llama.cpp
hexagon: support for HMX flash-attention head_dim not multiple of 64
#26539

Merged

hexagon: support for HMX flash-attention head_dim not multiple of 64 #26539

max-krasnyansky merged 1 commit into ggml-org:master from qualcomm:hexagon-hmx-fa-head-dim-pad

github-actions added testing

github-actions added ggml

github-actions added Hexagon

tboinovski1 changed the title ~~hexagon: HMX flash-attention head_dim padding (support DK=DV=72)~~ hexagon: support for HMX flash-attention head_dim not multiple of 64 104 days ago

tboinovski1 marked this pull request as ready for review 103 days ago

tboinovski1 requested a review from

ggerganov 103 days ago

tboinovski1 requested a review 103 days ago

tboinovski1 force pushed 96 days ago

tboinovski1 marked this pull request as draft 74 days ago

tboinovski1 force pushed 61 days ago

hexagon: HMX flash-attention head_dim padding (support DK=DV=72)

a78849c9

tboinovski1 force pushed to a78849c9 58 days ago

max-krasnyansky marked this pull request as ready for review 58 days ago

max-krasnyansky approved these changes on 2026-09-18

lhez approved these changes on 2026-09-18

max-krasnyansky merged 18a04f09 into master 58 days ago

Reviewers

lhez

max-krasnyansky

ggerganov

Assignees

No one assigned

Labels

testing ggml Hexagon

Milestone

No milestone

llama.cpp hexagon: support for HMX flash-attention head_dim not multiple of 64 #26539 Merged

hexagon: support for HMX flash-attention head_dim not multiple of 64 #26539

llama.cpp
hexagon: support for HMX flash-attention head_dim not multiple of 64
#26539

Merged