attention调用方式

在vllm/worker/model_runner.py中，class GPUModelRunnerBase的初始化过程决定使用attention的类型

		self.attn_backend = get_attn_backend(
            self.model_config.get_head_size(),
            self.model_config.get_sliding_window(),
            self.model_config.dtype,
            self.kv_cache_dtype,
            self.block_size,
            self.model_config.is_attention_free,) if needs_attn_backend else None
        if self.attn_backend:
            self.attn_state = self.attn_backend.get_state_cls()(
                weakref.proxy(self))
        else:
            self.attn_state = CommonAttentionState(weakref.proxy(self))

跳转到vllm/attention/selector.py中，get_attn_backend定义如下：

@lru_cache(maxsize=None)
def get_attn_backend(
    head_size: int,
    sliding_window: Optional[int],
    dtype: torch.dtype,
    kv_cache_dtype: Optional[str],
    block_size: int,
    is_attention_free: bool,  # model中不包含attention计算
    is_blocksparse: bool = False,
) -> Type[AttentionBackend]:
    """Selects which attention backend to use and lazily imports it."""

    if is_blocksparse:
        logger.info("Using BlocksparseFlashAttention backend.")
        from vllm.attention.backends.blocksparse_attn import BlocksparseFlashAttentionBackend
        return BlocksparseFlashAttentionBackend

    backend = which_attn_to_use(head_size, sliding_window, dtype,
                                kv_cache_dtype, block_size, is_attention_free)
    if backend == _Backend.FLASH_ATTN:
        from vllm.attention.backends.flash_attn import FlashAttentionBackend
        return FlashAttentionBackend
    if backend == _Backend.XFORMERS:
        logger.info("Using XFormers backend.")
        from vllm.attention.backends.xformers import XFormersBackend
        return XFormersBackend
    elif backend == _Backend.ROCM_FLASH:
        logger.info("Using ROCmFlashAttention backend.")
        from vllm.attention.backends.rocm_flash_attn import ROCmFlashAttentionBackend
        return ROCmFlashAttentionBackend
    elif backend == _Backend.TORCH_SDPA:
        assert is_cpu(), RuntimeError("Torch SDPA backend is only used for the CPU device.")
        logger.info("Using Torch SDPA backend.")
        from vllm.attention.backends.torch_sdpa import TorchSDPABackend
        return TorchSDPABackend
    elif backend == _Backend.OPENVINO:
        logger.info("Using OpenVINO Attention backend.")
        from vllm.attention.backends.openvino import OpenVINOAttentionBackend
        return OpenVINOAttentionBackend
    elif backend == _Backend.IPEX:
        assert is_xpu(), RuntimeError("IPEX attention backend is only used for the XPU device.")
        logger.info("Using IPEX attention backend.")
        from vllm.attention.backends.ipex_attn import IpexAttnBackend
        return IpexAttnBackend
    elif backend == _Backend.FLASHINFER:
        logger.info("Using Flashinfer backend.")
        from vllm.attention.backends.flashinfer import FlashInferBackend
        return FlashInferBackend
    elif backend == _Backend.PALLAS:
        logger.info("Using Pallas backend.")
        from vllm.attention.backends.pallas import PallasAttentionBackend
        return PallasAttentionBackend
    elif backend == _Backend.NO_ATTENTION:
        from vllm.attention.backends.placeholder_attn import PlaceholderAttentionBackend
        return PlaceholderAttentionBackend
    else:
        raise ValueError("Invalid attention backend.")