# SPDX-License-Identifier: Apache-2.0 import contextlib import copy import hashlib import math import os import queue import random import struct import threading import time from collections import defaultdict, deque from collections.abc import Iterator from concurrent.futures import ThreadPoolExecutor from dataclasses import dataclass from typing import TYPE_CHECKING, Any, List, Optional, OrderedDict, Tuple import msgspec import numpy as np import numpy.typing as npt import torch import torch_npu import zmq from mooncake.engine import TransferEngine # type: ignore from vllm import envs from vllm.config import VllmConfig from vllm.distributed import get_pcp_group from vllm.distributed.kv_transfer.kv_connector.v1.base import ( KVConnectorBase_V1, KVConnectorHandshakeMetadata, KVConnectorMetadata, KVConnectorRole) from vllm.distributed.parallel_state import ( get_decode_context_model_parallel_rank, get_decode_context_model_parallel_world_size, get_pp_group, get_tensor_model_parallel_rank, get_tensor_model_parallel_world_size, get_tp_group) from vllm.distributed.utils import get_pp_indices from vllm.logger import logger from vllm.utils.network_utils import get_ip, make_zmq_path, make_zmq_socket from vllm.v1.core.sched.output import SchedulerOutput from vllm.v1.kv_cache_interface import KVCacheConfig from vllm.v1.request import RequestStatus from vllm_ascend.ascend_config import get_ascend_config, init_ascend_config from vllm_ascend.distributed.mooncake_transfer_engine import global_te from vllm_ascend.distributed.utils import get_transfer_timeout_value from vllm_ascend.utils import is_vl_model if TYPE_CHECKING: from vllm.attention.backends.abstract import AttentionMetadata from vllm.forward_context import ForwardContext from vllm.v1.core.kv_cache_manager import KVCacheBlocks from vllm.v1.request import Request GET_META_MSG = b"get_meta_msg" DONE_RECVING_MSG = b"done_recving_msg" class MooncakeAgentMetadata(msgspec.Struct, omit_defaults=True, dict=True): engine_id: str te_rpc_port: int kv_caches_base_addr: list[int] num_blocks: int local_ip: str = "" @dataclass class ReqMeta: local_block_ids: list[int] num_external_tokens: int remote_block_ids: list[int] remote_host: str remote_port: int remote_engine_id: str remote_pcp_size: int remote_dcp_size: int remote_multi_nodes_meta_mapping: dict[str, dict[str, Any]] num_prompt_blocks: int @dataclass class SizedDict(OrderedDict): def __init__(self, max_size=16000, *args, **kwargs): self.max_size = max_size super().__init__(*args, **kwargs) def __setitem__(self, key, value): super().__setitem__(key, value) if len(self) > self.max_size: self.popitem(last=False) def __getitem__(self, key): try: return super().__getitem__(key) except KeyError: value: dict[int, list[int]] = {} self[key] = value return value class KVCacheTaskTracker: def __init__(self): super().__init__() self.done_task_lock = threading.Lock() self.finished_requests: set[str] = set() # Only used in prefill node. Tracks requests whose kv blocks freeing is # intentionally delayed. Each entry is a tuple of (request_id, # timestamp). If a request remains in this queue for too long, it will # be force-freed. self.record_finished_requests: set[str] = set() self.delayed_free_requests: OrderedDict[str, float] = OrderedDict() def add_not_transfer_request(self, request_id: str): with self.done_task_lock: self.finished_requests.add(request_id) def update_done_task_count(self, request_id: str): with self.done_task_lock: self.finished_requests.add(request_id) if request_id in self.delayed_free_requests: self._remove_delayed_requests(request_id) else: self.record_finished_requests.add(request_id) def get_and_clear_finished_requests(self) -> set[str]: """ Get and clear the requests that have been completed. Returns: A set of request IDs that have been completed. """ with self.done_task_lock: finished_requests = self.finished_requests.copy() expired_requests = self._retrieve_expired_requests() finished_requests.update(expired_requests) self.finished_requests.clear() return finished_requests def add_delayed_request(self, request_id: str, delay_start_time: float): """Add a delayed free request.""" with self.done_task_lock: if request_id not in self.record_finished_requests: self.delayed_free_requests[request_id] = delay_start_time else: self.record_finished_requests.discard(request_id) def _retrieve_expired_requests(self): """Retrieve all expired delayed requests.""" expired_requests: set[str] = set() # Free delayed requests if they exceed the timeout current_time = time.time() while self.delayed_free_requests: request_id = next(iter(self.delayed_free_requests)) delay_start_time = self.delayed_free_requests[request_id] if (current_time - delay_start_time > envs.VLLM_NIXL_ABORT_REQUEST_TIMEOUT): self.delayed_free_requests.popitem(last=False) expired_requests.add(request_id) logger.info("Force freed request: %s", request_id) else: break return expired_requests def _remove_delayed_requests(self, request_id: str): """Remove all delayed free requests matching the given request_id.""" self.delayed_free_requests.pop(request_id) class KVCacheSendingThread(threading.Thread): def __init__(self, vllm_config: VllmConfig, tp_rank: int, prefill_tp_size: int, local_engine_id: str, side_channel_host: str, side_channel_port: int, metadata: MooncakeAgentMetadata, ready_event: threading.Event, kv_caches: dict[str, Any], pcp_rank: int): super().__init__(daemon=True, name="KVCacheSendingThread") self.tp_rank = tp_rank self.prefill_tp_size = prefill_tp_size self.pp_rank = get_pp_group().rank_in_group self.pp_size = vllm_config.parallel_config.pipeline_parallel_size self.tp_size = get_tensor_model_parallel_world_size() self.local_engine_id = local_engine_id self.side_channel_host = side_channel_host self.side_channel_port = side_channel_port self.metadata = metadata self.ready_event = ready_event self.kv_caches = kv_caches self.pcp_rank = pcp_rank self.port_send_num: dict[str, int] = {} self.task_tracker = KVCacheTaskTracker() def get_and_clear_finished_requests(self) -> set[str]: """ Get and clear the requests that have been completed. Returns: A set of request IDs that have been completed. """ return self.task_tracker.get_and_clear_finished_requests() def add_not_transfer_request(self, request_id: str): self.task_tracker.add_not_transfer_request(request_id) def add_delayed_request(self, request_id: str, delay_start_time: float): return self.task_tracker.add_delayed_request(request_id, delay_start_time) def run(self): """Run the thread to handle KV cache transfer requests.""" try: # Listen for new requests for metadata. NOTE(rob): we need each rank # to have a unique port. This hack to keeps us moving. We will # switch when moving to etcd or where we have a single ZMQ socket in # the scheduler. device_index = self.pp_rank * self.tp_size + self.tp_rank + self.pcp_rank * self.prefill_tp_size handshake_port = self.side_channel_port + device_index path = make_zmq_path("tcp", self.side_channel_host, handshake_port) logger.info("Starting listening on path: %s", path) with zmq_ctx(zmq.ROUTER, path) as sock: # type: ignore self.ready_event.set() self.run_busy_loop(sock) except Exception as e: logger.error("Mooncake KVCacheSendingThread exception: %s", e, exc_info=True) def run_busy_loop(self, sock: zmq.Socket): # type: ignore encoder = msgspec.msgpack.Encoder() encoded_data = encoder.encode(self.metadata) size_in_bytes = len(encoded_data) logger.debug("Size of encoded MooncakeAgentMetadata: %s bytes", str(size_in_bytes)) decoder = msgspec.msgpack.Decoder(type=tuple) while True: try: frames = sock.recv_multipart() if len(frames) < 2: logger.error("Invalid message format: %s", frames) continue identity = frames[0] payload = [f for f in frames[1:] if f != b""] if len(payload) != 1: logger.error("Invalid message format: %s", frames) continue msg = decoder.decode(payload[0]) if