MCPcopy Create free account

hub / github.com/LMCache/LMCache / functions

Functions13,371 in github.com/LMCache/LMCache

↓ 2 callersMethodclear_session
End all open spans for *session_id* and remove them. Intended for use in ``shutdown()`` to clean up leaked spans when the EventBus st
lmcache/v1/mp_observability/subscribers/tracing/span_registry.py:111
↓ 2 callersMethodclear_worker_kv
Clear all KV chunks for a specific worker and location. Returns True if successful, False if worker not found.
lmcache/v1/cache_controller/utils.py:648
↓ 2 callersMethodclose
(self)
tests/v1/test_remote_mla_worker_id_as0.py:38
↓ 2 callersMethodclose
(self)
tests/v1/multiprocess/test_engine_driven_transfer.py:456
↓ 2 callersMethodclose
Shut down the thread pool.
tests/v1/storage_backend/test_valkey_connector.py:93
↓ 2 callersMethodclose
Close the StorageManager, stop the bus, and close the reader. Idempotent. The order matters: the StorageManager may publish teardown
lmcache/cli/commands/trace/_driver.py:226
↓ 2 callersMethodclose
Deregister the cuFile handle and close the fd.
lmcache/v1/gpu_connector/_cufile_async.py:365
↓ 2 callersMethodclose
(self)
lmcache/v1/transfer_channel/nixl_channel.py:596
↓ 2 callersMethodclose
(self)
lmcache/v1/distributed/l2_adapters/s3_l2_adapter.py:781
↓ 2 callersMethodclose
(self)
lmcache/v1/distributed/l2_adapters/p2p_l2_adapter.py:364
↓ 2 callersMethodclose
(self)
lmcache/v1/distributed/l2_adapters/fs_l2_adapter.py:436
↓ 2 callersMethodclose
(self)
lmcache/v1/lookup_client/lmcache_async_lookup_client.py:290
↓ 2 callersMethodclose
(self)
lmcache/v1/lookup_client/lmcache_lookup_client.py:175
↓ 2 callersMethodclose
Close file handles.
lmcache/v1/lookup_client/record_strategies/file_hash.py:125
↓ 2 callersMethodclose
Stop the reaper, if one is running.
lmcache/v1/multiprocess/modules/management.py:123
↓ 2 callersMethodclose
(self)
lmcache/v1/server/storage_backend/local_backend.py:69
↓ 2 callersMethodclose
(self)
lmcache/v1/rpc/zmq_transport.py:143
↓ 2 callersMethodclose
(self)
lmcache/v1/storage_backend/storage_manager.py:1385
↓ 2 callersMethodclose
Clean up resources
lmcache/v1/storage_backend/connector/fs_connector.py:401
↓ 2 callersMethodclose_lookup_client
Close the current lookup client. Returns: dict: Result with old client type info.
lmcache/v1/manager.py:333
↓ 2 callersMethodclose_lookup_server
Close the current lookup server. Returns: dict: Result with old server type info.
lmcache/v1/manager.py:400
↓ 2 callersFunctioncollect_field
Read one ``msgid``/``msgstr`` value plus any continuation lines. Returns: The joined string value, and the next line number to read.
tools/translate_docs_zh.py:117
↓ 2 callersMethodcommit_store
(self, *_args: Any, **_kwargs: Any)
tests/v1/multiprocess/test_engine_driven_transfer.py:453
↓ 2 callersMethodcompress
Execute a compress operation with error handling.
lmcache/v1/cache_controller/executor.py:159
↓ 2 callersFunctioncompute_v_scales
Compute FP32 V scales according to the requested scope. Args: v: the V tensor. Layout-agnostic; `head_axis` and `page_axis`
lmcache/v1/kv_codec/asym_k16_v8.py:69
↓ 2 callersFunctionconcrete_shape
Numeric shape of a format; ``size(label)`` gives each axis's dimension. E.g. ``NL_X_TWO_NB_BS_NH_HS`` with ``NL=32, NB=2048, BS=16, NH=8, HS=128`
lmcache/v1/gpu_connector/kv_format/specs/base.py:62
↓ 2 callersMethodconcrete_shape_str
``describe_shape`` with real dims, e.g. ``80 x [2, 2048, 128, 8, 128]``.
lmcache/v1/gpu_connector/kv_format/specs/base.py:172
↓ 2 callersFunctioncoordinatorMetricsPort
coordinatorMetricsPort returns the configured metrics port (default 9090).
