biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Support mnnvl all2allv from Flashinfer (#21003)

Browse Source 
Signed-off-by: Shu Wang <shuw@nvidia.com>
Signed-off-by: Shu Wang. <shuw@nvidia.com>
Signed-off-by: Tyler Michael Smith <tyler@neuralmagic.com>
Signed-off-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
Co-authored-by: Tyler Michael Smith <tyler@neuralmagic.com>
Co-authored-by: Tyler Michael Smith <tlrmchlsmth@gmail.com>
This commit is contained in:
Shu Wang
2025-09-24 13:38:16 -05:00
committed by GitHub
parent 2dda3e35d0
commit 54e42b72db
10 changed files with 410 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
tests/kernels/moe/modular_kernel_tools/mk_objects.py
View File

@@ -222,7 +222,8 @@ if (has_flashinfer_cutlass_fused_moe()
from vllm.model_executor.layers.fused_moe.flashinfer_cutlass_moe import ( # noqa: E501
FlashInferExperts)
from vllm.model_executor.layers.fused_moe.flashinfer_cutlass_prepare_finalize import ( # noqa: E501
FlashInferCutlassMoEPrepareAndFinalize)
FlashInferCutlassMoEPrepareAndFinalize,
create_flashinfer_prepare_finalize)
register_prepare_and_finalize(
FlashInferCutlassMoEPrepareAndFinalize,
@@ -373,7 +374,7 @@ def make_prepare_finalize(
assert prepare_finalize is not None
return prepare_finalize
elif prepare_finalize_type == FlashInferCutlassMoEPrepareAndFinalize:
return FlashInferCutlassMoEPrepareAndFinalize(
return create_flashinfer_prepare_finalize(
use_dp=moe.moe_parallel_config.dp_size > 1)
else:
return MoEPrepareAndFinalizeNoEP()