CUDA: more warps for mmvq on NVIDIA#5394

JohannesGaessler merged 1 commit intoggml-org:masterggml-org/llama.cpp:masterfrom

JohannesGaessler:cuda-faster-mmvq-9JohannesGaessler/llama.cpp:cuda-faster-mmvq-9Copy head branch name to clipboard