msg[0] == GET_META_MSG: sock.send_multipart((identity, b"", encoded_data)) elif msg[0] == DONE_RECVING_MSG: logger.debug("Got DONE_RECVING_MSG for request %s", msg[1]) request_id = msg[1] remote_port_send_num = msg[2] if remote_port_send_num: if request_id not in self.port_send_num: self.port_send_num[request_id] = 0 self.port_send_num[request_id] += 1 device_index = self.pp_rank * self.tp_size + \ self.tp_rank + self.pcp_rank * \ self.prefill_tp_size handshake_port = self.side_channel_port + device_index if self.port_send_num[request_id] >= \ remote_port_send_num[handshake_port]: self.task_tracker.update_done_task_count( request_id) del self.port_send_num[request_id] else: self.task_tracker.update_done_task_count(request_id) # Acknowledge the request completion. while True: try: # Send ACK to the sender. sock.send_multipart( (identity, b"", b"ACK"), flags=zmq.NOBLOCK) # type: ignore break except zmq.Again: # type: ignore # If the socket is not ready, retry sending. logger.debug( "Socket not ready, retrying to send ACK for " "request %s", msg[1]) time.sleep(0.01) else: logger.error( "Connection listener got unexpected message %s", msg) except Exception as e: logger.error("Connection listener got exception %s: %s", type(e), e) class KVCacheRecvingThread(threading.Thread): def __init__(self, tp_rank: int, tp_size: int, _prefill_pp_size: int, engine: TransferEngine, local_engine_id: str, local_handshake_port: int, side_channel_port: int, local_kv_caches_base_addr: list[int], block_len: list[int], ready_event: threading.Event, vllm_config: VllmConfig, kv_caches: dict[str, Any], prefill_pp_layer_partition: Optional[str] = None): super().__init__(daemon=True, name="KVCacheRecvingThread") self.tp_rank = tp_rank self.tp_size = tp_size self._prefill_pp_size = _prefill_pp_size self.local_engine_id = local_engine_id self.local_handshake_port = local_handshake_port self.side_channel_port = side_channel_port self.engine = engine self.ready_event = ready_event self.kv_caches = kv_caches self.kv_caches_base_addr: dict[str, dict[int, list[int]]] = \ SizedDict() self.kv_caches_base_addr[local_engine_id][local_handshake_port] = \ local_kv_caches_base_addr self.remote_te_port: dict[str, dict[int, int]] = \ SizedDict() self.block_len = block_len # TODO(jianzs): find a better way to detect MLA. self.use_mla = len(block_len) == 2 self.use_sparse = len(block_len) == 3 self.request_queue: queue.Queue[Any] = queue.Queue() self.executor = ThreadPoolExecutor(max_workers=32) self.task_tracker = KVCacheTaskTracker() self.encoder = msgspec.msgpack.Encoder() self.decoder = msgspec.msgpack.Decoder(MooncakeAgentMetadata) self.remote_sockets_lock = threading.Lock() self.remote_sockets: dict[ # type: ignore str, deque[zmq.Socket]] = defaultdict( # type: ignore deque) self.remote_poller = zmq.Poller() # type: ignore self.timeout = 1.0 # seconds self.vllm_config = vllm_config self.model_config = self.vllm_config.model_config self.block_size = self.vllm_config.cache_config.block_size self.num_layers = self.model_config.hf_text_config.num_hidden_layers self.pp_layer_indices = { rank: get_prefill_pp_indices(self.num_layers, rank, self._prefill_pp_size, prefill_pp_layer_partition) for rank in range(self._prefill_pp_size) } if not is_vl_model(vllm_config): if self.use_mla: self.k_head_dim = self.model_config.hf_text_config.kv_lora_rank self.v_head_dim = self.model_config.hf_text_config.qk_rope_head_dim self.num_kv_heads = 1 else: self.k_head_dim = self.model_config.hf_text_config.head_dim self.v_head_dim = self.model_config.hf_text_config.head_dim self.num_kv_heads = max( self.model_config.hf_text_config.num_key_value_heads // self.tp_size, 1) self.proc_not_transfer_request: dict[str, bool] = {} def add_request(self, request_id: str, local_block_ids: list[int], remote_block_ids: list[int], remote_engine_id: str, remote_host: str, remote_handshake_port: int, offset: int, tp_num_need_pulls: int, remote_port_send_num: dict[int, int] = {}, all_task_done: bool = False): """Add a new request to the queue for processing.""" logger.debug(f"Adding request {request_id} to the queue.") self.request_queue.put({ "request_id": request_id, "local_block_ids": local_block_ids, "remote_block_ids": remote_block_ids, "remote_engine_id": remote_engine_id, "remote_host": remote_host, "remote_handshake_port": remote_handshake_port, "offset": offset, "tp_num_need_pulls": tp_num_need_pulls, "remote_port_send_num": remote_port_send_num, "all_task_done": all_task_done }) def get_and_clear_finished_requests(self) -> set[str]: """ Get and clear the requests that have been completed. Returns: A set of request IDs that have been completed. """ return self.task_tracker.get_and_clear_finished_requests() def run(self): """Run the thread to handle KV cache transfer requests.""" self.ready_event.set() while True: try: request_data = self.request_queue.get() if request_data is None: logger.warning("Received a None request!") self.request_queue.task_done() continue self._handle_request(request_data) except Exception as e: logger.error(f"Error in KVCacheTransferThread: {e}") def _handle_request(self, req_meta: dict[str, Any]): request_id = req_meta["request_id"] remote_host = req_meta["remote_host"] remote_handshake_port = req_meta["remote_handshake_port"] remote_port_send_num = req_meta["remote_port_send_num"] all_task_done = req_meta["all_task_done"] try: logger.debug( f"Starting to transfer KV cache for request {request_id}.") self._transfer_kv_cache(req_meta) logger.debug( f"Finished transferring KV cache for request {request_id}.") except Exception as e: logger.error( "Failed to transfer KV cache for request " f"{request_id}: {e}", exc_info=True) finally: if all_task_done: self.task_tracker.update_done_task_count(request_id) if request_id in self.proc_not_transfer_request: del self.proc_not_transfer_request[request_id] self.request_queue.task_done() # Always send the done signal to the remote host to ensure proper # resource cleanup. Failing to do so may cause a memory leak on the # remote host. self._send_done_recv_signal(request_id, remote_host, remote_handshake_port, remote_port_send_num) self._send_done_signal_to_free_remote_port(request_id, remote_host, remote_port_send_num) def _send_done_signal_to_free_remote_port(self, request_id, remote_host, remote_port_send_num): if self.side_channel_port != self.local_handshake_port \ or not remote_port_send_num: return if request_id not in self.proc_not_transfer_request: self.proc_not_transfer_request[request_id] = True if self.proc_not_transfer_request[request_id]: for remote_port in remote_port_send_num.keys(): if remote_port_send_num[remote_port] == 0: self._send_done_recv_signal(request_id, remote_host, remote_port, remote_port_send_num) self.proc_not_transfer_request[request_id] = False def _transfer_kv_cache(self, req_meta: dict[str, Any]): """Handle a KV cache transfer request.""" request_id = req_meta["request_id"] remote_block_ids = req_meta["remote_block_ids"] local_block_ids = req_meta["local_block_ids"] remote_engine_id = req_meta["remote_engine_id"] remote_host = req_meta["remote_host"] remote_handshake_port = req_meta["remote_handshake_port"] offset = req_meta["offset"] tp_num_need_pulls = req_meta["tp_num_need_pulls"] # Full prefix cache hit: do not need to read remote blocks, just notify # P worker that we have the blocks we need. num_local_blocks = len(local_block_ids) if num_local_blocks == 0: return num_remote_blocks = len(remote_block_ids) assert num_local_blocks <= num_remote_blocks if num_local_blocks < num_remote_blocks: remote_block_ids = remote_block_ids[-num_local_blocks:] # Check if we have the remote metadata cached. if remote_engine_id not in self.kv_caches_base_addr or \ remote_handshake_port not in self.kv_caches_base_addr[remote_engine_id]: self._get_remote_metadata(remote_host, remote_handshake_port) if tp_num_need_pulls == 1: grouped_remote_block_ids, grouped_local_block_ids = \ group_concurrent_contiguous(remote_block_ids, local_block_ids) else: remote_block_ids = list(map(lambda x: [x], remote_block_ids)) local_block_ids = list(map(lambda x: [x], local_block_ids)) grouped_remote_block_ids, grouped_local_block_ids = remote_block_ids, local_block_ids num_transfer_groups = len(grouped_remote_block_ids) # tp_num_need_pulls: number of KV caches each Decode node needs to pull from each PP stage # Due to GQA, different KV heads are distributed across different ranks, so there are offsets # indicating which KV head to pull global_offset = offset # Global offset of request across all ranks prefill_pp_rank = offset // tp_num_need_pulls # PP rank where current request resides inner_offset = offset % tp_num_need_pulls # Offset within each PP stage remote_kv_caches_base_addrs = \ self.kv_caches_base_addr[remote_engine_id][remote_handshake_port] first_layer_index, end_layer_index = self.pp_layer_indices[ prefill_pp_rank] # support MTP layer kv transfer if self.vllm_config.speculative_config is not None: # all MTP layer use the same kv cache layer, so only need to transfer once if prefill_pp_rank == self._prefill_pp_size - 1: end_layer_index = end_layer_index + 1 num_cache_per_layer = len(list( self.kv_caches.values())[0]) # Number of KV caches per layer local_kv_caches_base_addrs = \ self.kv_caches_base_addr[self.local_engine_id][self.local_handshake_port][first_layer_index*num_cache_per_layer : end_layer_index*num_cache_per_layer] logger.debug( f"transfer kv cache first_layer_index:{first_layer_index} , end_layer_index:{end_layer_index}" ) remote_transfer_port = self.remote_te_port[remote_engine_id][ remote_handshake_port] num_blocks = len(local_block_ids) session_id = f"{remote_host}:{remote_transfer_port}" req_start_time = time.perf_counter() src_list, dst_list, length_list = [], [], [] for k, (src_layer_base_addr, dst_layer_base_addr) in enumerate( zip(local_kv_caches_base_addrs, remote_kv_caches_base_addrs)): if self.use_mla: block_len = (self.block_len[k % 2]) elif self.use_sparse: block_len = (self.block_len[k % 3]) else: block_len = (self.block_len[0]) inner_block_len = block_len // tp_num_need_pulls for remote_block_id, local_block_id in zip( grouped_remote_block_ids, grouped_local_block_ids): src = src_layer_base_addr + local_block_id[ 0] * block_len + inner_offset * inner_block_len dst = dst_layer_base_addr + remote_block_id[0] * inner_block_len length = inner_block_len * len(local_block_id) src_list.append(src) dst_list.append(dst) length_list.append(length) ret = self.engine.batch_transfer_sync_read(session_id, src_list, dst_list, length_list) if ret < 0: logger.error("Mooncake transfer failed for request %s", req_meta["request_id"]) raise RuntimeError(f"Mooncake transfer failed, ret: {ret}") req_end_time = time.perf_counter() req_transfer_elapsed = (req_end_time - req_start_time) * 1000 logger.info( "KV cache transfer for request %s took %.2f ms (%d groups," " %d blocks). local_ip %s local_device_id %s remote_session_id %s", request_id, req_transfer_elapsed, num_transfer_groups, num_blocks, get_ip(), self.tp_rank, session_id) # Determine if the current position is the offset position at the end of # the KV transmission. is_kv_transfer_end = ( global_offset == tp_num_need_pulls * self._prefill_pp_size - 1) need_cat_cache = tp_num_need_pulls > 1 and is_kv_transfer_end need_nz_cache = get_ascend_config().enable_kv_nz and is_kv_transfer_end if need_nz_cache or need_cat_cache: self.reformat_kv_cache(grouped_local_block_ids, tp_num_need_pulls, need_cat_cache, need_nz_cache) def reformat_kv_cache(self, block_ids: list[list[int]], tp_num_need_pulls: int, need_cat_cache: bool = False, need_nz_cache: bool = False): # Get necessary parameters k_cache = list(self.kv_caches.values())[0][0] dtype = k_cache.dtype device = k_cache.device flat_block_ids = [item for sublist in block_ids for item in sublist] block_ids_tensor = torch.tensor(flat_block_ids, dtype=torch.int32, device=device) num_blocks = len(flat_block_ids) num_tokens = num_blocks * self.block_size # Create device tensors for copy operations block_table = block_ids_tensor.view(1, -1) block_len_tensor = torch.tensor([num_tokens], dtype=torch.int32, device=device) seq_start_tensor = torch.tensor([0], dtype=torch.int32, device=device) # Initialize buffers k_buffer = torch.empty( (num_tokens, self.num_kv_heads, self.k_head_dim), dtype=dtype, device=device) v_buffer = torch.empty( (num_tokens, self.num_kv_heads, self.v_head_dim), dtype=dtype, device=device) # Create slot mapping for reshape operations block_offsets = torch.arange(0, self.block_size, dtype=torch.int32, device=device) slot_mapping = (block_offsets.reshape( (1, self.block_size)) + block_ids_tensor.reshape( (num_blocks, 1)) * self.block_size).flatten() # FIXME: Right now, if we skip synchronization at this point, the system # will crash in GQA scenarios. However, we still haven't identified the # root cause. torch.npu.synchronize() # Process each layer in the KV cache for _, (k_cache_layer, v_cache_layer) in self.kv_caches.items(): # Load cache data into buffers torch_npu.atb.npu_paged_cache_load(k_cache_layer, v_cache_layer, block_table, block_len_tensor, seq_starts=seq_start_tensor, key=k_buffer, value=v_buffer) if need_cat_cache: self._cat_kv_cache(k_cache_layer, v_cache_layer, k_buffer, v_buffer, tp_num_need_pulls, num_blocks, num_tokens, slot_mapping) if need_nz_cache: self._nz_kv_cache(k_cache_layer, v_cache_layer, k_buffer, v_buffer, slot_mapping) # Clean up buffers del k_buffer, v_buffer def _cat_kv_cache(self, k_cache_layer, v_cache_layer, k_buffer, v_buffer, tp_num_need_pulls, num_blocks, num_tokens, slot_mapping): def _transpose_kv_cache_between_head( buffer: torch.Tensor) -> torch.Tensor: buffer = buffer.view(num_blocks, tp_num_need_pulls, self.block_size, -1) buffer.transpose_(1, 2) return buffer.contiguous().view(num_tokens, self.num_kv_heads, -1) # Transpose KV cache k_buffer = _transpose_kv_cache_between_head(k_buffer) v_buffer = _transpose_kv_cache_between_head(v_buffer) # Reshape and cache the processed buffers torch_npu._npu_reshape_and_cache(key=k_buffer, value=v_buffer, key_cache=k_cache_layer, value_cache=v_cache_layer, slot_indices=slot_mapping) def _nz_kv_cache(self, k_cache_layer, v_cache_layer, k_buffer, v_buffer, slot_mapping): nz_fmt_last_dim = 16 k_cache_layer = k_cache_layer.view( -1, self.k_head_dim * self.num_kv_heads // nz_fmt_last_dim, self.block_size, nz_fmt_last_dim) v_cache_layer = v_cache_layer.view( -1, self.v_head_dim * self.num_kv_heads // nz_fmt_last_dim, self.block_size, nz_fmt_last_dim) torch_npu.npu_scatter_pa_kv_cache(k_buffer, v_buffer, k_cache_layer, v_cache_layer, slot_mapping) def _get_remote_metadata(self, remote_host: str, remote_handshake_port: int) -> None: """Get the metadata from the remote host.""" sock: Optional[zmq.Socket] = None # type: ignore try: sock = self._get_remote_socket(remote_host, remote_handshake_port) ensure_zmq_send(sock, self.encoder.encode((GET_META_MSG, ""))) metadata_bytes = ensure_zmq_recv(sock, self.remote_poller) agent_meta = self.decoder.decode(metadata_bytes) engine_id = agent_meta.engine_id assert engine_id != self.local_engine_id, ( f"Conflict engine id {engine_id} with local engine id " f"{self.local_engine_id}.") self.kv_caches_base_addr[engine_id][remote_handshake_port] = \ agent_meta.kv_caches_base_addr self.remote_te_port[engine_id][remote_handshake_port] = \ agent_meta.te_rpc_port finally: if sock is not None: self._return_remote_socket(sock, remote_host, remote_handshake_port) logger.debug("Returned socket to pool for %s:%d", remote_host, remote_handshake_port) def _send_done_recv_signal(self, request_id: str, remote_host: str, remote_handshake_port: int, remote_port_send_num: dict[int, int]): logger.debug("Sending done recving signal for