operator/internal/resources/coordinator_service.go:63
↓ 2 callersFunctioncounter_delta
Return the per-tagged-key delta between two snapshots.
tests/v1/mp_observability/subscribers/metrics/counter_helpers.py:55
↓ 2 callersFunctioncounter_delta
Compute ``after - before`` for every metric name in either snapshot.
tests/v1/mp_observability/subscribers/metrics/otel_setup.py:85
↓ 2 callersMethodcreate_backends
Create new storage backends based on current config. Backends that are already present will be skipped **before** instantiat
lmcache/v1/storage_backend/storage_manager.py:1259
↓ 2 callersMethodcreate_bucket
Create the bucket if it does not already exist.
lmcache/v1/storage_backend/connector/hfbucket_connector.py:131
↓ 2 callersFunctioncreate_buffer_memory_obj
Create a tensor-backed memory object from a slice of a buffer. Args: buffer: Backing tensor containing the bytes to expose. offse
tests/v1/distributed/test_mooncake_store_l2_adapter.py:90
↓ 2 callersFunctioncreate_config_class
Create a configuration class dynamically with common functionality. Args: config_name: Name of the configuration class config_def
lmcache/v1/config_base.py:218
↓ 2 callersMethodcreate_controller
Create KVController with test settings.
tests/v1/cache_controller/conftest.py:154
↓ 2 callersFunctioncreate_l1_memory_desc
Create an L1 memory descriptor for a contiguous test buffer.
lmcache/cli/commands/bench/l2_adapter_bench/data.py:146
↓ 2 callersMethodcreate_lookup_client
Create a new lookup client using the current configuration. Args: dryrun: If True, only return what would be created wit
lmcache/v1/manager.py:350
↓ 2 callersMethodcreate_lookup_server
Create a new lookup server using the current configuration. Args: dryrun: If True, only return what would be created wit
lmcache/v1/manager.py:417
↓ 2 callersFunctioncreate_memory_objects
( kv_dim: int, nl: int, tokens_per_object: int, hidden_dim: int, num_objects: int, dty
tests/v1/test_mp_mem_kernels.py:145
↓ 2 callersMethodcreate_message
( self, benchmark: "ZMQControllerBenchmark", test_data: "TestData" )
lmcache/tools/controller_benchmark/handlers/evict.py:31
↓ 2 callersFunctioncreate_periodic_thread
Factory function to create a simple PeriodicThread. This is a convenience function for creating periodic threads without needing to defi
lmcache/v1/periodic_thread.py:540
↓ 2 callersFunctioncreate_storage_manager_with_config
Create storage manager with default configuration
lmcache/v1/check/utils.py:130
↓ 2 callersFunctioncreate_test_key
Create a test CacheEngineKey.
tests/v1/cache_controller/conftest.py:62
↓ 2 callersFunctioncreate_test_key
Create a test CacheEngineKey
tests/v1/storage_backend/test_p2p_backend_with_controller.py:209
↓ 2 callersFunctioncreate_test_memory_obj_for_storage_manager
Create a test MemoryObj for testing with StorageManager.
lmcache/v1/check/utils.py:100
↓ 2 callersFunctioncreate_test_metadata
Create LMCache metadata with a deterministic full chunk layout.
tests/v1/storage_backend/test_hfbucket_connector.py:125
↓ 2 callersFunctioncreate_test_metadata
()
tests/v1/storage_backend/test_local_cpu_backend.py:69
↓ 2 callersFunctioncreate_vllm_tensors
( engine_kv_format, nl: int, nb: int, bs: int, nh: int, hs: int, dtype: torch.dtyp
tests/v1/test_mp_mem_kernels.py:69
↓ 2 callersFunctioncreate_zero_vllm_tensors
( engine_kv_format, nl: int, nb: int, bs: int, nh: int, hs: int, dtype: torch.dtyp
tests/v1/test_mp_mem_kernels.py:107
↓ 2 callersMethoddata_ptrs
Return device pointers for ``layer_indices`` in kernel-expected order. Per-layer formats: one pointer per layer. SGLang two-list MHA: all K
lmcache/v1/gpu_connector/kv_format/specs/base.py:164
↓ 2 callersFunctiondecode_function_gpu
Given the path to the encoded KV bytestream, decode the KV cache Inputs: cdf: the cdf tensor, in shape [2 * nlayers, nchannels, bins
lmcache/storage_backend/serde/cachegen_decoder.py:78
↓ 2 callersFunctiondecode_po_string
Unwrap one quoted ``.po`` line into a plain Python string.
