Skip to content

GitLab

Commit f42ed372 authored by zhuwenwen's avatar zhuwenwen
Browse files

暂时去掉profilling标志位，避免影响其他模型

parent 81301303
Show whitespace changes
Inline Side-by-side
Showing with 3 additions and 3 deletions
vllm/v1/worker/gpu_model_runner.py
View file @ f42ed372
...@@ -2611,8 +2611,8 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin): ...@@ -2611,8 +2611,8 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin):
def profile_run(self) -> None: def profile_run(self) -> None:
# set profiling flag to avoid torch compile # set profiling flag to avoid torch compile
set_profilling(True) # set_profilling(True)
self._sync_device() # self._sync_device()
# Profile with multimodal encoder & encoder cache. # Profile with multimodal encoder & encoder cache.
if self.supports_mm_inputs: if self.supports_mm_inputs:
...@@ -2681,7 +2681,7 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin): ...@@ -2681,7 +2681,7 @@ class GPUModelRunner(LoRAModelRunnerMixin, KVConnectorModelRunnerMixin):
del hidden_states, output del hidden_states, output
self.encoder_cache.clear() self.encoder_cache.clear()
gc.collect() gc.collect()
set_profilling(False) # set_profilling(False)
def capture_model(self) -> None: def capture_model(self) -> None:
if self.compilation_config.cudagraph_mode == CUDAGraphMode.NONE: if self.compilation_config.cudagraph_mode == CUDAGraphMode.NONE:
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment