biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Core] Use sha256 bytes instead of BlockHash to reduce GC overhead (#23673)

Browse Source 
Signed-off-by: linzebing <linzebing1995@gmail.com>
This commit is contained in:
Zebing Lin
2025-09-09 00:34:37 -04:00
committed by GitHub
parent bba1042c6f
commit 82dfb12e52
15 changed files with 298 additions and 283 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tests/v1/kv_connector/unit/utils.py
View File

@@ -13,6 +13,7 @@ from vllm.distributed.kv_transfer.kv_connector.factory import (
KVConnectorFactory)
from vllm.distributed.kv_transfer.kv_connector.v1.shared_storage_connector import ( # noqa
SharedStorageConnector)
from vllm.utils import sha256
from vllm.v1.core.kv_cache_manager import KVCacheBlocks
from vllm.v1.core.kv_cache_utils import (get_request_block_hasher,
init_none_hash)
@@ -127,11 +128,11 @@ def create_request(request_id: int,
use_all_1s_for_prompt_tokens: bool = False,
num_remote_blocks: int = 3,
block_size: int = 16,
hash_fn: Callable = hash) -> Request:
hash_fn: Callable = sha256) -> Request:
"""Make dummy request for testing."""
global _none_hash_initialized
if not _none_hash_initialized:
init_none_hash(hash)
init_none_hash(hash_fn)
_none_hash_initialized = True
kv_transfer_params: Optional[dict[str, Any]] = None