MCPcopy Create free account

hub / github.com/LMCache/LMCache / functions

Functions13,371 in github.com/LMCache/LMCache

↓ 1 callersMethodremove_put_task
(self, key: CacheEngineKey)
lmcache/v1/storage_backend/local_disk_backend.py:76
↓ 1 callersMethodremove_transfer_channel_client
Discard the client for ``peer_advertise_url`` and free its resources. Call this when reads from the peer are no longer needed (e.g. the
lmcache/v1/distributed/transfer_channel/abstract.py:122
↓ 1 callersFunctionrenderMpHitStats
(stats)
lmcache/lmcache_frontend/static/js/app.js:650
↓ 1 callersFunctionrenderMpOverview
(container, data)
lmcache/lmcache_frontend/static/js/app.js:508
↓ 1 callersFunctionrepeat_prompts
Repeat each prompt in the list for a specified number of times. The order of prompts in the output list depends on the mode. Args:
benchmarks/long_doc_qa/long_doc_qa.py:332
↓ 1 callersFunctionreplace_msgstr
Replace one entry's ``msgstr`` in ``lines``, in place.
tools/translate_docs_zh.py:169
↓ 1 callersMethodreport_status
Return blend module status information. Returns: A dict containing registered CB GPU instance IDs and per-instance KV
lmcache/v1/multiprocess/modules/blend.py:376
↓ 1 callersMethodreport_status
Return module-specific status information. Returns: A dict with a ``worker_liveness`` summary when reaping targets ar
lmcache/v1/multiprocess/modules/management.py:104
↓ 1 callersMethodrequest_finished
Called exactly once when a request has finished, before its blocks are freed. The connector may assumes responsibility for f
lmcache/integration/vllm/lmcache_mp_connector.py:1062
↓ 1 callersMethodrequirements_file
ROCm core requirements file.
setup_extensions/build_profiles/rocm.py:123
↓ 1 callersMethodreset
(self)
lmcache/integration/sglang/multi_process_adapter.py:533
↓ 1 callersMethodreset
Reset strategy and clear async queue.
lmcache/v1/lookup_client/record_strategies/base.py:298
↓ 1 callersMethodreset_histograms
Reset all Prometheus Histogram metrics by calling clear(). After clearing, re-initialize with labels so metrics remain visible.
lmcache/observability.py:1892
↓ 1 callersFunctionreset_observability_metrics
Reset observability metrics to their initial state.
lmcache/observability.py:1943
↓ 1 callersMethodreset_store
(cls)
tests/conftest.py:166
↓ 1 callersFunctionresolveInjectedPullSecrets
resolveInjectedPullSecrets returns the pull-secret references to inject: the per-pod annotation override (a comma-separated list of Secret names) when
operator/internal/webhook/pod_injector.go:317
↓ 1 callersFunctionresolvePayloadImage
resolvePayloadImage builds the "<repository>:<tag>" reference and pull policy for the payload init container from the engine's injection.payloadImage.
operator/internal/webhook/pod_injector.go:402
↓ 1 callersFunctionresolvePortMapping
resolvePortMapping replaces a "0:REMOTE" mapping (or bare "0") with "<picked>:REMOTE" using a kernel-allocated ephemeral port. Mappings that already s
operator/test/utils/portforward.go:110
↓ 1 callersFunctionresolveTargetContainer
resolveTargetContainer returns the index of the container to inject into and whether one was found. The per-pod annotation override (annotationName) t
operator/internal/webhook/pod_injector.go:286
↓ 1 callersFunctionresolve_block_stride_and_log_layout
Resolve the per-block stride for a KV layer group and log its layout. Single entry point for :class:`KVLayerGroupsManager` to obtain the ``bl
lmcache/v1/gpu_connector/utils.py:428
↓ 1 callersFunctionresolve_documents
Resolve ``--documents`` args into placeholder mapping and trailing text. Args: prompt_template: The prompt string containing placeholders
lmcache/cli/commands/query/_prompt.py:61
↓ 1 callersMethodresolve_profile
Resolve the active build profile. Returns ``None`` when building sdist, native extensions are disabled, GPU extensions are disabled,
setup_extensions/policy.py:142
↓ 1 callersMethodreverse_encode
(self, positions, q, k)
lmcache/v1/compute/positional_encoding.py:41
↓ 1 callersFunctionround_robin_pick_clients
()
examples/disagg_prefill_mp/disagg_proxy_server.py:149
↓ 1 callersMethodrun
Run warmup + benchmark, closing the HTTP client in the same loop. Blocks until the workload is complete. The ``RequestSender``'s http
lmcache/cli/commands/bench/engine_bench/workloads/base.py:74
↓ 1 callersMethodrun
(self)
lmcache/v1/internal_api_server/api_server.py:111
↓ 1 callersMethodrun
This is adapted from https://github.com/flashinfer-ai/flashinfer/blob/cc5ab77370dd9a489357a47e34315d9a8f3ad5fb/flashinfer/sparse.py#L1073
lmcache/v1/compute/attention/flash_infer_sparse.py:32
↓ 1 callersMethodrun_all_benchmarks
Run all benchmarks for specified thread counts.
