llama.cpp
vulkan: fix stale prealloc_y reuse across flash attention and soft_max
#29591

Merged

vulkan: fix stale prealloc_y reuse across flash attention and soft_max #29591

0cc4m merged 1 commit into ggml-org:master from fxgsell:vk-prealloc-y-reuse

fxgsell requested a review from

ggerganov 10 days ago

fxgsell requested a review 10 days ago

github-actions added testing

github-actions added Vulkan

github-actions added ggml

vulkan: fix stale prealloc_y reuse across flash attention and soft_max

5a20ad1a

fxgsell force pushed from 3ddd26a1 to 5a20ad1a 10 days ago

jeffbolznv approved these changes on 2026-09-28

jeffbolznv approved these changes on 2026-10-01

0cc4m approved these changes on 2026-10-05

0cc4m merged 806eee98 into master 4 days ago

Reviewers

0cc4m

jeffbolznv

ggerganov

Assignees

No one assigned

Labels

testing Vulkan ggml

Milestone

No milestone