llama.cpp
CUDA: better error for FA kernel with 0 occupancy
#16643

Merged

CUDA: better error for FA kernel with 0 occupancy #16643

JohannesGaessler merged 1 commit into ggml-org:master from JohannesGaessler:cuda-fa-better-error

CUDA: better error for FA kernel with 0 occupancy

6c4f5f77

github-actions added Nvidia GPU

github-actions added ggml

IMbackK approved these changes on 2025-10-17

am17an approved these changes on 2025-10-20

slaren approved these changes on 2025-10-20

JohannesGaessler merged 51d1a8c9 into master 175 days ago

Reviewers

slaren

am17an

IMbackK

Assignees

No one assigned

Labels

Nvidia GPU ggml

Milestone

No milestone