b9015

ggml-org has released llama.cpp b9015, a minor but impactful update to the popular open-source C++ library for running large language models locally. The release removes the dead GGML_VK_MAX_NODES definition from Vulkan code, cleaning up the codebase and reducing potential confusion. This change is part of ongoing efforts to optimize and modernize the Vulkan backend, which is crucial for running LLMs on a wide range of GPUs without proprietary dependencies.

Beyond the Vulkan cleanup, this release significantly expands build support across operating systems and hardware. New prebuilt binaries are available for macOS Intel (x64), Linux s390x, Windows with CUDA 12 and 13, and openEuler (both x86 and aarch64 with ACL Graph). Existing builds for macOS Apple Silicon, Linux (x64, arm64, Vulkan, ROCm, SYCL), Android arm64, and other configurations are also continued. This breadth of platform coverage makes llama.cpp one of the most versatile tools for running AI models locally, whether on consumer desktops, servers, or edge devices.