PR #16546 CUDA: enable FA for FP32 KV cache

CUDA: enable FA for FP32 KV cache #16546

JohannesGaessler merged 1 commit into ggml-org:master from JohannesGaessler:cuda-fa-f32

CUDA: enable FA for FP32 KV cache

31f2d456

github-actions added Nvidia GPU

github-actions added ggml

ggerganov approved these changes on 2025-10-14

JohannesGaessler merged 9c7185dd into master 21 days ago

Reviewers

ggerganov

Assignees

No one assigned

Labels

Nvidia GPU ggml

Milestone

No milestone