llama.cpp b9584 fixes Windows CI, adds KleidiAI on macOS
New release for the popular local LLM runner with broader hardware support.
Deep Dive
llama.cpp released version b9584, fixing Windows CI issues. Builds include macOS Apple Silicon (arm64) with and without KleidiAI, macOS Intel (x64), iOS XCFramework, Linux (CPU, Vulkan, ROCm, OpenVINO, SYCL FP32 disabled), Android arm64 (CPU), Windows (CPU, CUDA 12/13, Vulkan, HIP, SYCL disabled), and several disabled openEuler builds.
Key Points
- Windows CI fix resolves build failures for CPU, CUDA, and Vulkan binaries.
- New macOS Apple Silicon build includes KleidiAI (Neural Engine acceleration).
- Support expanded to Linux s390x and Android arm64, plus HIP for AMD GPUs.
Why It Matters
Ensures reliable local LLM inference on Windows and newer Apple Silicon, broadening access for developers.