request %s to %s:%d", request_id, remote_host, remote_handshake_port) sock: Optional[zmq.Socket] = None # type: ignore try: sock = self._get_remote_socket(remote_host, remote_handshake_port) data_bytes = self.encoder.encode( (DONE_RECVING_MSG, request_id, remote_port_send_num)) ensure_zmq_send(sock, data_bytes) resp = ensure_zmq_recv(sock, self.remote_poller, timeout=self.timeout) logger.debug( f"Received response for request {request_id}: {resp.decode('utf-8')}" ) if resp != b"ACK": logger.error("Failed to receive ACK for request %s from %s:%d", request_id, remote_host, remote_handshake_port) raise RuntimeError( f"Failed to receive ACK, resp: {resp.decode('utf-8')}") except RuntimeError as e: if isinstance(sock, zmq.Socket): # type: ignore sock.close() sock = None logger.warning( f"Unexpected error occurred in socket, {e}, closing the original channel" ) finally: if sock is not None: self._return_remote_socket(sock, remote_host, remote_handshake_port) logger.debug("Returned socket to pool for %s:%d", remote_host, remote_handshake_port) def _get_remote_socket( self, remote_host: str, remote_handshake_port: int) -> zmq.Socket: # type: ignore """Get a socket to the remote host.""" remote_path = make_zmq_path("tcp", remote_host, remote_handshake_port) with self.remote_sockets_lock: if self.remote_sockets[remote_path]: return self.remote_sockets[remote_path].popleft() ctx = zmq.Context() # type: ignore sock = make_zmq_socket( ctx=ctx, path=remote_path, socket_type=zmq.REQ, # type: ignore bind=False) sock.setsockopt( zmq.SNDTIMEO, # type: ignore int(self.timeout * 1000)) self.remote_poller.register(sock, zmq.POLLIN) # type: ignore return sock def _return_remote_socket( self, sock: zmq.Socket, # type: ignore remote_host: str, remote_handshake_port: int) -> None: """Return the remote socket to the pool.""" remote_path = make_zmq_path("tcp", remote_host, remote_handshake_port) with self.remote_sockets_lock: self.remote_sockets[remote_path].append(sock) class MooncakeConnectorMetadata(KVConnectorMetadata): def __init__(self): self.requests: dict[str, ReqMeta] = {} self.requests_to_send: dict[str, float] = {} def add_new_req( self, request_id: str, local_block_ids: list[int], num_external_tokens: int, kv_transfer_params: dict[str, Any], ): self.requests[request_id] = ReqMeta( local_block_ids=local_block_ids, num_external_tokens=num_external_tokens, remote_block_ids=kv_transfer_params["remote_block_ids"], remote_engine_id=kv_transfer_params["remote_engine_id"], remote_host=kv_transfer_params["remote_host"], remote_port=kv_transfer_params["remote_port"], remote_pcp_size=kv_transfer_params.get("remote_pcp_size", 1), remote_dcp_size=kv_transfer_params.get("remote_dcp_size", 1), remote_multi_nodes_meta_mapping=kv_transfer_params.get( "remote_multi_nodes_meta_mapping", {}), num_prompt_blocks=kv_transfer_params.get("num_prompt_blocks", 0), ) class MooncakeConnector(KVConnectorBase_V1): def __init__(self, vllm_config: VllmConfig, role: KVConnectorRole, kv_cache_config: Optional[KVCacheConfig] = None): assert vllm_config.kv_transfer_config is not None self.engine_id = vllm_config.kv_transfer_config.engine_id self._connector_metadata = MooncakeConnectorMetadata() if role == KVConnectorRole.SCHEDULER: self.connector_scheduler: Optional[MooncakeConnectorScheduler] = \ MooncakeConnectorScheduler(vllm_config, str(self.engine_id)) self.connector_worker: Optional[MooncakeConnectorWorker] = None elif role == KVConnectorRole.WORKER: self.connector_scheduler = None self.connector_worker = MooncakeConnectorWorker( vllm_config, str(self.engine_id)) ############################################################ # Scheduler Side Methods ############################################################ def get_num_new_matched_tokens( self, request: "Request", num_computed_tokens: int) -> tuple[int, bool]: assert self.connector_scheduler is not None return self.connector_scheduler.get_num_new_matched_tokens( request, num_computed_tokens) def update_state_after_alloc(self, request: "Request", blocks: "KVCacheBlocks", num_external_tokens: int): assert self.connector_scheduler is not None return self.connector_scheduler.update_state_after_alloc( request, blocks, num_external_tokens) def build_connector_meta( self, scheduler_output: SchedulerOutput, ) -> KVConnectorMetadata: assert self.connector_scheduler is not None return self.connector_scheduler.build_connector_meta(scheduler_output) def request_finished( self, request: "Request", block_ids: list[int], ) -> tuple[bool, Optional[dict[str, Any]]]: assert self.connector_scheduler is not None return self.connector_scheduler.request_finished(request, block_ids) ############################################################ # Worker Side Methods ############################################################ def register_kv_caches(self, kv_caches: dict[str, torch.Tensor]): assert self.connector_worker is not None self.connector_worker.register_kv_caches(kv_caches) def get_finished(self, finished_req_ids: set[str]) -> tuple[set[str], set[str]]: """Get the finished recving and sending requests.""" assert self.connector_worker is not None return self.connector_worker.get_finished() def start_load_kv(self, forward_context: "ForwardContext", **kwargs) -> None: assert self.connector_worker is not None assert isinstance(self._connector_metadata, MooncakeConnectorMetadata) self.connector_worker.start_load_kv(self._connector_metadata) def wait_for_layer_load(self, layer_name: str) -> None: """MooncakeConnector does not do layerwise saving.""" pass def save_kv_layer(self, layer_name: str, kv_layer: torch.Tensor, attn_metadata: "AttentionMetadata", **kwargs) -> None: """MooncakeConnector does not save explicitly.""" pass def wait_for_save(self): """MooncakeConnector does not save explicitly.""" pass def get_handshake_metadata(self) -> KVConnectorHandshakeMetadata | None: """ Get the KVConnector handshake metadata for this connector. This metadata is used for out-of-band connector handshake between P/D workers. Returns: KVConnectorHandshakeMetadata: the handshake metadata. None if no handshake metadata is available. """ assert self.connector_worker is not None return self.connector_worker.xfer_handshake_metadata def set_xfer_handshake_metadata( self, metadata: dict[int, KVConnectorHandshakeMetadata]) -> None: """ Set the KV connector handshake metadata for this connector. Args: metadata (dict): the handshake metadata to set. """ assert self.connector_scheduler is not None self.connector_scheduler.set_xfer_handshake_metadata(metadata) class MooncakeConnectorScheduler: """Implementation of Scheduler side methods""" def __init__(self, vllm_config: VllmConfig, engine_id: str): self.vllm_config = vllm_config init_ascend_config(vllm_config) self.ascend_config = get_ascend_config() self.block_size = vllm_config.cache_config.block_size self.engine_id = engine_id self.local_ip = get_ip() logger.info("Initializing Mooncake Scheduler %s", engine_id) self.side_channel_host = get_ip() self.pcp_size = vllm_config.parallel_config.prefill_context_parallel_size self.dcp_size = vllm_config.parallel_config.decode_context_parallel_size self.max_device_id = vllm_config.parallel_config.tensor_parallel_size * \ vllm_config.parallel_config.data_parallel_size * \ self.pcp_size * \ vllm_config.parallel_config.pipeline_parallel_size # Handshake base port self.side_channel_port = ( vllm_config.kv_transfer_config.kv_port + vllm_config.parallel_config.data_parallel_rank * vllm_config.parallel_config.tensor_parallel_size * vllm_config.parallel_config.pipeline_parallel_size * self.pcp_size) # Requests that need to start recv. # New requests are added by update_state_after_alloc in # the scheduler. Used to make metadata passed to Worker. self._reqs_need_recv: dict[str, tuple[Request, list[int], int]] = {} self._reqs_need_send: dict[str, float] = {} # master-slave meta