[1/N] Elastic EP Milestone 2 (#34861)

Signed-off-by: Yongji Wu <wuyongji317@gmail.com>
Signed-off-by: Itay Alroy <ialroy@nvidia.com>
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Signed-off-by: Ron Tourgeman <rtourgeman@nvidia.com>
Co-authored-by: Yongji Wu <wuyongji317@gmail.com>
Co-authored-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Co-authored-by: Ron Tourgeman <rtourgeman@nvidia.com>
This commit is contained in:
Itay Alroy
2026-02-28 06:46:42 +02:00
committed by GitHub
parent 90805ff464
commit dea268336f
53 changed files with 3613 additions and 1016 deletions

View File

@@ -419,6 +419,7 @@ class EngineArgs:
enable_expert_parallel: bool = ParallelConfig.enable_expert_parallel
moe_backend: MoEBackend = KernelConfig.moe_backend
all2all_backend: All2AllBackend = ParallelConfig.all2all_backend
enable_elastic_ep: bool = ParallelConfig.enable_elastic_ep
enable_dbo: bool = ParallelConfig.enable_dbo
ubatch_size: int = ParallelConfig.ubatch_size
dbo_decode_token_threshold: int = ParallelConfig.dbo_decode_token_threshold
@@ -896,6 +897,9 @@ class EngineArgs:
"--ubatch-size",
**parallel_kwargs["ubatch_size"],
)
parallel_group.add_argument(
"--enable-elastic-ep", **parallel_kwargs["enable_elastic_ep"]
)
parallel_group.add_argument(
"--dbo-decode-token-threshold",
**parallel_kwargs["dbo_decode_token_threshold"],
@@ -1698,6 +1702,7 @@ class EngineArgs:
is_moe_model=model_config.is_moe,
enable_expert_parallel=self.enable_expert_parallel,
all2all_backend=self.all2all_backend,
enable_elastic_ep=self.enable_elastic_ep,
enable_dbo=self.enable_dbo,
ubatch_size=self.ubatch_size,
dbo_decode_token_threshold=self.dbo_decode_token_threshold,