vllm.v1.kv_offload.tiering ¶
Modules:
-
async_lookup–AsyncLookupManager: per-tier async lookup manager for secondary tier
-
base–Abstract interfaces and data types for the secondary tiering layer.
-
example– -
fs– -
manager–TieringOffloadingManager: Multi-tier KV cache offloading orchestrator.
-
obj– -
spec–TieringOffloadingSpec: Spec for multi-tier KV cache offloading.