information for cross-nodes self.multi_nodes_meta_mapping: dict[str, dict[str, Any]] = {} def get_num_new_matched_tokens( self, request: "Request", num_computed_tokens: int) -> tuple[int, bool]: """ For remote prefill, pull all prompt blocks from remote asynchronously relative to engine execution. Args: request (Request): the request object. num_computed_tokens (int): the number of locally computed tokens for this request Returns: * the number of tokens that can be loaded from the external KV cache beyond what is already computed. * true if the external KV cache tokens will be loaded asynchronously (between scheduler steps). """ params = request.kv_transfer_params logger.debug( "MooncakeConnector get_num_new_matched_tokens: " "num_computed_tokens=%s, kv_transfer_params=%s", num_computed_tokens, params) if params is not None and params.get("do_remote_prefill"): # Remote prefill: get all prompt blocks from remote. assert num_computed_tokens % self.block_size == 0 # Note: We use the full token count as transmit data here. count = max(len(request.prompt_token_ids) - num_computed_tokens, 0) return count, count > 0 # No remote prefill for this request. return 0, False def update_state_after_alloc(self, request: "Request", blocks: "KVCacheBlocks", num_external_tokens: int): params = request.kv_transfer_params logger.debug( "MooncakeConnector update_state_after_alloc: " "num_external_tokens=%s, kv_transfer_params=%s", num_external_tokens, params) if params is not None and params.get("do_remote_prefill"): if params.get("remote_block_ids"): if all(p in params for p in ("remote_engine_id", "remote_host", "remote_port")): local_block_ids = (blocks.get_unhashed_block_ids() if num_external_tokens > 0 else []) # Get unhashed blocks to pull from remote. self._reqs_need_recv[request.request_id] = ( request, local_block_ids, num_external_tokens) else: logger.warning( "Got invalid KVTransferParams: %s. This " "request will not utilize KVTransfer", params) else: assert num_external_tokens == 0 # Only trigger 1 KV transfer per request. params["do_remote_prefill"] = False def build_connector_meta( self, scheduler_output: SchedulerOutput, ) -> KVConnectorMetadata: meta = MooncakeConnectorMetadata() # Loop through scheduled reqs and convert to ReqMeta. for req_id, (req, block_ids, num_external_tokens) in self._reqs_need_recv.items(): assert req.kv_transfer_params is not None # For the case where there are no remote blocks to pull # (block_ids is empty), we don't need to schedule # an async read on the worker side. meta.add_new_req( request_id=req_id, local_block_ids=block_ids, num_external_tokens=num_external_tokens, kv_transfer_params=req.kv_transfer_params, ) # Clear the list once workers start the transfers self._reqs_need_recv.clear() meta.requests_to_send = self._reqs_need_send self._reqs_need_send = {} return meta def request_finished( self, request: "Request", block_ids: list[int], ) -> tuple[bool, Optional[dict[str, Any]]]: """ Once a request is finished, determine whether request blocks should be freed now or will be sent asynchronously and freed later. """ params = request.kv_transfer_params logger.debug( "MooncakeConnector request_finished, request_status=%s, " "kv_transfer_params=%s", request.status, params) if (params is None or not params.get("do_remote_decode") or request.status != RequestStatus.FINISHED_LENGTH_CAPPED): return False, None computed_block_ids = block_ids delay_free_blocks = len(computed_block_ids) > 0 if delay_free_blocks: logger.info("Delaying free of %d blocks for request %s", len(computed_block_ids), request.request_id) self._reqs_need_send[request.request_id] = time.time() num_prompt_blocks = math.ceil( len(request.prompt_token_ids) / self.block_size) return delay_free_blocks, dict( do_remote_prefill=True, do_remote_decode=False, remote_block_ids=computed_block_ids, remote_engine_id=self.engine_id, remote_host=self.side_channel_host, remote_port=self.side_channel_port, remote_pcp_size=self.pcp_size, remote_dcp_size=self.dcp_size, last_token_id=request.output_token_ids[-1], remote_multi_nodes_meta_mapping=self.multi_nodes_meta_mapping, num_prompt_blocks=num_prompt_blocks, ) def set_xfer_handshake_metadata( self, metadata: dict[int, KVConnectorHandshakeMetadata]) -> None: """ Set the KV connector handshake metadata for this connector. Args: metadata (dict): the handshake metadata to set. """ for local_rank, rank_metadata in metadata.items(): self.multi_nodes_meta_mapping[str(local_rank)] = { "host": rank_metadata.local_ip, "engine_id": rank_metadata.engine_id, } class MooncakeConnectorWorker: """Implementation of Worker side methods""" def __init__(self, vllm_config: VllmConfig, engine_id: str): self._get_prefill_decode_size(vllm_config) os.environ["ASCEND_TRANSFER_TIMEOUT"] = str( get_transfer_timeout_value()) if self._prefill_tp_size < self._decode_tp_size: raise ValueError( f"prefill_tp_size: {self._prefill_tp_size} must be greater than" f" or equal to the decode_tp_size: {self._decode_tp_size}") # Metadata. self.vllm_config = vllm_config self.ascend_config = get_ascend_config() self.engine_id = engine_id self.tp_rank = get_tensor_model_parallel_rank() self.tp_size = vllm_config.parallel_config.tensor_parallel_size self.tp_group = get_tp_group() self.pp_rank = get_pp_group().rank_in_group self.dp_rank = vllm_config.parallel_config.data_parallel_rank_local self.dp_size = vllm_config.parallel_config.data_parallel_size_local self.pp_size = vllm_config.parallel_config.pipeline_parallel_size self.kv_caches: dict[str, torch.Tensor] = {} self.side_channel_host = get_ip() self.pcp_size = get_pcp_group().world_size # Assert that pp_size and pcp_size cannot both be greater than 1 assert not (self.pp_size > 1 and self.pcp_size > 1), "pp and pcp cannot open in same time" self.pcp_rank = get_pcp_group( ).rank_in_group if self.pcp_size > 1 else 0 self.dcp_size = get_decode_context_model_parallel_world_size() self.dcp_rank = get_decode_context_model_parallel_rank( ) if self.dcp_size > 1 else 0 self.max_device_id = self.tp_size * self.dp_size * self.pcp_size * self.pp_size self.kv_role = vllm_config.kv_transfer_config.kv_role self.num_key_value_heads = self.vllm_config.model_config.hf_text_config.num_key_value_heads # Handshake base port self.side_channel_port = ( vllm_config.kv_transfer_config.kv_port + vllm_config.parallel_config.data_parallel_rank * vllm_config.parallel_config.tensor_parallel_size * vllm_config.parallel_config.pipeline_parallel_size * self.pcp_size) device_index = (self.pp_rank + self.pcp_rank) * self.tp_size + self.tp_rank self.handshake_port = self.side_channel_port + device_index self.sockets: dict = {} self.engine = global_te.get_transfer_engine(self.side_channel_host, device_name=None) self.te_rpc_port = self.engine.get_rpc_port() # Background thread for sending or receiving KV caches. self.kv_send_thread: Optional[KVCacheSendingThread] = None self.kv_recv_thread: Optional[KVCacheRecvingThread] = None # Handshake metadata of this worker self.xfer_handshake_metadata: MooncakeAgentMetadata | None = None # kv_transfer variables self.vllm_config = vllm_config self.block_size = vllm_config.cache_config.block_size if self.vllm_config.model_config.is_deepseek_mla: self.tp_num_need_pulls = 1 else: num_d_block_heads = max(1, self.num_key_value_heads // self.tp_size) num_p_block_heads = max( 1, self.num_key_value_heads // self._prefill_tp_size) self.tp_num_need_pulls = num_d_block_heads // num_p_block_heads self.local_remote_block_port_mapping: dict[ str, Optional[List[List[int]]]] = {} self.remote_port_send_num: dict[str, dict[int, int]] = {} def _get_prefill_decode_size(self, vllm_config: VllmConfig): # get prefill tp and dp size from extra config prefill_parallel_config: dict[ str, Any] = vllm_config.kv_transfer_config.get_from_extra_config( "prefill", {}) assert "tp_size" in prefill_parallel_config.keys() self._prefill_tp_size = prefill_parallel_config["tp_size"] assert "dp_size" in prefill_parallel_config.keys() self._prefill_dp_size = prefill_parallel_config["dp_size"] # get prefill pp size from extra config self._prefill_pp_size = prefill_parallel_config.get("pp_size", 1) # get decode tp and dp size from extra config decode_parallel_config: dict[ str, Any] = vllm_config.kv_transfer_config.get_from_extra_config( "decode", {}) assert "tp_size" in decode_parallel_config.keys() self._decode_tp_size = decode_parallel_config["tp_size"] assert "dp_size" in decode_parallel_config.keys() self._decode_dp_size = decode_parallel_config["dp_size"] # get prefill pp size from extra config self._decode_pp_size = decode_parallel_config.get("pp_size", 1) assert self._decode_pp_size == 1, "decode pp size must be 1" self._prefill_pp_layer_partition = prefill_parallel_config.get( "pp_layer_partition", None) def register_kv_caches(self, kv_caches: dict[str, torch.Tensor]): """Register the KV Cache data.""" _, first_kv_cache_tuple = next(iter(kv_caches.items())) first_kv_cache = first_kv_cache_tuple[0] # TODO(tms): Find a more robust way to detect and handle MLA self.use_mla = first_kv_cache_tuple[0].size( -1) != first_kv_cache_tuple[1].size(-1) and len( first_kv_cache_tuple) == 2 self.use_sparse = len(first_kv_cache_tuple) == 3 if self.use_mla: # MLA case.[num_block, block_size, 1, hidden_dim] self.num_blocks = first_kv_cache.shape[0] block_rank = 3 # [block_size, latent_dim] block_shape_norm = first_kv_cache_tuple[0].shape[-block_rank:] block_shape_pe = first_kv_cache_tuple[1].shape[-block_rank:] self.block_len = [ first_kv_cache[0].element_size() * math.prod(block_shape_norm), first_kv_cache[1].element_size() * math.prod(block_shape_pe) ] logger.info( "num_blocks: %s, block_shape_norm: %s, block_shape_pe: %s", self.num_blocks, block_shape_norm, block_shape_pe) elif self.use_sparse: self.num_blocks = first_kv_cache.shape[0] block_rank = 3 # [block_size, latent_dim] block_shape_norm = first_kv_cache_tuple[0].shape[-block_rank:] block_shape_pe = first_kv_cache_tuple[1].shape[-block_rank:] block_shape_k = first_kv_cache_tuple[2].shape[-block_rank:] self.block_len = [ first_kv_cache[0].element_size() * math.prod(block_shape_norm), first_kv_cache[1].element_size() * math.prod(block_shape_pe), first_kv_cache[2].element_size() * math.prod(block_shape_k) ] logger.info( "num_blocks: %s, block_shape_norm: %s, block_shape_pe: %s, block_shape_k: %s", self.num_blocks, block_shape_norm, block_shape_pe, block_shape_k) else: # eager:[num_block, block_size, num_head, hidden_dim] # torchair:[num_block, block_size, num_head*hidden_dim] self.num_blocks = first_kv_cache.shape[0] kv_elem_size = first_kv_cache.element_size() block_rank = len( first_kv_cache.shape ) - 1 # [block_size, kv_heads, head_dim] or [block_size, kv_heads*head_dim] block_shape = first_kv_cache.shape[-block_rank:] self.block_len = [kv_elem_size * math.prod(block_shape)] logger.info("num_blocks: %s, block_shape: %s", self.num_blocks, block_shape) logger.info( "Registering KV_Caches. use_mla: %s, use_sparse: %s, shape %s", self.use_mla, self.use_sparse, first_kv_cache.shape) self.kv_caches = kv_caches kv_caches_base_addr = [] ptrs = [] lengths = [] for cache_or_caches in kv_caches.values(): # Normalize to always be a list of caches if self.use_mla: for i, cache in enumerate(cache_or_caches, 0): base_addr = cache.data_ptr() region_len = self.num_blocks * self.block_len[i % 2] kv_caches_base_addr.append(base_addr) ptrs.append(base_addr) lengths.append(region_len) elif self.use_sparse: for i, cache in enumerate(cache_or_caches, 0): base_addr = cache.data_ptr() region_len = self.num_blocks * self.block_len[i % 3] kv_caches_base_addr.append(base_addr) ptrs.append(base_addr) lengths.append(region_len) else: cache_list = [ cache_or_caches ] if self.use_mla or self.use_sparse else cache_or_caches for cache in cache_list: base_addr = cache.data_ptr() region_len = self.num_blocks * self.block_len[0] kv_caches_base_addr.append(base_addr) ptrs.append(base_addr) lengths.append(region_len) global_te.register_buffer(ptrs, lengths) # After KV Caches registered, start the sending or receiving thread. metadata = MooncakeAgentMetadata( engine_id=self.engine_id, te_rpc_port=self.te_rpc_port, kv_caches_base_addr=kv_caches_base_addr, num_blocks=self.num_blocks, local_ip=get_ip(), ) self.xfer_handshake_metadata = metadata ready_event = threading.Event() if self.kv_role == 'kv_producer': self.kv_send_thread = KVCacheSendingThread( self.vllm_config, self.tp_rank, self._prefill_tp_size, self.engine_id, self.side_channel_host, self.side_channel_port, metadata, ready_event, self.kv_caches, self.pcp_rank) self.kv_send_thread.start() else: self.kv_recv_thread = KVCacheRecvingThread( self.tp_rank, self.tp_size, self._prefill_pp_size, self.engine, self.engine_id, self.handshake_port, self.side_channel_port, kv_caches_base_addr, self.block_len, ready_event, self.vllm_config, self.kv_caches, self._prefill_pp_layer_partition) self.kv_recv_thread.start() start_wait_time = time.time() thread = self.kv_send_thread if self.kv_role == 'kv_producer' else self.kv_recv_thread assert thread is not None while not ready_event.is_set(): if not thread.is_alive(): raise RuntimeError( "KV Cache sending/receiving thread failed to start.") if time.time() - start_wait_time > 5 * 60: raise RuntimeError( "Timeout waiting for KV Cache thread to be ready.") time.sleep(3) def get_finished(self) -> tuple[set[str], set[str]]: done_sending = ( self.kv_send_thread. get_and_clear_finished_requests( # type: ignore[union-attr] ) if self.kv_role == 'kv_producer' else set()) done_recving = ( self.kv_recv_thread. get_and_clear_finished_requests( # type: ignore[union-attr] ) if self.kv_role == 'kv_consumer' else set()) if self.tp_rank == 0: logger.debug( "Number of completed KV cache send requests: %d, receive " "requests: %d", len(done_sending), len(done_recving)) return done_sending, done_recving def _get_kv_split_metadata( self, req_id: str, meta: ReqMeta, ) -> tuple[list[list[int]], list[list[int]], list[list[int]]]: """ In cp/dcp scenario, kv_cache may be split, so we need to pull multiple blocks from multiple remote P node. Use this function to calculate remote port and remote block number of each remote P node that we need to pull. """ if meta.remote_pcp_size * meta.remote_dcp_size * self.pcp_size * self.dcp_size == 1: choosen_rank_list = self._get_remote_rank(req_id) remote_handshake_port_list = [[ x + meta.remote_port for x in choosen_rank_list ]] local_block_ids_list, remote_block_ids_list = [ meta.local_block_ids ], [meta.remote_block_ids] return remote_handshake_port_list, local_block_ids_list, remote_block_ids_list def context_parallel_parameters_check(): assert (meta.remote_pcp_size * meta.remote_dcp_size) % ( self.pcp_size * self.dcp_size) == 0 if not self.use_mla: p_node_heads_per_rank = math.ceil(self.num_key_value_heads / self._prefill_tp_size) d_node_heads_per_rank = math.ceil(self.num_key_value_heads / self.tp_size) assert d_node_heads_per_rank % p_node_heads_per_rank == 0 def get_kv_head_groups(tp_size): if self.use_mla: kv_head_groups = [] kv_head_ids = [0] kv_head_groups.append(tuple(kv_head_ids)) return kv_head_groups if self.num_key_value_heads // tp_size >= 1: kv_head_groups = [] for tp_rank in range(tp_size): kv_head_ids = [head_idx + tp_rank * (self.num_key_value_heads // tp_size) \ for head_idx in range(self.num_key_value_heads // tp_size)] kv_head_groups.append(tuple(kv_head_ids)) return kv_head_groups if tp_size // self.num_key_value_heads > 1: kv_head_groups = [] for kv_head_ids_ in range(self.num_key_value_heads): kv_head_groups.append(tuple([kv_head_ids_])) return kv_head_groups def get_cp_group_meta(tp_size, pcp_size, dcp_size, port_base): # key is kv_head_group, value is cp_groups and which cp_groups to select cp_group_meta: dict = {} kv_head_groups = get_kv_head_groups(tp_size) dcp_repeat_num = tp_size // len(kv_head_groups) // dcp_size for kv_head_group_idx, kv_head_group in enumerate(kv_head_groups): if kv_head_group not in cp_group_meta: cp_group_meta[kv_head_group] = {} cp_group_meta[kv_head_group]['cp_groups'] = [] cp_group_meta[kv_head_group]['select_cp_groups_id'] = 0 kv_head_group_offset = tp_size // len( kv_head_groups) * kv_head_group_idx for dcp_repeat_idx in range(dcp_repeat_num): # len(cp_group) == pcp_size * dcp_size cp_group = [] dcp_repeat_offset = dcp_size * dcp_repeat_idx for pcp_rank in range(pcp_size): pcp_rank_offset = tp_size * pcp_rank for dcp_rank in range(dcp_size): cp_group.append(dcp_rank + port_base + pcp_rank_offset + dcp_repeat_offset + kv_head_group_offset) cp_group_meta[kv_head_group]['cp_groups'].append(cp_group) return cp_group_meta def get_local_remote_block_port_mappings(): context_parallel_parameters_check() p_node_cp_group_meta = get_cp_group_meta(self._prefill_tp_size, meta.remote_pcp_size, meta.remote_dcp_size, meta.remote_port) d_node_cp_group_meta = get_cp_group_meta(self.tp_size, self.pcp_size, self.dcp_size, self.side_channel_port) local_remote_block_port_mappings: dict[int, list[list[int]]] = {} for d_node_head_key in d_node_cp_group_meta.keys(): for p_node_head_key in p_node_cp_group_meta.keys(): if not set(p_node_head_key).issubset(set(d_node_head_key)): continue d_node_head_group = d_node_cp_group_meta[d_node_head_key] p_node_head_group = p_node_cp_group_meta[p_node_head_key] for d_cp_group in d_node_head_group['cp_groups']: select_cp_groups_id = p_node_head_group[ 'select_cp_groups_id'] p_cp_groups = p_node_head_group['cp_groups'] p_cp_group = p_cp_groups[select_cp_groups_id] p_node_head_group['select_cp_groups_id'] = select_cp_groups_id + 1 \ if select_cp_groups_id + 1 < len(p_cp_groups) else 0 for d_idx, d_port in enumerate(d_cp_group): if d_port not in local_remote_block_port_mappings: local_remote_block_port_mappings[d_port] = [] p_port_remote_list = [] for p_idx, p_port in enumerate(p_cp_group): if p_idx % len(d_cp_group) == d_idx: p_port_remote_list.append(p_port) local_remote_block_port_mappings[d_port].append( p_port_remote_list) logger.info( "p_node_cp_group_meta is:: %s. d_node_cp_group_meta is:: %s. " "local_remote_block_port_mappings is:: %s. ", p_node_cp_group_meta, d_node_cp_group_meta, local_remote_block_port_mappings) return local_remote_block_port_mappings def get_remote_port_send_num(local_remote_block_port_mappings): remote_port_send_num: dict[int, int] = {} for port in range(self._prefill_tp_size * meta.remote_pcp_size): remote_port_send_num[meta.remote_port + port] = 0 for local_port in local_remote_block_port_mappings.keys(): remote_port_head_list = local_remote_block_port_mappings[ local_port] for remote_port_list in remote_port_head_list: for remote_port in remote_port_list: remote_port_send_num[remote_port] += 1 return remote_port_send_num if meta.remote_engine_id not in self.local_remote_block_port_mapping: self.local_remote_block_port_mapping[meta.remote_engine_id] = None if self.local_remote_block_port_mapping[meta.remote_engine_id] is None: local_remote_block_port_mappings = get_local_remote_block_port_mappings( ) self.local_remote_block_port_mapping[ meta.remote_engine_id] = local_remote_block_port_mappings[ self.handshake_port] self.remote_port_send_num[ meta.remote_engine_id] = get_remote_port_send_num( local_remote_block_port_mappings) local_remote_block_port_mapping = copy.deepcopy( self.local_remote_block_port_mapping[meta.remote_engine_id]) num_external_blocks = math.ceil(meta.num_external_tokens / self.block_size) assert math.ceil(num_external_blocks / (self.pcp_size * self.dcp_size)) == len(meta.local_block_ids), \ f"num_external_blocks({num_external_blocks}), cp_size({self.pcp_size * self.dcp_size}), " \ f"local_block_ids_len ({len(meta.local_block_ids)})" assert meta.num_prompt_blocks >= num_external_blocks, \ f"meta.num_prompt_blocks({meta.num_prompt_blocks}), num_external_blocks({num_external_blocks})" remote_cp_size = meta.remote_pcp_size * meta.remote_dcp_size remote_block_nums_all = [meta.num_prompt_blocks // remote_cp_size ] * remote_cp_size num_remain_blocks = meta.num_prompt_blocks % remote_cp_size for i in range(num_remain_blocks): remote_block_nums_all[i] += 1 last_block_location = (num_remain_blocks + remote_cp_size - 1) % remote_cp_size # Considering prefix cache, the remote_block_nums_all should be revised num_prefix_cached_blocks = meta.num_prompt_blocks - num_external_blocks remote_block_nums_all = [ num - num_prefix_cached_blocks // remote_cp_size for num in remote_block_nums_all ] num_remain_blocks = num_prefix_cached_blocks % remote_cp_size for i in range(num_remain_blocks): remote_block_nums_all[i] -= 1 # make sure the last block (which may be unfull) of P nodes is put to the last block of D node remote_block_nums: list[int] = [] final_block_idx: int | None = None local_cp_rank = self.dcp_rank + self.pcp_rank * self.dcp_size local_cp_size = self.dcp_size * self.pcp_size for cp_rank, block_num in enumerate(remote_block_nums_all): if cp_rank % local_cp_size == local_cp_rank: if last_block_location == cp_rank: final_block_idx = len(remote_block_nums) remote_block_nums.append(block_num) assert local_remote_block_port_mapping is not None if final_block_idx is not None: final_block_num = remote_block_nums.pop(final_block_idx) remote_block_nums.append(final_block_num) for mapping in local_remote_block_port_mapping: final_block_port = mapping.pop(final_block_idx) mapping.append(final_block_port) remote_handshake_port_list, local_block_ids_list, remote_block_ids_list = [], [], [] for idx in range(len(local_remote_block_port_mapping[0])): mapping_list = [] for mapping in local_remote_block_port_mapping: mapping_list.append(mapping[idx]) remote_handshake_port_list.append(mapping_list) # the local_block_ids_list and remote_block_ids_list are related with remote_handshake_port_list # such as: local_block_ids_list[[1],[2],[5],[6]], remote_block_ids_list[[1],[1],[1],[1]], # remote_handshake_port_list[[30000],[30001],[30004],[30005]] # D rank will get remote block 1 in port 30004 and save it in local block 5 local_block_offset = 0 for remote_kv_id in range(len(remote_handshake_port_list)): num_blocks_to_pull = remote_block_nums[remote_kv_id] remote_block_ids_list.append( meta.remote_block_ids[:num_blocks_to_pull]) local_block_ids_list.append( meta.local_block_ids[local_block_offset:local_block_offset + num_blocks_to_pull]) local_block_offset += num_blocks_to_pull assert self.tp_num_need_pulls == len(remote_handshake_port_list[0]), \ f"tp_num_need_pulls: {self.tp_num_need_pulls}, remote_handshake_port_list: {remote_handshake_port_list}" return remote_handshake_port_list, local_block_ids_list, remote_block_ids_list def start_load_kv(self, metadata: MooncakeConnectorMetadata): """Start loading KV blocks from remote engine.""" for req_id, meta in metadata.requests.items(): logger.debug( "start_load_kv for request %s from remote engine %s. " "Num local_block_ids: %s. Num remote_block_ids: %s. ", req_id, meta.remote_engine_id, len(meta.local_block_ids), len(meta.remote_block_ids)) if meta.remote_pcp_size * meta.remote_dcp_size > 1: remote_handshake_port_list, local_block_ids_list, remote_block_ids_list = self._get_kv_split_metadata( req_id, meta) for pcp_dcp_rank in range(len(remote_handshake_port_list)): for i in range(self.tp_num_need_pulls): assert self.kv_recv_thread is not None remote_host, remote_engine_id = self._get_remote_host_info_by_port( meta.remote_port, remote_handshake_port_list[pcp_dcp_rank][i], meta.remote_host, meta.remote_engine_id, meta.remote_multi_nodes_meta_mapping) self.kv_recv_thread.add_request( request_id=req_id, local_block_ids=local_block_ids_list[pcp_dcp_rank], remote_block_ids=remote_block_ids_list[ pcp_dcp_rank], remote_engine_id=remote_engine_id, remote_host=remote_host, remote_handshake_port=remote_handshake_port_list[ pcp_dcp_rank][i], offset=i, tp_num_need_pulls=self.tp_num_need_pulls, remote_port_send_num=self.remote_port_send_num[ meta.remote_engine_id], all_task_done=( pcp_dcp_rank == len(remote_handshake_port_list) - 1 and i == self.tp_num_need_pulls - 1)) else: #TODO: support prefill context parallel and pipeline parallel open at the same time choosen_rank_list = self._get_remote_rank(req_id) remote_handshake_port_list = [[x + meta.remote_port] for x in choosen_rank_list] for i in range(self.tp_num_need_pulls * self._prefill_pp_size): assert self.kv_recv_thread is not None remote_host, remote_engine_id = self._get_remote_host_info_by_port( meta.remote_port, remote_handshake_port_list[i][0], meta.remote_host, meta.remote_engine_id, meta.remote_multi_nodes_meta_mapping) self.kv_recv_thread.add_request( request_id=req_id, local_block_ids=meta.local_block_ids, remote_block_ids=meta.remote_block_ids, remote_engine_id=remote_engine_id, remote_host=remote_host, remote_handshake_port=remote_handshake_port_list[i][0], offset=i, tp_num_need_pulls=self.tp_num_need_pulls, all_task_done=(i == self.tp_num_need_pulls * self._prefill_pp_size - 1)) if self.kv_send_thread is not None and self.pcp_size * self.dcp_size == 1: for req_id, delay_start_time in metadata.requests_to_send.items(): if self.tp_rank in self._prefill_get_remote_rank(req_id): self.kv_send_thread.add_delayed_request( req_id, delay_start_time) else: self.kv_send_thread.add_not_transfer_request(req_id) if self.kv_send_thread is not None and self.pcp_size * self.dcp_size > 1: for req_id, delay_start_time in metadata.requests_to_send.items(): self.kv_send_thread.add_delayed_request( req_id, delay_start_time) def _get_remote_host_info_by_port(self, base_port: int, remote_handshake_port: int, remote_host: str, remote_engine_id: str, remote_multi_nodes_meta_mapping: dict): rank = str(remote_handshake_port - base_port) if remote_multi_nodes_meta_mapping is None or remote_multi_nodes_meta_mapping.get( rank, None) is None: return remote_host, remote_engine_id info = remote_multi_nodes_meta_mapping[rank] return info.get("host", remote_host), info.get("engine_id", remote_engine_id) def _prefill_get_remote_rank(self, req_id: str) -> List[int]: return sum(self._get_remote_ranks_for_req(req_id), []) def _get_remote_rank(self, req_id: str) -> List[int]: return self._get_remote_ranks_for_req(req_id)[self.tp_rank] def _get_remote_tp_ranks(self, tp_ori_data: np.ndarray, rand_group_index: list[int], num_groups: int) -> List[List[int]]: # random split prefill tp list tp_sampled_nums = [] if self._prefill_tp_size > self.num_key_value_heads or self.vllm_config.model_config.is_deepseek_mla or self.use_sparse: tp_ori_data = tp_ori_data.reshape(-1, num_groups) choosen_group = tp_ori_data[:, [rand_group_index]] flattened = choosen_group.reshape(-1).tolist() tp_sampled_nums = [ flattened[i:i + self.tp_num_need_pulls] for i in range(0, len(flattened), self.tp_num_need_pulls) ] # non-random split else: group_size = self._prefill_tp_size // self._decode_tp_size for i in range(self._decode_tp_size): slice = tp_ori_data[i * group_size:(i + 1) * group_size] tp_sampled_nums.append(slice.tolist()) return tp_sampled_nums def _get_remote_ranks_for_req(self, req_id: str) -> List[List[int]]: # Divide the ports according to the TP within the PP sampled_nums = [] if self._prefill_tp_size == self._decode_tp_size: sampled_nums = list( map( lambda tp: [ tp + pp * self._prefill_tp_size for pp in range(self._prefill_pp_size) ], range(self._prefill_tp_size))) return sampled_nums # use deepseek mla, num_key_value_heads == 128, but consider as 1 if self.vllm_config.model_config.is_deepseek_mla or self.use_sparse: num_kv_head = 1 else: num_kv_head = self.num_key_value_heads ori_data = np.arange(self._prefill_tp_size * self._prefill_pp_size) seed = string_to_int64_hash(req_id) rand = random.Random(seed) # random split prefill tp list ori_data = ori_data.reshape(self._prefill_pp_size, -1) num_groups = max( 1, len(ori_data[0]) // num_kv_head ) # The number of redundant copies for each KV head within the PP stage rand_group_index = rand.sample(range(num_groups), \ (max(self._decode_tp_size // num_kv_head, 1))) # random choose a group all_results = [ self._get_remote_tp_ranks(ori_data[pp_index], rand_group_index, num_groups) for pp_index in range(self._prefill_pp_size) ] for group_index in range(len(all_results[0])): group = [] for pp_index in range(self._prefill_pp_size): group.extend(all_results[pp_index][group_index]) sampled_nums.append(group) return sampled_nums @contextlib.contextmanager def zmq_ctx(socket_type: Any, addr: str) -> Iterator[zmq.Socket]: # type: ignore """Context manager for a ZMQ socket""" if socket_type not in (zmq.ROUTER, zmq.REQ, zmq.DEALER): # type: ignore raise ValueError(f"Unexpected socket type: {socket_type}") ctx: Optional[zmq.Context] = None # type: ignore try: ctx = zmq.Context() # type: ignore yield make_zmq_socket(ctx=ctx, path=addr, socket_type=socket_type, bind=socket_type == zmq.ROUTER) # type: ignore finally: if ctx is not None: ctx.destroy(linger=0) def group_concurrent_contiguous( src: List[int], dst: List[int] ) -> Tuple[List[npt.NDArray[np.int64]], List[npt.NDArray[np.int64]]]: """Vectorised NumPy implementation.""" src_indices: npt.NDArray[np.int64] = np.array(src, dtype=np.int64) dst_indices: npt.NDArray[np.int64] = np.array(dst, dtype=np.int64) if src_indices.size == 0: return [], [] brk = np.where((np.diff(src_indices) != 1) | (np.diff(dst_indices) != 1))[0] + 1 src_groups = np.split(src_indices, brk) dst_groups = np.split(dst_indices, brk) src_groups = [g.tolist() for g in src_groups] dst_groups = [g.tolist() for g in dst_groups] return src_groups, dst_groups def string_to_int64_hash(input_str): """ Hash the string using SHA-256 and convert it into an int64 integer. """ hashed_bytes = hashlib.sha256(input_str.encode("utf-8")).digest() trunked_bytes = hashed_bytes[:8] uint64_value = struct.unpack(" 0: logger.warning( f"Send failed: {e}, retrying... ({retries_left} " "attempts left)") time.sleep(0.1) else: logger.error(f"Send failed after all retries: {e}") raise RuntimeError(f"Failed to send data after {max_retries} " f"retries: {e}") def ensure_zmq_recv( socket: zmq.Socket, # type: ignore poller: zmq.Poller, # type: ignore timeout: float = 1.0, max_retries: int = 3) -> bytes: retries_left = max_retries while True: try: if dict(poller.poll(int(timeout * 1000))): # milliseconds data = socket.recv() return data else: raise zmq.ZMQError("Receive timeout") # type: ignore except zmq.ZMQError as e: # type: ignore retries_left -= 1 if retries_left > 0: logger.warning(f"Receive failed: {e}, retrying... " f"({retries_left} attempts left)") time.sleep(0.1) else: logger.error(f"Receive failed after all retries: {e}") raise RuntimeError( f"Failed to receive data after {max_retries} " f"retries: {e}") # decode node should know pp_partition_layer in prefill node, # it is configured in kv_transfer_config by partition_list_str, # default using vllm layer split algorithm. def get_prefill_pp_indices( num_hidden_layers: int, pp_rank: int, pp_size: int, partition_list_str: Optional[str] = None) -> tuple[int, int]: if partition_list_str is None: return get_pp_indices(num_hidden_layers, pp_rank, pp_size) else: try: partitions = [ int(layer) for layer in partition_list_str.split(",") ] except ValueError as err: raise ValueError("Invalid partition string: {}".format( partition_list_str)) from err if len(partitions) != pp_size: raise ValueError(f"{len(partitions)=} does not match {pp_size=}.") if sum(partitions) != num_hidden_layers: raise ValueError( f"{sum(partitions)=} does not match {num_hidden_layers=}.") start_layer = sum(partitions[:pp_rank]) end_layer = start_layer + partitions[pp_rank] return (start_layer, end_layer)