llama.cpp
Tensor-parallel: Fix delayed AllReduce on Gemma-4 MoE
#22129

Merged

Tensor-parallel: Fix delayed AllReduce on Gemma-4 MoE #22129

JohannesGaessler merged 3 commits into ggml-org:master from gaugarg-nv:gemma4_perf

Fix delayed AllReduce on Gemma-4 MoE

4ce8fde4

github-actions added ggml

am17an approved these changes on 2026-04-20

JohannesGaessler commented on 2026-04-20

Check for all sources before skipping nodes

07a15854

JohannesGaessler approved these changes on 2026-04-20

Address review comments

63c7607d

JohannesGaessler approved these changes on 2026-04-20

JohannesGaessler merged fd6ae4ca into master 50 days ago

gaugarg-nv deleted the gemma4_perf branch 50 days ago

Reviewers

JohannesGaessler

am17an

Assignees

No one assigned

Labels

ggml

Milestone

No milestone