From 41cf8bbd6015783d3de4f6beb0f8201a8e1a87bb Mon Sep 17 00:00:00 2001 From: Jiangyun Zhu Date: Wed, 3 Jun 2026 20:17:46 +0800 Subject: [PATCH] enable breakable cudagraph by default (#39) Signed-off-by: zjy0516 --- vllm/config/vllm.py | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/vllm/config/vllm.py b/vllm/config/vllm.py index f753647081c..cc50e6bd469 100644 --- a/vllm/config/vllm.py +++ b/vllm/config/vllm.py @@ -1051,20 +1051,26 @@ class VllmConfig: ) self.compilation_config.mode = CompilationMode.NONE - # DeepSeek V4's model classes don't carry @support_torch_compile — + # For model classes don't carry @support_torch_compile — # the breakable cudagraph is the supported PIECEWISE path. Auto-enable # it unless the user has explicitly opted out via the env var. if ( self.model_config is not None and "VLLM_USE_BREAKABLE_CUDAGRAPH" not in os.environ and any( - a in ("DeepseekV4ForCausalLM", "DeepSeekV4MTPModel") + a + in ( + "DeepseekV4ForCausalLM", + "DeepSeekV4MTPModel", + "MiniMaxM3SparseForCausalLM", + "MiniMaxM3SparseForConditionalGeneration", + ) for a in self.model_config.architectures ) ): os.environ["VLLM_USE_BREAKABLE_CUDAGRAPH"] = "1" logger.info_once( - "Auto-enabling VLLM_USE_BREAKABLE_CUDAGRAPH=1 for DeepSeek V4. " + "Auto-enabling VLLM_USE_BREAKABLE_CUDAGRAPH=1. " "Set VLLM_USE_BREAKABLE_CUDAGRAPH=0 to opt out." )