llama.cpp

c2224f00 - ggml-vulkan: fix VULKAN_CHECK_RESULTS flag, which was previously broken (#5813)