benchmarks/microbenchmark/ttl_lock_benchmark.py:370
↓ 1 callersFunctionrun_benchmark
( host: str, port: int, chunk_mb: float, num_workers: int, num_keys: int, username: st
examples/kv_cache_reuse/remote_backends/resp/benchmark_resp_client.py:18
↓ 1 callersMethodrun_benchmark
Run the main benchmark
lmcache/tools/controller_benchmark/benchmark.py:479
↓ 1 callersFunctionrun_interactive
Run the interactive configuration flow. Walks the user through missing required items, offers gates for general and workload-specific setting
lmcache/cli/commands/bench/engine_bench/interactive/__init__.py:202
↓ 1 callersFunctionrun_l2_adapter_bench
Run the L2 adapter benchmark. Args: command: The owning :class:`BaseCommand` instance, used only to obtain a configured :clas
lmcache/cli/commands/bench/l2_adapter_bench/command.py:132
↓ 1 callersFunctionrun_precompute
(args)
benchmarks/rag/precompute.py:212
↓ 1 callersFunctionrun_rag
(args)
benchmarks/rag/rag.py:353
↓ 1 callersFunctionrun_server_bench
Centralized orchestrator: run the server bench loop. Args: command: The owning :class:`BaseCommand` instance, used to obtain
lmcache/cli/commands/bench/server_bench/command.py:201
↓ 1 callersFunctionrun_single_process
Run benchmark in a single process and return results
lmcache/tools/controller_benchmark/__main__.py:35
↓ 1 callersFunctionrun_trace_replay
Construct a StorageManager from *args* and drive replay. Produces three kinds of output: * Per-record stream: every dispatch is logged at IN
lmcache/cli/commands/trace/replay_command.py:100
↓ 1 callersFunctionsaveSidebarScroll
()
docs/source/_static/custom.js:240
↓ 1 callersMethodsave_caches
Save one encoder cache entry from vLLM into LMCache.
lmcache/integration/vllm/vllm_ec_adapter.py:145
↓ 1 callersFunctionsave_metadata
(path: str, tmp: str, metadata: bytes)
lmcache/v1/storage_backend/gds_backend.py:148
↓ 1 callersFunctionselectProxy
(proxy)
lmcache/lmcache_frontend/static/js/app.js:176
↓ 1 callersMethodselect_l1_deletions
Decide which keys to delete from L1 after successful L2 store. Args: keys: Keys that were successfully stored to L2.
lmcache/v1/distributed/storage_controllers/store_policy.py:80
↓ 1 callersMethodselect_store_targets
Decide which keys to store to which L2 adapters. Args: keys: Keys that were just written to L1 and are c
lmcache/v1/distributed/storage_controllers/store_policy.py:60
↓ 1 callersMethodsend_and_recv_all
Send structured data to all ranks and collect responses. The transport is responsible for serializing each element of msg bef
lmcache/v1/rpc/transport.py:29
↓ 1 callersMethodsend_caching_message
(self)
lmcache/usage_context.py:338
↓ 1 callersFunctionsend_cfg
(max_len)
tests/v1/storage_backend/test_pd_backend_async.py:824
↓ 1 callersMethodsend_engine_message
(self)
lmcache/usage_context.py:155
↓ 1 callersMethodsend_env_message
(self)
lmcache/usage_context.py:150
↓ 1 callersMethodsend_heartbeat
Send heartbeat via dedicated heartbeat DEALER socket Args: message: HeartbeatMsg to send Returns: Tuple of (
lmcache/tools/controller_benchmark/benchmark.py:388
↓ 1 callersMethodsend_init_side_msg
Send side messages during initialization. :param socket: The ZMQ socket used for sending the message. :param init_side_msg:
lmcache/v1/transfer_channel/abstract.py:92
↓ 1 callersFunctionsend_ping
Send a PING request and return the result. Args: mq_client: The message queue client. timeout: Seconds to wait for the server's r
lmcache/integration/vllm/vllm_multi_process_adapter.py:271
↓ 1 callersFunctionsend_request_to_decoder
Send a request to decoder service.
