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 11, 01:15 UTC
[UX] improve profiler error message (#31125)
Signed-off-by: Boyuan Feng <boyuan@meta.com>
Boyuan Feng committed
Dec 22, 2025 at 08:45 UTC
8dd0db687b40a2fe7827975d93a296c9070aaba1
1 file changed
+6
-1
vllm/v1/worker/gpu_worker.py
+6
-1
@@ -634,7 +634,12 @@ class Worker(WorkerBase):
634
635
def profile(self, is_start: bool = True):
636
if self.profiler is None:
637
- raise RuntimeError("Profiling is not enabled.")
637
+ raise RuntimeError(
638
+ "Profiling is not enabled. Please set --profiler-config to enable "
639
+ "profiling. Example: "
640
+ "'--profiler-config.profiler=torch --profiler-config.torch_profiler_dir"
641
+ "=YOUR_DIR_PATH_TO_DUMP_TRACE'"
642
+ )
643
if is_start:
644
self.profiler.start()
645
else: