{"record":{"id":"991e587cf89f365f","repo":"vllm-project/vllm","slug":"cudagraphmode-cudagraph-mode-name-is-not-support","errorCode":null,"errorMessage":"CUDAGraphMode.{cudagraph_mode.name} is not supported with {min_cg_attn_backend} backend (support: {min_cg_support}); please try cudagraph_mode=PIECEWISE, and make sure compilation mode is VLLM_COMPILE","messagePattern":"CUDAGraphMode\\.(.+?) is not supported with (.+?) backend \\(support: (.+?)\\); please try cudagraph_mode=PIECEWISE, and make sure compilation mode is VLLM_COMPILE","errorType":"validation","errorClass":"ValueError","httpStatus":null,"severity":"error","filePath":"vllm/config/compilation.py","lineNumber":1403,"sourceCode":"            return CUDAGraphMode.NONE\n\n        # Check cudagraph for mixed batch is supported\n        if (\n            cudagraph_mode.mixed_mode() == CUDAGraphMode.FULL\n            and min_cg_support != AttentionCGSupport.ALWAYS\n        ):\n            msg = (\n                f\"CUDAGraphMode.{cudagraph_mode.name} is not supported \"\n                f\"with {min_cg_attn_backend} backend (support: \"\n                f\"{min_cg_support})\"\n            )\n            if min_cg_support == AttentionCGSupport.NEVER:\n                # if not supported any full cudagraphs, just raise it.\n                msg += (\n                    \"; please try cudagraph_mode=PIECEWISE, and \"\n                    \"make sure compilation mode is VLLM_COMPILE\"\n                )\n                raise ValueError(msg)\n\n            # attempt to resolve the full cudagraph related mode\n            if self.splitting_ops_contain_attention():\n                msg += \"; setting cudagraph_mode=FULL_AND_PIECEWISE\"\n                cudagraph_mode = CUDAGraphMode.FULL_AND_PIECEWISE\n            else:\n                msg += \"; setting cudagraph_mode=FULL_DECODE_ONLY\"\n                cudagraph_mode = CUDAGraphMode.FULL_DECODE_ONLY\n            logger.warning(msg)\n\n        # check that if we are doing decode full-cudagraphs it is supported\n        if (\n            cudagraph_mode.decode_mode() == CUDAGraphMode.FULL\n            and min_cg_support == AttentionCGSupport.NEVER\n        ):\n            msg = (\n                f\"CUDAGraphMode.{cudagraph_mode.name} is not supported \"\n                f\"with {min_cg_attn_backend} backend (support: \"","sourceCodeStart":1385,"sourceCodeEnd":1421,"githubUrl":"https://github.com/vllm-project/vllm/blob/c794754062d49a8fdb63ab3c5215b488b865030c/vllm/config/compilation.py#L1385-L1421","documentation":"Error \"CUDAGraphMode.{cudagraph_mode.name} is not supported with {min_cg_attn_backend} backend (support: {min_cg_support}); please try cudagraph_mode=PIECEWISE, and make sure compilation mode is VLLM_COMPILE\" thrown in vllm-project/vllm.","triggerScenarios":"Raised at vllm/config/compilation.py:1403 when validation fails: CUDAGraphMode not supported with the selected attention backend; try cudagraph_mode=PIECEWISE with compilation mode VLLM_COMPILE. Typically triggered by an incompatible or incomplete vLLM configuration, an unsupported platform/backend combination, or a runtime resource/dependency that is missing.","commonSituations":"Commonly encountered at vllm/config/compilation.py:1403 during vLLM startup/config validation or runtime setup when: (1) conflicting CLI flags or config fields are combined, (2) the current platform (CUDA/ROCm/CPU/XPU) or installed optional packages do not support the requested feature, or (3) a required value is absent or out of range. Resolve by correcting the configuration as described in the message, or by selecting a supported alternative.","solutions":["Set cudagraph_mode=PIECEWISE in the compilation config.","Ensure compilation mode is VLLM_COMPILE (-O3).","Or select an attention backend with stronger CUDA graph support."],"exampleFix":null,"handlingStrategy":null,"validationCode":null,"typeGuard":null,"tryCatchPattern":null,"preventionTips":[],"tags":[],"backgroundTag":null,"analyzedSha":"c794754062d49a8fdb63ab3c5215b488b865030c","analyzedAt":"2026-08-14T21:17:39.825Z","schemaVersion":2},"datasetVersion":"2026-08-15T22:17:37.221Z"}