Fix prefill OOM error in the case of large page size (#5081)

2025-05-05 16:02:55 -07:00
parent 683707c314
commit f8e460930a
3 changed files with 11 additions and 2 deletions
--- a/python/sglang/srt/mem_cache/chunk_cache.py
+++ b/python/sglang/srt/mem_cache/chunk_cache.py
@@ -24,9 +24,11 @@ class ChunkCache(BasePrefixCache):
        self,
        req_to_token_pool: ReqToTokenPool,
        token_to_kv_pool_allocator: TokenToKVPoolAllocator,
+        page_size: int,
    ):
        self.req_to_token_pool = req_to_token_pool
        self.token_to_kv_pool_allocator = token_to_kv_pool_allocator
+        self.page_size = page_size

    def reset(self):
        pass