[Platform] Refactor Platform attention backend selection to avoid breakpoint for OOT platform (#30212)

Signed-off-by: Isotr0py <mozf@mail2.sysu.edu.cn>
Signed-off-by: Isotr0py <2037008807@qq.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
This commit is contained in:
Isotr0py
2025-12-15 17:36:07 +00:00
committed by GitHub
co-authored by gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
parent 970713d4a4
commit ec154c36ee
7 changed files with 73 additions and 137 deletions
+2 -10
View File
@@ -18,8 +18,8 @@ from vllm.logger import init_logger
if TYPE_CHECKING:
from torch.distributed import PrefixStore, ProcessGroup
from vllm.attention.selector import AttentionSelectorConfig
from vllm.config import VllmConfig
from vllm.config.cache import CacheDType
from vllm.inputs import ProcessorInputs, PromptType
from vllm.pooling_params import PoolingParams
from vllm.sampling_params import SamplingParams
@@ -226,15 +226,7 @@ class Platform:
def get_attn_backend_cls(
cls,
selected_backend: "AttentionBackendEnum",
head_size: int,
dtype: torch.dtype,
kv_cache_dtype: "CacheDType | None",
block_size: int,
use_mla: bool,
has_sink: bool,
use_sparse: bool,
use_mm_prefix: bool,
attn_type: str | None = None,
attn_selector_config: "AttentionSelectorConfig",
) -> str:
"""Get the attention backend class of a device."""
return ""