apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 08:37:06 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[FDConfig]Remove splitwise_role and engine_worker_queue_port in FDConfig (#4147)
Some checks failed
CE Compile Job / ce_job_pre_check (push) Has been cancelled
Details
CE Compile Job / print_ce_job_pre_check_outputs (push) Has been cancelled
Details
CE Compile Job / FD-Clone-Linux (push) Has been cancelled
Details
CE Compile Job / Show Code Archive Output (push) Has been cancelled
Details
CE Compile Job / BUILD_SM8090 (push) Has been cancelled
Details
CE Compile Job / BUILD_SM8689 (push) Has been cancelled
Details
CE Compile Job / CE_UPLOAD (push) Has been cancelled
Details
Deploy GitHub Pages / deploy (push) Has been cancelled
Details

Browse Source 
* remove splitwise_role and engine_worker_queue_port

* fix xpu

* fix xpu

* fix xpu

* fix unittest

* resolve conflct
This commit is contained in:
YuanRisheng
2025-09-19 17:01:52 +08:00
committed by GitHub
parent ee9d8a840a
commit 24180fba0a
23 changed files with 129 additions and 89 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
fastdeploy/engine/expert_service.py
View File

@@ -50,13 +50,13 @@ class ExpertService:
self.cfg = cfg
start_pos = (local_data_parallel_id * self.cfg.parallel_config.tensor_parallel_size) % cfg.worker_num_per_node
end_pos = start_pos + self.cfg.parallel_config.tensor_parallel_size
if cfg.splitwise_role != "mixed":
if cfg.scheduler_config.splitwise_role != "mixed":
self.cfg.cache_config.rdma_comm_ports = self.cfg.cache_config.rdma_comm_ports[start_pos:end_pos]
self.cfg.local_device_ids = self.cfg.device_ids.split(",")[start_pos:end_pos]
llm_logger.info(f"local_data_parallel_id: {local_data_parallel_id}")
self.cfg.disaggregate_info = None
if cfg.splitwise_role != "mixed":
if cfg.scheduler_config.splitwise_role != "mixed":
if len(self.cfg.cache_config.pd_comm_port) == 1:
self.cfg.cache_config.pd_comm_port[0] = (
int(self.cfg.cache_config.pd_comm_port[0]) + local_data_parallel_id
@@ -84,21 +84,21 @@ class ExpertService:
self.api_server_pid = ipc_signal_suffix
self.engine.start_zmq_service(ipc_signal_suffix)
else:
ipc_signal_suffix = self.cfg.engine_worker_queue_port[0]
ipc_signal_suffix = self.cfg.parallel_config.engine_worker_queue_port[0]
llm_logger.info(f"start expert service {local_data_parallel_id}")
if self.cfg.splitwise_role != "mixed":
if self.cfg.scheduler_config.splitwise_role != "mixed":
self.engine.start_cache_service(self.cfg.local_device_ids, ipc_signal_suffix)
self.engine.split_mode_get_tasks()
if self.cfg.scheduler_config.name == "splitwise":
self.cfg.init_cache_info()
role = self.cfg.splitwise_role
role = self.cfg.scheduler_config.splitwise_role
host_ip = self.cfg.host_ip
disaggregate = self.cfg.disaggregate_info
self.engine.scheduler.start(role, host_ip, disaggregate)
if self.cfg.splitwise_role != "mixed":
if self.cfg.scheduler_config.splitwise_role != "mixed":
self.splitwise_receive_thread = threading.Thread(
target=self.engine.split_connector.start_receiver, args=()
)