On 2025-04-15 15:35, Christian Kastner wrote: > This should install four builds with varying instruction support, with > glibc automatically picking the best version supported by the current CPU. > > This worked fine when ggml was still imbedded in llama.cpp [1], so I > guess I messed up somewhere when I factored ggml out.
Indeed, I messed up what I thought was a simple refactoring in my last upload. I just uploaded new upstream version with a fix to NEW. I also uploaded a new upstream version of llama.cpp to NEW, as the existing one was incompatible with the new ggml. Best, Christian
