llama.cpp

Vulkan: VK_KHR_cooperative_matrix support to speed up prompt processing

#10597

Merged