A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[V1] Remove unnecessary check for main thread (#23298)
Signed-off-by: Robert Shaw <robshaw@redhat.com> Co-authored-by: Robert Shaw <robshaw@redhat.com>
Robert Shaw committed
Aug 21, 2025 at 10:08 UTC
c8e33c72c62e12bffe7a42f92e531dc5fb92f7cf
1 file changed
-6
vllm/engine/arg_utils.py
-6
@@ -8,7 +8,6 @@ import dataclasses
8
import functools
9
import json
10
import sys
11
-import threading
11
from dataclasses import MISSING, dataclass, fields, is_dataclass
12
from itertools import permutations
13
from typing import (TYPE_CHECKING, Annotated, Any, Callable, Dict, List,
@@ -1527,11 +1526,6 @@ class EngineArgs:
1526
#############################################################
1527
# Experimental Features - allow users to opt in.
1528
1530
- # Signal Handlers requires running in main thread.
1531
- if (threading.current_thread() != threading.main_thread()
1532
- and _warn_or_fallback("Engine in background thread")):
1533
- return False
1534
-
1529
if self.pipeline_parallel_size > 1:
1530
supports_pp = getattr(self.distributed_executor_backend,
1531
'supports_pp', False)