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

[Moe Refactor] Make Inplace Flag for FusedMoEModularKernel part of the constructor (#33375)

Browse Source 
Signed-off-by: Bill Nell <bnell@redhat.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
This commit is contained in:
bnellnm
2026-02-05 13:07:18 -05:00
committed by GitHub
parent 1ee95841bd
commit a57c8228ff
37 changed files with 132 additions and 109 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
tests/kernels/moe/test_deepep_deepgemm_moe.py
View File

@@ -194,8 +194,11 @@ def make_ll_modular_kernel(
quant_config=quant_config,
moe_config=make_dummy_moe_config(),
)
mk = FusedMoEModularKernel(prepare_finalize=a2a, fused_experts=fused_experts)
return mk
return FusedMoEModularKernel(
prepare_finalize=a2a,
fused_experts=fused_experts,
inplace=False,
)
def make_ht_modular_kernel(
@@ -224,8 +227,11 @@ def make_ht_modular_kernel(
moe_config=make_dummy_moe_config(),
quant_config=quant_config,
)
mk = FusedMoEModularKernel(prepare_finalize=a2a, fused_experts=fused_experts)
return mk
return FusedMoEModularKernel(
prepare_finalize=a2a,
fused_experts=fused_experts,
inplace=False,
)
def make_modular_kernel(
@@ -318,7 +324,6 @@ def deepep_deepgemm_moe_impl(
w2=w2,
topk_weights=test_tensors.topk_weights,
topk_ids=test_tensors.topk,
inplace=False,
activation="silu",
global_num_experts=num_experts,
expert_map=build_expert_map(),