[KV Connector] Opt DecodeBenchConnector into SupportsHMA (#41770)

Signed-off-by: Zijing Liu <liuzijing2014@gmail.com>
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Zijing Liu
2026-05-07 16:10:00 -07:00
committed by GitHub
co-authored by Claude Opus 4.7
parent 10ebb40d62
commit 09a7cc5ba9
@@ -40,7 +40,10 @@ from vllm.distributed.kv_transfer.kv_connector.v1 import (
KVConnectorBase_V1,
KVConnectorRole,
)
from vllm.distributed.kv_transfer.kv_connector.v1.base import KVConnectorMetadata
from vllm.distributed.kv_transfer.kv_connector.v1.base import (
KVConnectorMetadata,
SupportsHMA,
)
from vllm.logger import init_logger
from vllm.utils.math_utils import cdiv
from vllm.v1.attention.backend import AttentionMetadata
@@ -71,7 +74,7 @@ class DecodeBenchConnectorMetadata(KVConnectorMetadata):
reqs_to_fill: dict[str, tuple[tuple[list[int], ...], int]]
class DecodeBenchConnector(KVConnectorBase_V1):
class DecodeBenchConnector(KVConnectorBase_V1, SupportsHMA):
"""
A KV Connector for decode instance performance testing.
@@ -164,6 +167,17 @@ class DecodeBenchConnector(KVConnectorBase_V1):
self.connector_scheduler.request_finished(request)
return False, None
def request_finished_all_groups(
self,
request: "Request",
block_ids: tuple[list[int], ...],
) -> tuple[bool, dict[str, Any] | None]:
# HMA-enabled path: same cleanup as the single-group variant since
# this connector owns no external state per block.
assert self.connector_scheduler is not None
self.connector_scheduler.request_finished(request)
return False, None
class DecodeBenchConnectorScheduler:
"""Scheduler-side implementation for DecodeBenchConnector."""