vllm-project/vllm · error · ValueError

Mamba SSU algorithm selection is only supported with the Fla

Error message

Mamba SSU algorithm selection is only supported with the FlashInfer backend. Please set `--mamba-backend flashinfer`, or omit `--mamba-ssu-algorithm`.

What it means

Error "Mamba SSU algorithm selection is only supported with the FlashInfer backend. Please set `--mamba-backend flashinfer`, or omit `--mamba-ssu-algorithm`." thrown in vllm-project/vllm.

Source

Thrown at vllm/config/mamba.py:77

    @classmethod
    def validate_backend_before(cls, value: Any) -> Any:
        """Enable parsing of the `backend` enum type from string."""
        if isinstance(value, str):
            return MambaBackendEnum[value.upper()]
        return value

    def validate_ssu_algorithm(self) -> None:
        if self.ssu_algorithm is None:
            return
        valid_algorithms = get_args(MambaSSUAlgorithm)
        if self.ssu_algorithm not in valid_algorithms:
            valid = ", ".join(valid_algorithms)
            raise ValueError(
                f"Unknown Mamba SSU algorithm: '{self.ssu_algorithm}'. "
                f"Valid options are: {valid}"
            )
        if self.backend != MambaBackendEnum.FLASHINFER:
            raise ValueError(
                "Mamba SSU algorithm selection is only supported with the "
                "FlashInfer backend. Please set `--mamba-backend flashinfer`, "
                "or omit `--mamba-ssu-algorithm`."
            )

    def __post_init__(self):
        self.validate_ssu_algorithm()
        if self.enable_stochastic_rounding:
            from vllm.platforms import current_platform

            if not current_platform.is_cuda():
                raise ValueError(
                    "Stochastic rounding for Mamba cache is only supported "
                    "on NVIDIA CUDA platforms. Please do not specify  "
                    "`--enable-mamba-cache-stochastic-rounding`."
                )
            if (
                self.backend == MambaBackendEnum.TRITON

View on GitHub (pinned to c794754062)

Solutions

1. Set --mamba-backend flashinfer when using --mamba-ssu-algorithm.
2. Or omit --mamba-ssu-algorithm.

When it happens

Trigger: Raised at vllm/config/mamba.py:77 when validation fails: Mamba SSU algorithm selection is only supported with the FlashInfer backend. Typically triggered by an incompatible or incomplete vLLM configuration, an unsupported platform/backend combination, or a runtime resource/dependency that is missing.

Common situations: Commonly encountered at vllm/config/mamba.py:77 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.

AI-assisted analysis of vllm-project/vllm@c794754062 (2026-08-14). Data as JSON: /api/errors/9bf5216038debc02. Report an issue: GitHub.