mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-11-11 14:22:09 +01:00
* vulkan: fix matrix multiplication index calculation for odd m/n and odd k in combination with batching * add odd m/n + odd k test with batching