xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-16 04:25:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

Scheduled removal of ParallelConfig's direct child EPLB fields (#29324)

Browse Source 
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor 2025-11-25 01:58:58 +00:00 committed by GitHub
parent a178a0b40b
commit a4ad43ad5a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 0 additions and 74 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

50
vllm/config/parallel.py
View File

@ -141,22 +141,6 @@ class ParallelConfig:
- "deepep_high_throughput": Use deepep high-throughput kernels - "deepep_high_throughput": Use deepep high-throughput kernels
- "deepep_low_latency": Use deepep low-latency kernels - "deepep_low_latency": Use deepep low-latency kernels
- "flashinfer_all2allv": Use flashinfer alltoallv kernels for mnnvl""" - "flashinfer_all2allv": Use flashinfer alltoallv kernels for mnnvl"""
num_redundant_experts: int | None = None
"""`num_redundant_experts` is deprecated and has been replaced with
`eplb_config.num_redundant_experts`. This will be removed in v0.12.0.
Please use `eplb_config.num_redundant_experts` instead."""
eplb_window_size: int | None = None
"""`eplb_window_size` is deprecated and has been replaced with
`eplb_config.window_size`. This will be removed in v0.12.0.
Please use `eplb_config.window_size` instead."""
eplb_step_interval: int | None = None
"""`eplb_step_interval` is deprecated and has been replaced with
`eplb_config.step_interval`. This will be removed in v0.12.0.
Please use `eplb_config.step_interval` instead."""
eplb_log_balancedness: bool | None = None
"""`eplb_log_balancedness` is deprecated and has been replaced with
`eplb_config.log_balancedness`. This will be removed in v0.12.0.
Please use `eplb_config.log_balancedness` instead."""
max_parallel_loading_workers: int | None = None max_parallel_loading_workers: int | None = None
"""Maximum number of parallel loading workers when loading model """Maximum number of parallel loading workers when loading model
@ -516,40 +500,6 @@ class ParallelConfig:
"--all2all-backend command-line argument instead." "--all2all-backend command-line argument instead."
) )
# Forward deprecated fields to their new location
if self.num_redundant_experts is not None:
self.eplb_config.num_redundant_experts = self.num_redundant_experts
logger.warning_once(
"num_redundant_experts is deprecated and has been replaced "
"with eplb_config.num_redundant_experts. This will be removed "
"in v0.12.0. Changing this field after initialization will "
"have no effect."
)
if self.eplb_window_size is not None:
self.eplb_config.window_size = self.eplb_window_size
logger.warning_once(
"eplb_window_size is deprecated and has been replaced "
"with eplb_config.window_size. This will be removed "
"in v0.12.0. Changing this field after initialization will "
"have no effect."
)
if self.eplb_step_interval is not None:
self.eplb_config.step_interval = self.eplb_step_interval
logger.warning_once(
"eplb_step_interval is deprecated and has been replaced "
"with eplb_config.step_interval. This will be removed "
"in v0.12.0. Changing this field after initialization will "
"have no effect."
)
if self.eplb_log_balancedness is not None:
self.eplb_config.log_balancedness = self.eplb_log_balancedness
logger.warning_once(
"eplb_log_balancedness is deprecated and has been replaced "
"with eplb_config.log_balancedness. This will be removed "
"in v0.12.0. Changing this field after initialization will "
"have no effect."
)
# Continue with the rest of the initialization # Continue with the rest of the initialization
self.world_size = ( self.world_size = (
self.pipeline_parallel_size self.pipeline_parallel_size

24
vllm/engine/arg_utils.py
View File

@ -855,30 +855,6 @@ class EngineArgs:
"--expert-placement-strategy", "--expert-placement-strategy",
**parallel_kwargs["expert_placement_strategy"], **parallel_kwargs["expert_placement_strategy"],
) )
parallel_group.add_argument(
"--num-redundant-experts",
type=int,
help="[DEPRECATED] --num-redundant-experts will be removed in v0.12.0.",
deprecated=True,
)
parallel_group.add_argument(
"--eplb-window-size",
type=int,
help="[DEPRECATED] --eplb-window-size will be removed in v0.12.0.",
deprecated=True,
)
parallel_group.add_argument(
"--eplb-step-interval",
type=int,
help="[DEPRECATED] --eplb-step-interval will be removed in v0.12.0.",
deprecated=True,
)
parallel_group.add_argument(
"--eplb-log-balancedness",
action=argparse.BooleanOptionalAction,
help="[DEPRECATED] --eplb-log-balancedness will be removed in v0.12.0.",
deprecated=True,
)
parallel_group.add_argument( parallel_group.add_argument(
"--max-parallel-loading-workers", "--max-parallel-loading-workers",