allocate_blocks_cache

Applicable Products

Note: For Atlas A2 training products/Atlas A2 inference products, only the Atlas 800I A2 inference server and A200I A2 Box heterogeneous subrack are supported.

Function Description

In the PagedAttention scenario, allocates the cache with multiple blocks. After the cache is successfully allocated, the memory can be released through deallocate_blocks_cache.

Prototype

1

allocate_blocks_cache(cache_desc: CacheDesc, blocks_cache_key: Optional[BlocksCacheKey] = None)

Parameters

Example

1
2
3
4

from llm_datadist import *
...
blocks_cache_key = BlocksCacheKey(1, 0)
blocks_cache = cache_manager.allocate_blocks_cache(cache_desc, blocks_cache_key)

Returns

In normal cases, the cache is returned.

If the input data type is incorrect, the TypeError or ValueError exception is reported.

If the execution time exceeds the value of sync_kv_timeout, LLMException is thrown.

Constraints

The memory pool must be configured before this API call.