CUDA: route single-column f32 mul_mat through mmvf (transpose-free operand swap)#49

mgehre-amd merged 1 commit into

gfx11ROCm/llama.cpp:gfx11from

rogarcia.cuda-mmvf-single-col-swapROCm/llama.cpp:rogarcia.cuda-mmvf-single-col-swapCopy head branch name to clipboard