llama.cpp
CUDA: avoid mul + bias fusion when buffers are split
#16935

Merged

CUDA: avoid mul + bias fusion when buffers are split #16935

am17an merged 1 commit into ggml-org:master from am17an:cuda-fix-sm-row

CUDA: avoid mul + bias fusion when doing fusion

1d3a6152

am17an requested a review from

slaren 125 days ago

github-actions added Nvidia GPU

github-actions added ggml

am17an requested a review from

JohannesGaessler 125 days ago

am17an changed the title ~~CUDA: avoid mul + bias fusion when doing fusion~~ CUDA: avoid mul + bias fusion when buffers are split 125 days ago

JohannesGaessler approved these changes on 2025-11-02

IMbackK approved these changes on 2025-11-02

am17an merged 2759ccdb into master 124 days ago

am17an deleted the cuda-fix-sm-row branch 124 days ago

Reviewers

JohannesGaessler

IMbackK

slaren

Assignees

No one assigned

Labels

Nvidia GPU ggml

Milestone

No milestone