llama.cpp
CUDA: skip compilation of superfluous FA kernels
#21768

Merged

CUDA: skip compilation of superfluous FA kernels #21768

IMbackK merged 1 commit into ggml-org:master from JohannesGaessler:cuda-faster-compile

CUDA: skip compilation of superfluous FA kernels

ebbffa0d

JohannesGaessler requested a review 55 days ago

github-actions added Nvidia GPU

github-actions added ggml

am17an approved these changes on 2026-04-11

IMbackK approved these changes on 2026-04-11

IMbackK merged ff5ef827 into master 55 days ago

Reviewers

IMbackK

am17an

Assignees

No one assigned

Labels

Nvidia GPU ggml

Milestone

No milestone