llama.cpp
opencl: use a better matmul path on two Adreno GPU generations
#27640

Merged

opencl: use a better matmul path on two Adreno GPU generations #27640

max-krasnyansky merged 3 commits into ggml-org:master from qualcomm:hq/opencl-adreno-path-selection-r0822

opencl: default the Adreno xmem F16xF32 GEMM on for X2E

a26f7261

opencl: bypass the tiled f32 GEMM on the Adreno A7X

7b68a59c

github-actions added ggml

github-actions added OpenCL

opencl: enable xmem GEMM for adreno by default

84b78229

lhez marked this pull request as ready for review 8 days ago

lhez requested a review 8 days ago

lhez approved these changes on 2026-08-29

lhez requested a review from

max-krasnyansky 8 days ago

max-krasnyansky approved these changes on 2026-08-29

max-krasnyansky merged c841aeeb into master 7 days ago

Reviewers

max-krasnyansky

lhez

Assignees

No one assigned

Labels

ggml OpenCL

Milestone

No milestone