tools/translate_docs_zh.py:100
↓ 2 callersFunctiondecode_value
Decode a msgpack-deserialized value back to its native form. Raises: ValueError: If a wrapped value carries an unknown tag.
lmcache/v1/mp_observability/trace/codecs.py:114
↓ 2 callersMethoddecompress
Execute a decompress operation with error handling.
lmcache/v1/cache_controller/executor.py:222
↓ 2 callersMethoddefault_cxx_flags
ABI-aware default flags for downstream consumers.
setup_extensions/build_profiles/cuda.py:80
↓ 2 callersMethoddelete
(self, keys: list[ObjectKey])
lmcache/v1/distributed/l2_adapters/serde_wrapper.py:312
↓ 2 callersMethoddelete_files
Delete files in a single batch operation.
lmcache/v1/storage_backend/connector/hfbucket_connector.py:167
↓ 2 callersFunctiondequantize_v_fp8
Inverse of quantize_v_fp8. Returns out_dtype.
lmcache/v1/kv_codec/asym_k16_v8.py:200
↓ 2 callersMethoddescribe
(self)
lmcache/v1/cache_controller/message.py:571
↓ 2 callersMethoddeserialize
(s: bytes)
lmcache/v1/protocol.py:186
↓ 2 callersFunctiondiscover_subclasses
Yield concrete subclasses of *base_class* found in direct submodules of *package*. Args: package: The package to scan (module or dott
setup_extensions/policy.py:31
↓ 2 callersMethoddownload_files
Download files in a single batch operation.
lmcache/v1/storage_backend/connector/hfbucket_connector.py:159
↓ 2 callersFunctiondownsample_and_stage_block_ids
Cut the block id lists to skip the unneeded blocks in a chunk and stage it into GPU tensors for later use. This mainly targets the case where
lmcache/v1/multiprocess/modules/lmcache_driven_transfer.py:126
↓ 2 callersMethoddrop_instance_state
(self, instance_id: int)
tests/v1/multiprocess/test_worker_liveness.py:202
↓ 2 callersMethoddrop_key
Manually drop all HS layers for ``key``. Test/admin use.
lmcache/v1/hidden_state_store.py:293
↓ 2 callersMethoddynamic_delete_file
Delete a storage file from disk.
lmcache/v1/distributed/l2_adapters/nixl_store_dynamic_l2_adapter.py:257
↓ 2 callersFunctionencode_function
Given the path to the original key value cache, encode the KV cache
lmcache/storage_backend/serde/cachegen_encoder.py:279
↓ 2 callersFunctionencode_header
Serialize a header to msgpack bytes.
lmcache/v1/mp_observability/trace/format.py:115
↓ 2 callersFunctionencode_value
Encode ``v`` to a msgpack-friendly representation. Recursively encodes lists, tuples, and dicts. Tuples are preserved via a tag so they can
lmcache/v1/mp_observability/trace/codecs.py:79
↓ 2 callersMethodenqueue_register
Queue a best-effort register of stored ranges. Args: ranges: Wire-form ``StoreRange`` dicts to register.
lmcache/v1/mp_coordinator/blend_client.py:139
↓ 2 callersFunctionestimate_memory
Estimate memory requirements for caching. Args: unique_chunks: Number of unique chunks chunk_size: Number of tokens per chunk
examples/chunk_statistics/analyze_chunk_hashes.py:210
↓ 2 callersFunctionestimate_num_tokens
(text: str)
benchmarks/multi_round_qa/data_preprocessing.py:15
↓ 2 callersMethodestimate_serialized_size
(self, layout_desc: MemoryLayoutDesc)
tests/v1/distributed/serde/test_multi.py:376
↓ 2 callersMethodestimate_serialized_size
Return buffer size for fp8 output: exactly 1 byte per element. fp8 has a fixed 1:1 element-to-byte mapping, so the size is determinis
lmcache/v1/distributed/serde/fp8.py:50
↓ 2 callersMethodexists
(fname: Path)
lmcache/v1/storage_backend/connector/hf3fs_connector.py:593
↓ 2 callersMethodexists
Check if key exists in file system
lmcache/v1/storage_backend/connector/fs_connector.py:165
↓ 2 callersMethodexport_csv
Write per-qualname stats to a CSV file. Columns: ``qualname, count, errors, mean_ms, p50_ms, p90_ms, p99_ms, min_ms, max_ms``. One r
lmcache/cli/commands/trace/_stats.py:208
↓ 2 callersMethodexport_json
Write per-qualname stats + replay duration to a JSON file. Schema:: { "duration_s": <float>, "ops":
lmcache/cli/commands/trace/_stats.py:249
↓ 2 callersMethodexport_json
Write summary JSON with config and aggregated metrics.
