Skip to main content

Python class

DummyKVCache

DummyKVCache

class max.kv_cache.DummyKVCache(*args, **kwargs)

source

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.

Parameters:

alloc()

alloc(*args, **kwargs)

source

No-op.

Parameters:

Return type:

None

claim()

claim(request_id, replica_idx)

source

No-op.

Parameters:

Return type:

None

contains()

contains(request_id, replica_idx)

source

Returns True for any request.

Parameters:

Return type:

bool

get_metrics()

get_metrics(replica_idx)

source

Returns empty metrics.

Parameters:

replica_idx (int)

Return type:

KVCacheMetrics

get_num_host_pages()

get_num_host_pages(replica_idx)

source

Returns 1.

Parameters:

replica_idx (int)

Return type:

int

get_num_pages()

get_num_pages(replica_idx)

source

Returns 1.

Parameters:

replica_idx (int)

Return type:

int

get_num_used_host_pages()

get_num_used_host_pages(replica_idx)

source

Returns 1.

Parameters:

replica_idx (int)

Return type:

int

get_num_used_pages()

get_num_used_pages(replica_idx)

source

Returns 1.

Parameters:

replica_idx (int)

Return type:

int

get_pct_used_blocks_after_allocation()

get_pct_used_blocks_after_allocation(*args, **kwargs)

source

Returns a fixed low percentage (0.01).

Parameters:

Return type:

float

release()

release(request_id, replica_idx)

source

No-op.

Parameters:

Return type:

None

reset_metrics()

reset_metrics()

source

No-op.

Return type:

None

step()

step(*args, **kwargs)

source

No-op.

Parameters:

Return type:

None