EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[minor] simplify the TokenToKVPoolAllocator (#7414)

Browse Source
This commit is contained in:
Liangsheng Yin
2025-06-22 12:37:18 +08:00
committed by GitHub
parent b7a2df0a44
commit 05c9bc8956
14 changed files with 165 additions and 149 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
python/sglang/srt/mem_cache/hiradix_cache.py
View File

@@ -7,12 +7,12 @@ from typing import List, Optional
import torch
from sglang.srt.managers.cache_controller import HiCacheController
from sglang.srt.mem_cache.allocator import BaseTokenToKVPoolAllocator
from sglang.srt.mem_cache.base_prefix_cache import MatchResult
from sglang.srt.mem_cache.memory_pool import (
MHATokenToKVPool,
MLATokenToKVPool,
ReqToTokenPool,
TokenToKVPoolAllocator,
)
from sglang.srt.mem_cache.memory_pool_host import (
MHATokenToKVPoolHost,
@@ -28,7 +28,7 @@ class HiRadixCache(RadixCache):
def __init__(
self,
req_to_token_pool: ReqToTokenPool,
token_to_kv_pool_allocator: TokenToKVPoolAllocator,
token_to_kv_pool_allocator: BaseTokenToKVPoolAllocator,
tp_cache_group: torch.distributed.ProcessGroup,
page_size: int,
hicache_ratio: float,