lmcache/cli/commands/bench/engine_bench/stats.py:239
↓ 2 callersMethodextract_cell_value
(self, row: Any)
lmcache/v1/storage_backend/connector/bigtable_schema.py:37
↓ 2 callersFunctionextract_mm_features
Normalize multimodal information from a Request into parallel lists. This helper reads either: 1) `request.mm_features` (objects each
lmcache/integration/vllm/utils.py:286
↓ 2 callersFunctionextract_normal_content
Extract normal content from the response chunk. Args: chunk: The response chunk from OpenAI Chat Completions API. Returns:
benchmarks/long_doc_qa/long_doc_qa.py:133
↓ 2 callersFunctionextract_reasoning_content
Extract reasoning content from the response chunk. Args: chunk: The response chunk from OpenAI Chat Completions API. Returns:
benchmarks/long_doc_qa/long_doc_qa.py:110
↓ 2 callersFunctionextract_request_configs
(sampling_params: SamplingParams)
lmcache/integration/vllm/vllm_v1_adapter.py:98
↓ 2 callersFunctionextract_world_size_and_kv_rank
Convert the rank for the MLA.
lmcache/integration/vllm/lmcache_mp_connector_0201.py:87
↓ 2 callersFunctionextract_world_size_and_kv_rank
Convert the rank for the MLA.
lmcache/integration/vllm/lmcache_mp_connector_0180.py:80
↓ 2 callersFunctionfallbackMissingLanguagePage
* Fall back to the target language home page if the equivalent page is absent. * * @param {HTMLAnchorElement} link The language link to validate. *
docs/source/_static/custom.js:74
↓ 2 callersFunctionfetch_child_nodes_from_proxy
Fetch child nodes from a single proxy node. Args: proxy_node: Proxy dict with ``name``/``host``/``port`` keys. ``/api/nodes``
lmcache/lmcache_frontend/app.py:105
↓ 2 callersFunctionfindEnvVar
findEnvVar returns a pointer to the named env var (so callers can inspect ValueFrom) or nil if absent. We return *corev1.EnvVar rather than the value
operator/test/e2e/auth_smoke_test.go:148
↓ 2 callersMethodfind_key
Find a key in this worker's kv_store. Returns: (KVChunkInfo, peer_init_url, keys) if found, None otherwise. KVChunkInfo will
lmcache/v1/cache_controller/utils.py:197
↓ 2 callersMethodfind_key_simple
Find a key in this worker's kv_store, returning only KVChunkInfo. KVChunkInfo will have instance_id="" since WorkerNode doesn't know
lmcache/v1/cache_controller/utils.py:218
↓ 2 callersFunctionfirstPodForDeployment
firstPodForDeployment returns the first non-terminating pod selected by the Deployment's pod selector, or nil if none exist yet. Returns the typed Pod
operator/test/e2e/cacheblend_integration_smoke_test.go:255
↓ 2 callersFunctionflush_redis
Flush the test Redis database.
tests/v1/distributed/test_resp_l2_adapter_integration.py:103
↓ 2 callersFunctionfmt_bytes
Format a byte count as a human-readable string.
lmcache/cli/commands/describe.py:61
↓ 2 callersMethodformat
Render metrics as indented JSON. Args: title: The report title. sections: Ordered list of ``Section`` objects.
lmcache/cli/metrics/formatter.py:158
↓ 2 callersFunctionformatBytesStr
(bytes)
lmcache/lmcache_frontend/static/js/app.js:693
↓ 2 callersMethodfree
(self, memory_obj: MemoryObj, allocator_type: Optional[str] = None)
lmcache/v1/memory_management.py:2911
↓ 2 callersMethodfree
(self, memory_obj: MemoryObj, allocator_type: Optional[str] = None)
lmcache/v1/memory_management.py:3092
↓ 2 callersFunctionfree_pinned_numa_ptr
Non-CUDA equivalent of freeing a previously allocated NUMA pointer.
lmcache/python_ops_fallback.py:413
↓ 2 callersMethodfrom_bytes
(self, bs: bytes)
lmcache/storage_backend/serde/cachegen_decoder.py:154
← previousnext →2,601–2,700 of 13,371, ranked by callers