From 5255ea71190d3c2386f5635303ec50f36487f243 Mon Sep 17 00:00:00 2001 From: Lucas Wilkinson Date: Fri, 14 Nov 2025 20:24:12 +0000 Subject: [PATCH] remove dcp assert Signed-off-by: Lucas Wilkinson --- vllm/v1/worker/gpu_model_runner.py | 10 ---------- 1 file changed, 10 deletions(-) diff --git a/vllm/v1/worker/gpu_model_runner.py b/vllm/v1/worker/gpu_model_runner.py index 9b3e5b668aab..d49331bc91e4 100644 --- a/vllm/v1/worker/gpu_model_runner.py +++ b/vllm/v1/worker/gpu_model_runner.py @@ -628,16 +628,6 @@ def _may_reorder_batch(self, scheduler_output: "SchedulerOutput") -> None: return if self.reorder_batch_threshold is not None: - # NOTE(lucas): currently no backend supports the custom masking - # required for DCP with q_len > 1, so we assert here. Remove this - # assert once the custom mask is support is added to FA3. - if ( - self.dcp_world_size > 1 - and envs.VLLM_ATTENTION_BACKEND != "FLASH_ATTN_MLA" - ): - assert self.reorder_batch_threshold == 1, ( - "DCP not support reorder_batch_threshold > 1 now." - ) reorder_batch_to_split_decodes_and_prefills( self.input_batch, scheduler_output,