mstar.engine.resources.kv#
Modules
What a model declares about a KV cache: its shape, its spec, its step. |
|
Page keys for cross-request prefix reuse. |
|
Page copies between KV caches: the device cache and the host page pool, or two device caches (CUDA IPC KV transfer). |
|
The KV plan: what KVManager.plan hands its dependents. |
|
The prefix index: which cached page holds which key, and which to drop first. |
|
Ring KV storage: a fixed horizon of frame slots, overwritten in place. |