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 5, 16:00 UTC
[Kernel] [Helion] Use warning_once in get_gpu_name to prevent log spam (#38743)
Signed-off-by: Yanan Cao <gmagogsfm@gmail.com> Co-authored-by: Claude Sonnet 4 <noreply@anthropic.com>
Yanan Cao committed
Apr 1, 2026 at 21:30 UTC
73f48ce559e230fd0d738c52cb2e99bd0dd08754
1 file changed
+3
-4
vllm/kernels/helion/utils.py
+3
-4
@@ -2,11 +2,10 @@
2
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
3
"""Utility functions for Helion kernel management."""
4
5
-import logging
6
-
5
+from vllm.logger import init_logger
6
from vllm.platforms import current_platform
7
9
-logger = logging.getLogger(__name__)
8
+logger = init_logger(__name__)
9
10
# Maps known variant GPU names (after lowercase/underscore normalization)
11
# to their canonical form.
@@ -49,7 +48,7 @@ _GPU_NAME_ALIASES: dict[str, str] = {
48
49
def get_gpu_name(device_id: int | None = None) -> str:
50
if device_id is None:
52
- logger.warning(
51
+ logger.warning_once(
52
"get_gpu_name() called without device_id, defaulting to 0. "
53
"This may return the wrong device name in multi-node setups."
54
)