Skip to content

ggml-cuda: add dequant-float matvec (mmvdq) for Q4_K/Q5_K/Q6_K on RDN…#61

Merged
jimw567 merged 4 commits into
gfx11from
annier.mmvq-to-mmvdq
Jul 26, 2026
Merged

ggml-cuda: add dequant-float matvec (mmvdq) for Q4_K/Q5_K/Q6_K on RDN…#61
jimw567 merged 4 commits into
gfx11from
annier.mmvq-to-mmvdq