apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-12-24 13:28:13 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

supports internode_ll_two_stage (#4162)

Browse Source 
* supports internode_ll_two_stage

* supports internode_ll_two_stage

* supports internode_ll_two_stage

* supports internode_ll_two_stage

* supports D internode_ll_two_stage

* fix codestype

* fix xpu internode_ll_two_stage

* fix xpu internode_ll_two_stage
This commit is contained in:
lzy
2025-11-04 16:35:40 +08:00
committed by GitHub
parent 8a40374bfe
commit af7e0f27f3
6 changed files with 165 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
fastdeploy/engine/args_utils.py
View File

@@ -237,6 +237,11 @@ class EngineArgs:
Flag to enable the custom all-reduce kernel.
"""
use_internode_ll_two_stage: bool = False
"""
Flag to use the internode_ll_two_stage kernel.
"""
engine_worker_queue_port: str = "0"
"""
Port for worker queue communication.
@@ -721,6 +726,12 @@ class EngineArgs:
default=EngineArgs.disable_custom_all_reduce,
help="Flag to disable custom all-reduce.",
)
parallel_group.add_argument(
"--use-internode-ll-two-stage",
action="store_true",
default=EngineArgs.use_internode_ll_two_stage,
help="Flag to use the internode_ll_two_stage kernel.",
)
parallel_group.add_argument(
"--max-num-seqs",
type=int,