From 1e31b07fa7371ee8a05f58a546882733afddb2a7 Mon Sep 17 00:00:00 2001 From: XiaoxinWang <963372609@qq.com> Date: Wed, 29 Oct 2025 10:30:23 +0800 Subject: [PATCH] fix qwen3next full graph break. (#3812) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit ### What this PR does / why we need it? fix qwen3next full graph break. linearattention doesnot has aclgraph_support attr，so change to cudagraph_support to support vllm. image ### Does this PR introduce _any_ user-facing change? ### How was this patch tested? - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/releases/v0.11.1 Signed-off-by: wangxiaoxin-sherie Co-authored-by: wangxiaoxin-sherie --- vllm_ascend/worker/model_runner_v1.py | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/vllm_ascend/worker/model_runner_v1.py b/vllm_ascend/worker/model_runner_v1.py index 522c3efc..3b1226df 100644 --- a/vllm_ascend/worker/model_runner_v1.py +++ b/vllm_ascend/worker/model_runner_v1.py @@ -3775,7 +3775,12 @@ class NPUModelRunner(LoRAModelRunnerMixin): for attn_group in self._attn_group_iterator(): builder = attn_group.get_metadata_builder() - if builder.aclgraph_support.value < min_ag_support.value: + graph_support = None + if hasattr(builder, 'aclgraph_support'): + graph_support = builder.aclgraph_support.value + else: + graph_support = builder.cudagraph_support.value + if graph_support < min_ag_support.value: min_ag_support = builder.aclgraph_support min_ag_builder_name = builder.__class__.__name__