examples/disagg_prefill_mp/disagg_proxy_server.py:181
↓ 1 callersFunctionsend_request_to_prefiller
Send a request to prefiller with request-id header.
examples/disagg_prefill_mp/disagg_proxy_server.py:166
↓ 1 callersMethodserialize
(self)
lmcache/v1/protocol.py:224
↓ 1 callersMethodserialize
Serialize ``src`` group into ``dst`` byte buffer. Args: src: Source group with length :attr:`group_size`. Slots m
lmcache/v1/distributed/serde/multi.py:108
↓ 1 callersMethodserialize
Serialize src KV data into dst byte buffer (in-place). Args: src: Source MemoryObj containing KV-shaped data (read-locked).
lmcache/v1/distributed/serde/base.py:55
↓ 1 callersFunctionserialized_layout_desc
Compute a flat byte-buffer MemoryLayoutDesc for the serialized output. Returns a single-group uint8 layout whose size is determined by the se
lmcache/v1/distributed/serde/utils.py:17
↓ 1 callersFunctionserver_main
Run the benchmark server: register a buffer and serve its catalog. Args: cfg: The benchmark configuration (``role == "server"``). Ra
lmcache/tools/transfer_channel_benchmark/benchmark.py:94
↓ 1 callersFunctionserviceAccountToken
serviceAccountToken returns a token for the specified service account in the given namespace. It uses the Kubernetes TokenRequest API to generate a to
operator/test/e2e/e2e_test.go:225
↓ 1 callersFunctionserviceMonitorCRDInstalled
serviceMonitorCRDInstalled reports whether the Prometheus Operator's ServiceMonitor CRD is registered on the target cluster. Specs that only make sens
operator/test/e2e/smoke_helpers_test.go:253
↓ 1 callersMethodset_context
(self, context_strings)
examples/frontend/chat_session.py:28
↓ 1 callersMethodset_current_retrieve_stats
(self, stats: RetrieveRequestStats)
lmcache/observability.py:323
↓ 1 callersMethodset_freeze
Set freeze mode. When enabled, only local_cpu backend will be used for retrieval.
lmcache/v1/storage_backend/storage_manager.py:860
↓ 1 callersMethodset_health_monitor
Set the health monitor reference. This is called by LMCacheManager after creating the HealthMonitor to inject the reference
lmcache/v1/cache_engine.py:245
↓ 1 callersFunctionset_hi_multiplier
()
benchmarks/ttft-estimator/ttft-estimator.py:83
↓ 1 callersMethodset_internal_state
Tell the session is the 'offset' seconds after the start
benchmarks/multi_round_qa/multi-round-qa.py:296
↓ 1 callersMethodset_raw_device_for_testing
Replace the raw device handle used by this core. Args: raw_device: Object implementing the Rust raw-device methods.
lmcache/v1/storage_backend/raw_block/core.py:439
↓ 1 callersMethodset_used_size
Narrow this buffer's logical size to the first ``n`` bytes. Optional hook for callers that have just written ``n`` bytes into a buffe
lmcache/v1/memory_management.py:271
↓ 1 callersFunctionsetup
()
tests/benchmarks/test_xpu_layerwise_connector_benchmark.py:385
↓ 1 callersFunctionsetup
()
tests/benchmarks/test_xpu_connector_benchmark.py:287
↓ 1 callersMethodsetup
Initialize the lock objects.
benchmarks/microbenchmark/ttl_lock_benchmark.py:83
↓ 1 callersFunctionsetup_env
Set up LMCache environment variables for CacheBlend.
