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 8, 17:45 UTC
[Log] Optimize duplicate startup log (#33944)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Wentao Ye committed
Feb 6, 2026 at 12:49 UTC
67a746e87f52d8728a40fbead123c7c17c94d4dc
3 files changed
+10
-7
vllm/compilation/backends.py
+2
-1
@@ -263,11 +263,12 @@ class CompilerManager:
263
now = time.time()
264
elapsed = now - compilation_start_time
265
compilation_config.compilation_time += elapsed
266
- logger.info(
266
+ logger.info_once(
267
"Directly load the compiled graph(s) for compile range %s "
268
"from the cache, took %.3f s",
269
str(compile_range),
270
elapsed,
271
+ scope="local",
272
)
273
return compiled_graph
274
vllm/utils/deep_gemm.py
+5
-3
@@ -91,14 +91,16 @@ def is_deep_gemm_e8m0_used() -> bool:
91
_lazy_init()
92
93
if _fp8_gemm_nt_impl is None:
94
- logger.info_once("DeepGEMM E8M0 disabled: _fp8_gemm_nt_impl not found")
94
+ logger.info_once(
95
+ "DeepGEMM E8M0 disabled: _fp8_gemm_nt_impl not found", scope="local"
96
+ )
97
return False
98
99
if envs.VLLM_USE_DEEP_GEMM_E8M0:
98
- logger.info_once("DeepGEMM E8M0 enabled on current platform.")
100
+ logger.info_once("DeepGEMM E8M0 enabled on current platform.", scope="local")
101
return True
102
101
- logger.info_once("DeepGEMM E8M0 disabled on current configuration.")
103
+ logger.info_once("DeepGEMM E8M0 disabled on current configuration.", scope="local")
104
return False
105
106
vllm/v1/worker/gpu_worker.py
+3
-3
@@ -119,9 +119,6 @@ class Worker(WorkerBase):
119
120
self.use_v2_model_runner = envs.VLLM_USE_V2_MODEL_RUNNER
121
122
- if self.use_v2_model_runner:
123
- logger.info_once("Using V2 Model Runner", scope="global")
124
-
122
def sleep(self, level: int = 1) -> None:
123
from vllm.device_allocator.cumem import CuMemAllocator
124
@@ -240,6 +237,9 @@ class Worker(WorkerBase):
237
current_platform.dist_backend,
238
)
239
240
+ if self.use_v2_model_runner:
241
+ logger.info_once("Using V2 Model Runner", scope="local")
242
+
243
# Set random seed.
244
set_random_seed(self.model_config.seed)
245