind_kv_cache,
unique_filepath,
)
from ..utils import create_new_process_for_each_test, flat_product
@@ -466,18 +465,6 @@ def test_load_config_file(tmp_path):
os.remove(str(config_file_path))
def test_unique_filepath():
temp_dir = tempfile.mkdtemp()
path_fn = lambda i: Path(temp_dir) / f"file_{i}.txt"
paths = set()
for i in range(10):
path = unique_filepath(path_fn)
path.write_text("test")
paths.add(path)
assert len(paths) == 10
assert len(list(Path(temp_dir).glob("*.txt"))) == 10
def test_flat_product():
# Check regular itertools.product behavior
result1 = list(flat_product([1, 2, 3], ["a", "b"]))
+1 -1
View File
@@ -21,8 +21,8 @@ from vllm.distributed.parallel_state import (
from vllm.model_executor.layers.mamba.mamba_mixer2 import MambaMixer2
from vllm.platforms import current_platform
from vllm.sampling_params import SamplingParams
from vllm.utils import update_environment_variables
from vllm.utils.mem_constants import GiB_bytes
from vllm.utils.system_utils import update_environment_variables
from vllm.v1.core.kv_cache_utils import estimate_max_model_len, get_kv_cache_configs
from vllm.v1.core.sched.output import CachedRequestData, NewRequestData, SchedulerOutput
from vllm.v1.kv_cache_interface import (