biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Refactor]A simple device-related refactor (#11163)

Browse Source 
Signed-off-by: noemotiovon <noemotiovon@gmail.com>
Co-authored-by: noemotiovon <noemotiovon@gmail.com>
This commit is contained in:
Chenguang Li
2024-12-13 21:39:00 +08:00
committed by GitHub
parent 969da7d70b
commit d1fa714cb1
7 changed files with 51 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
vllm/utils.py
View File

@@ -24,7 +24,6 @@ from asyncio import FIRST_COMPLETED, AbstractEventLoop, Future, Task
from collections import UserDict, defaultdict
from collections.abc import Iterable, Mapping
from functools import lru_cache, partial, wraps
from platform import uname
from typing import (TYPE_CHECKING, Any, AsyncGenerator, Awaitable, Callable,
Dict, Generic, Hashable, List, Literal, Optional,
OrderedDict, Set, Tuple, Type, TypeVar, Union, overload)
@@ -344,12 +343,6 @@ def random_uuid() -> str:
return str(uuid.uuid4().hex)
@lru_cache(maxsize=None)
def in_wsl() -> bool:
# Reference: https://github.com/microsoft/WSL/issues/4071
return "microsoft" in " ".join(uname()).lower()
def make_async(
func: Callable[P, T],
executor: Optional[concurrent.futures.Executor] = None
@@ -729,25 +722,7 @@ def print_warning_once(msg: str) -> None:
@lru_cache(maxsize=None)
def is_pin_memory_available() -> bool:
if in_wsl():
# Pinning memory in WSL is not supported.
# https://docs.nvidia.com/cuda/wsl-user-guide/index.html#known-limitations-for-linux-cuda-applications
print_warning_once("Using 'pin_memory=False' as WSL is detected. "
"This may slow down the performance.")
return False
elif current_platform.is_xpu():
print_warning_once("Pin memory is not supported on XPU.")
return False
elif current_platform.is_neuron():
print_warning_once("Pin memory is not supported on Neuron.")
return False
elif current_platform.is_hpu():
print_warning_once("Pin memory is not supported on HPU.")
return False
elif current_platform.is_cpu() or current_platform.is_openvino():
return False
return True
return current_platform.is_pin_memory_available()
class DeviceMemoryProfiler: