xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-31 07:56:31 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Core] Minimize number of dict lookup in _maybe_evict_cached_block (#21281)

Browse Source 
Signed-off-by: Jialin Ouyang <Jialin.Ouyang@gmail.com>
This commit is contained in:
Jialin Ouyang 2025-07-21 22:37:34 -07:00 committed by GitHub
parent e7b2042681
commit af376ca19d
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 20 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

35
vllm/v1/core/block_pool.py
View File

@ -243,22 +243,27 @@ class BlockPool:
True if the block is evicted, False otherwise.
"""
block_hash = block.block_hash
if block_hash and block_hash in self.cached_block_hash_to_block:
block.reset_hash()
del self.cached_block_hash_to_block[block_hash][block.block_id]
if block_hash is None:
# The block doesn't have hash, eviction is not needed
return False
blocks_by_id = self.cached_block_hash_to_block.get(block_hash)
if blocks_by_id is None:
# block_hash not found in cached_block_hash_to_block,
# eviction is not needed
return False
block.reset_hash()
blocks_by_id.pop(block.block_id, None)
if blocks_by_id:
del self.cached_block_hash_to_block[block_hash]
if len(self.cached_block_hash_to_block[block_hash]) == 0:
del self.cached_block_hash_to_block[block_hash]
if self.enable_kv_cache_events:
# FIXME (Chen): Not sure whether we should return `hash_value`
# or `(hash_value, group_id)` here. But it's fine now because
# we disable hybrid kv cache manager when kv cache event is
# enabled, so there is only one group.
self.kv_event_queue.append(
BlockRemoved(block_hashes=[block_hash.get_hash_value()]))
return True
return False
if self.enable_kv_cache_events:
# FIXME (Chen): Not sure whether we should return `hash_value`
# or `(hash_value, group_id)` here. But it's fine now because
# we disable hybrid kv cache manager when kv cache event is
# enabled, so there is only one group.
self.kv_event_queue.append(
BlockRemoved(block_hashes=[block_hash.get_hash_value()]))
return True
def touch(self, blocks: tuple[list[KVCacheBlock], ...]) -> None:
"""Touch a block increases its reference count by 1, and may remove