llama.cpp
HIP: Tune RDNA4 MMQ tiles for Q6_K / Q2_K (gfx1201)
#25587

Closed

HIP: Tune RDNA4 MMQ tiles for Q6_K / Q2_K (gfx1201) #25587

qiongz wants to merge 3 commits into ggml-org:master from qiongz:hip-rdna4-mmq-mmvq-mtp-opt

MMQ: RDNA4 gfx1201 Q6_K prefill + small-N decode tuning

ff4c006e

MMQ: per-type geometry dispatch for RDNA4 (fix Q4/Q5/Q8 prefill)

9f07b0d4

MMQ: fix Q3_K prefill and Q2_K decode regressions on RDNA4

fe2df7f3

qiongz requested a review 3 days ago

github-actions added ggml

github-actions added CUDA

JohannesGaessler closed this 3 days ago

Reviewers

No reviews

Assignees

No one assigned

Labels

ggml CUDA

Milestone

No milestone