The llama.cpp project has released version b10066, introducing support for the `kernel_gemm_moe_q6_k_f32_ns` kernel via OpenCL. This update enables the loading and usage of this specific binary kernel library within the framework.

  • The release includes binaries for macOS (Apple Silicon and Intel), iOS, Linux (Ubuntu x64, arm64, s390x with CPU, Vulkan, ROCm 7.2, OpenVINO, SYCL FP32/FP16), Windows (CPU, Vulkan, CUDA 12/13, OpenVINO, SYCL, HIP), and Android.
  • KleidiAI support for macOS Apple Silicon is disabled in this build.
  • openEuler support is also disabled for this release.

This update provides users with broader hardware compatibility through the new OpenCL kernel implementation.