Skip to content

Apply back moe_sum_reduce for fused_marlin_moe #14829

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ispobock merged 3 commits into from
Dec 10, 2025
Merged

Apply back moe_sum_reduce for fused_marlin_moe #14829

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
bacec3b
apply
ispobock Dec 10, 2025
181c7e5
Merge branch 'main' into apply-moe-sum-reduce
ispobock Dec 10, 2025
4b131a7
Merge branch 'main' into apply-moe-sum-reduce
ispobock Dec 10, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 10 additions & 4 deletions python/sglang/srt/layers/moe/fused_moe_triton/fused_marlin_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
_is_cuda = is_cuda()

if _is_cuda:
from sgl_kernel import silu_and_mul
from sgl_kernel import moe_sum_reduce, silu_and_mul


def get_scalar_type(num_bits: int, has_zp: bool):
Expand Down Expand Up @@ -204,9 +204,15 @@ def fused_marlin_moe(
).view(-1, topk, K)

output = hidden_states if inplace else torch.empty_like(hidden_states)
torch.sum(intermediate_cache3.view(*intermediate_cache3.shape), dim=1, out=output)
if routed_scaling_factor is not None:
output *= routed_scaling_factor

if routed_scaling_factor is None:
routed_scaling_factor = 1.0

moe_sum_reduce(
intermediate_cache3,
output,
routed_scaling_factor,
)
return output


Expand Down
Loading