Skip to content

[Kernels] Isolate modular kernel code from FusedMoEMethodBase subclasses. #27123

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
merged 16 commits into from
Nov 4, 2025
Merged

[Kernels] Isolate modular kernel code from FusedMoEMethodBase subclasses. #27123

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
2199fff
[Kernel] Isolate modular kernel dispatching
bnellnm Oct 17, 2025
7bfd9c3
cleanup
bnellnm Oct 17, 2025
2f3f631
remove uses of self.fused_experts
bnellnm Oct 17, 2025
9d1f2fb
remove uses of self.fused_experts
bnellnm Oct 17, 2025
38799ff
comment
bnellnm Oct 17, 2025
ad9e3b6
cleanups
bnellnm Oct 17, 2025
b12ed23
fix
bnellnm Oct 17, 2025
9b0af52
comment
bnellnm Oct 18, 2025
3048d6e
fix inplace
bnellnm Oct 18, 2025
9d6f340
fix warmup code
bnellnm Oct 18, 2025
f1d0b24
remove unused import
bnellnm Oct 18, 2025
e8be030
clean up object types and initialization
bnellnm Oct 20, 2025
bc0cf46
review comments
bnellnm Oct 24, 2025
d1b98d5
add type annotation
bnellnm Oct 24, 2025
5830b98
rebase
bnellnm Nov 3, 2025
a0621a3
fix merge
bnellnm Nov 3, 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
4 changes: 2 additions & 2 deletions vllm/distributed/device_communicators/base_device_communicator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -266,14 +266,14 @@ def prepare_communication_buffer_for_model(self, model: torch.nn.Module) -> None
module
for module in model.modules()
# TODO(bnell): Should use isinstance but can't. Maybe search for
# presence of quant_method.init_prepare_finalize?
# presence of quant_method.maybe_init_modular_kernel?
if (
module.__class__.__name__ == "FusedMoE"
or module.__class__.__name__ == "SharedFusedMoE"
)
]
for module in moe_modules:
module.quant_method.init_prepare_finalize(module)
module.maybe_init_modular_kernel()

def dispatch(
self,
Expand Down
Loading