xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-15 07:24:57 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Core] Avoid repeated len(block_token_ids) check in hash_request_tokens (#21781)

Browse Source 
Signed-off-by: linzebing <linzebing1995@gmail.com>
This commit is contained in:
Zebing Lin 2025-08-01 03:23:29 -04:00 committed by GitHub
parent b4e081cb15
commit e0f63e4a35
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 2 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
vllm/v1/core/kv_cache_utils.py
View File

@ -567,12 +567,10 @@ def hash_request_tokens(hash_function: Any, block_size: int,
ret = [] ret = []
parent_block_hash_value = None parent_block_hash_value = None
for start in range(0, len(token_ids), block_size): # Only full blocks will be hashed
for start in range(0, len(token_ids) - block_size + 1, block_size):
end = start + block_size end = start + block_size
block_token_ids = token_ids[start:end] block_token_ids = token_ids[start:end]
# Do not hash the block if it is not full.
if len(block_token_ids) < block_size:
break
if req_need_extra_keys: if req_need_extra_keys:
# MM and LoRA requests need extra keys for block-hash computation. # MM and LoRA requests need extra keys for block-hash computation.