From 33a2262845a297a6a89279037f7a775311c72093 Mon Sep 17 00:00:00 2001 From: Rui Qiao Date: Tue, 15 Jul 2025 15:56:58 -0700 Subject: [PATCH] [Misc] Minor comment reorganization in capture_model() Signed-off-by: Rui Qiao --- vllm/v1/worker/gpu_model_runner.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/vllm/v1/worker/gpu_model_runner.py b/vllm/v1/worker/gpu_model_runner.py index af216539c90..7f9df2d1530 100644 --- a/vllm/v1/worker/gpu_model_runner.py +++ b/vllm/v1/worker/gpu_model_runner.py @@ -2273,12 +2273,12 @@ def capture_model(self) -> None: start_free_gpu_memory = torch.cuda.mem_get_info()[0] # Trigger CUDA graph capture for specific shapes. - # Capture the large shapes first so that the smaller shapes - # can reuse the memory pool allocated for the large shapes. with graph_capture(device=self.device): full_cg = self.full_cuda_graph - # Only rank 0 should print progress bar during capture + # Capture the large shapes first so that the smaller shapes + # can reuse the memory pool allocated for the large shapes. compilation_cases = reversed(self.cudagraph_batch_sizes) + # Only rank 0 should print progress bar during capture if is_global_first_rank(): compilation_cases = tqdm( list(compilation_cases),