For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).
Python class
DummyKVCache
DummyKVCache
class max.pipelines.kv_cache.DummyKVCache(*args, **kwargs)
Bases: PagedKVCacheManager
No-op KV cache implementation for testing or when cache is disabled.
Initializes the dummy cache with a single replica and no host swapping.
alloc()
alloc(*args, **kwargs)
No-op; returns an already-complete transfer (nothing to onload).
claim()
claim(request_id, replica_idx)
No-op.
contains()
contains(request_id, replica_idx)
Returns True for any request.
get_metrics_aggregated()
get_metrics_aggregated()
Returns empty aggregated metrics.
-
Return type:
get_num_disk_pages()
get_num_disk_pages(replica_idx)
Returns 1.
get_num_host_pages()
get_num_host_pages(replica_idx)
Returns 1.
get_num_pages()
get_num_pages(replica_idx)
Returns 1.
get_num_used_disk_pages()
get_num_used_disk_pages(replica_idx)
Returns 1.
get_num_used_host_pages()
get_num_used_host_pages(replica_idx)
Returns 1.
get_num_used_pages()
get_num_used_pages(replica_idx)
Returns 0; this cache never allocates, so no page is ever in use.
release()
release(request_id, replica_idx)
No-op.
reset_metrics()
reset_metrics()
No-op.
-
Return type:
-
None
step()
step(*args, **kwargs)
No-op.