diff --git a/vllm/v1/worker/gpu_model_runner.py b/vllm/v1/worker/gpu_model_runner.py
index c3d84ab377..bfdb0f7225 100644
--- a/vllm/v1/worker/gpu_model_runner.py
+++ b/vllm/v1/worker/gpu_model_runner.py
@@ -540,9 +540,6 @@ class GPUModelRunner(LoRAModelRunnerMixin):
         # because M (max_model_len) is not necessarily divisible by block_size.
         block_table_indices = (req_indices * self.max_num_blocks_per_req +
                                positions_np // self.block_size)
-        # NOTE(woosuk): We use torch.index_select instead of np.take here
-        # because torch.index_select is much faster than np.take for large
-        # tensors.
         block_table_cpu = self.input_batch.block_table.get_cpu_tensor()
         block_numbers = block_table_cpu.flatten()[block_table_indices].numpy()
         block_offsets = positions_np % self.block_size