llama.cpp
vulkan: Support FA with any multiple of 8 head sizes
#15537

Merged

vulkan: Support FA with any multiple of 8 head sizes #15537

0cc4m merged 1 commit into ggml-org:master from jeffbolznv:flash_attention_multiple_of_8

vulkan: Support FA with any multiple of 8 head sizes

9a23fc21

jeffbolznv requested a review from

0cc4m 129 days ago

github-actions added testing

github-actions added Vulkan

github-actions added ggml

0cc4m approved these changes on 2025-08-24

0cc4m merged c9a24fb9 into master 129 days ago

Reviewers

0cc4m

Assignees

No one assigned

Labels

testing Vulkan ggml

Milestone

No milestone