llama.cpp

2b86e5ca - ggml-cpu: fix fallback for RVV kernels without zvfh (#21157)

ggml-cpu: fix fallback for RVV kernels without zvfh (#21157) * ggml-cpu: refactor sgemm; fix rvv checks * ggml-cpu: refactor rvv kernels; set zvfbfwma default to off