tests/v1/compute/attention/test_triton_sparse_e2e.py:32
↓ 1 callersFunctionsetup_environment_variables
Set up LMCache-related environment variables for the Rust raw block backend. Configures environment variables for LMCache including chunk size, s
examples/kv_cache_reuse/local_backends/rust_backend_offload.py:21
↓ 1 callersFunctionsetup_environment_variables
(vllm_version: str, use_disk: bool = False)
examples/kv_cache_reuse/local_backends/offload.py:17
↓ 1 callersFunctionsetup_environment_variables
( use_disk: bool = False, blend_special_str: str = " # # ", enable_sparse: bool = False, )
examples/blend_kv_v1/blend.py:20
↓ 1 callersMethodsetup_metrics
Setup Prometheus metrics for this strategy. Args: prometheus_logger: Prometheus logger instance to register metrics with
lmcache/v1/lookup_client/record_strategies/base.py:132
↓ 1 callersFunctionsetup_signal_handlers
Setup signal handlers for graceful shutdown
lmcache/v1/standalone/__main__.py:406
↓ 1 callersFunctionshm_open_pool_as_mmap
(name: str, nbytes: int)
lmcache/cli/commands/bench/server_bench/helpers.py:84
↓ 1 callersFunctionshould_use_lazy_alloc
Determine if lazy allocation should be used based on CUDA availability.
tests/v1/distributed/test_store_controller.py:70
↓ 1 callersFunctionshould_use_lazy_alloc
()
tests/v1/distributed/test_prefetch_controller.py:74
↓ 1 callersFunctionshould_use_lazy_alloc
()
tests/v1/distributed/test_report_status.py:41
↓ 1 callersFunctionshould_use_lazy_alloc
Determine if lazy allocation should be used based on CUDA availability.
tests/v1/distributed/serde/test_serde_e2e.py:50
↓ 1 callersFunctionshutdown
()
tests/v1/test_remote_mla_worker_id_as0.py:164
↓ 1 callersMethodshutdown
Shutdown the connector by delegating to LMCacheManager.
lmcache/integration/vllm/vllm_v1_adapter.py:1340
↓ 1 callersMethodshutdown_async
Async shutdown — safe to call from a coroutine on the event loop. Args: wait: If True, drain the queue before shutting down.
lmcache/v1/storage_backend/job_executor/pq_executor.py:193
↓ 1 callersMethodslave_for
( self, service_name, socket_timeout=None, username=None, password=None, **kwargs )
tests/conftest.py:258
↓ 1 callersFunctionslot_identity_from_encoded_key
Return the stable slot-header identity for an encoded key. Args: encoded: Encoded raw-block key string. namespace: Key namespace
lmcache/v1/storage_backend/raw_block/key_codec.py:137
↓ 1 callersMethodsnapshot_indexed_keys
Return a detached snapshot of encoded keys currently in the index.
lmcache/v1/storage_backend/raw_block/core.py:588
↓ 1 callersMethodstart
(self)
tests/v1/test_health_monitor_fallback_recovery.py:78
↓ 1 callersMethodstart
(self)
examples/online_session/openai_chat_completion_client.py:75
↓ 1 callersMethodstart
(self)
lmcache/v1/distributed/storage_controllers/eviction_controller.py:52
↓ 1 callersMethodstart
(self)
lmcache/v1/internal_api_server/api_server.py:116
↓ 1 callersMethodstart_all
(self)
lmcache/v1/cache_controller/controller_manager.py:504
↓ 1 callersMethodstart_load_caches
Load needed encoder caches from LMCache into vLLM encoder_cache.
lmcache/integration/vllm/vllm_ec_adapter.py:115
↓ 1 callersMethodstart_monitoring
Start the background monitoring thread.
lmcache/v1/pin_monitor.py:216
↓ 1 callersMethodstep
Execute one step of the workload. Args: time_offset: seconds since benchmark start. Returns: Next wakeup tim
lmcache/cli/commands/bench/engine_bench/workloads/base.py:47
↓ 1 callersMethodstep
(self, timestamp: float, request_executor: RequestExecutor)
benchmarks/multi_round_qa/multi-round-qa.py:314
↓ 1 callersMethodstep
(self, timestamp: float, executor: RequestExecutor)
benchmarks/multi_round_qa/multi-round-qa.py:448
↓ 1 callersMethodstep
(self, timestamp: float, executor: RequestExecutor)
benchmarks/rag/rag.py:287
↓ 1 callersMethodstop
(self)
tests/v1/test_health_monitor_fallback_recovery.py:87
← previousnext →4,601–4,700 of 13,371, ranked by callers