[V1] Hybrid allocator without prefix caching (vllm-project#20661)

Signed-off-by: nopperl <54780682+nopperl@users.noreply.github.com>
Signed-off-by: Brayden Zhong <b8zhong@uwaterloo.ca>

Author: nopperl

vllm/v1/core/kv_cache_coordinator.py

@@ -171,6 +171,35 @@ def find_longest_cache_hit(
         pass
 
 
+class KVCacheCoordinatorNoPrefixCache(KVCacheCoordinator):
+    """
+    KV cache coordinator to use if prefix caching is disabled or unsupported.
+    In contrast to UnitaryKVCacheCoordinator and HybridKVCacheCoordinator,
+    supports arbitrary numbers of KV cache groups (including 0 groups).
+    Does not implement any features related to prefix caching.
+    """
+
+    def __init__(self, kv_cache_config: KVCacheConfig, max_model_len: int,
+                 use_eagle: bool, caching_hash_fn: Callable,
+                 enable_kv_cache_events: bool):
+        super().__init__(kv_cache_config, max_model_len, use_eagle, False,
+                         caching_hash_fn, enable_kv_cache_events)
+        self.num_single_type_manager = len(self.single_type_managers)
+
+    def get_num_common_prefix_blocks(self, request_id: str,
+                                     num_running_requests: int) -> list[int]:
+        return [0] * self.num_single_type_manager
+
+    def find_longest_cache_hit(
+        self,
+        block_hashes: list[BlockHash],
+        max_cache_hit_length: int,
+    ) -> tuple[tuple[list[KVCacheBlock], ...], int]:
+        blocks: tuple[list[KVCacheBlock], ...] = tuple(
+            [] for _ in range(self.num_single_type_manager))
+        return blocks, 0
+
+
 class UnitaryKVCacheCoordinator(KVCacheCoordinator):
     """
     KV cache coordinator for models with only one KV cache group. This is the
@@ -359,6 +388,10 @@ def get_kv_cache_coordinator(
         kv_cache_config: KVCacheConfig, max_model_len: int, use_eagle: bool,
         enable_caching: bool, caching_hash_fn: Callable,
         enable_kv_cache_events: bool) -> KVCacheCoordinator:
+    if not enable_caching:
+        return KVCacheCoordinatorNoPrefixCache(kv_cache_config, max_model_len,
+                                               use_eagle, caching_hash_fn,
+                                               enable_kv_cache_events)
     if len(kv_cache_config.kv_cache_groups) == 1:
         return UnitaryKVCacheCoordinator(kv_cache_config, max_model_len,
                                          use_eagle, enable_caching,