[KVCache] Support Pluggable KVCacheSpec (#37505)

Signed-off-by: MengqingCao <cmq0113@163.com>
Signed-off-by: Mengqing Cao <cmq0113@163.com>
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: zjy0516 <riverclouds.zhu@qq.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
This commit is contained in:
Mengqing Cao
2026-06-03 09:05:16 -07:00
committed by GitHub
co-authored by zjy0516 mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
parent df7252c343
commit 0c6631f02a
11 changed files with 698 additions and 59 deletions
+4
View File
@@ -52,6 +52,7 @@ from vllm.v1.core.kv_cache_utils import (
)
from vllm.v1.core.sched.interface import PauseState, SchedulerInterface
from vllm.v1.core.sched.output import SchedulerOutput
from vllm.v1.core.single_type_kv_cache_manager import register_all_kvcache_specs
from vllm.v1.engine import (
EEP_NOTIFICATION_CALL_ID,
EEPNotificationType,
@@ -235,6 +236,9 @@ class EngineCore:
def _initialize_kv_caches(self, vllm_config: VllmConfig) -> KVCacheConfig:
start = time.time()
# register all kvcache specs in enginecore process.
register_all_kvcache_specs(vllm_config)
# Get all kv cache needed by the model
kv_cache_specs = self.model_executor.get_kv_cache_specs()