●Change8

b10192

📦 llama-cppView on GitHub →
✨ 9 features⚡ 2 deprecations

Summary

This release provides updated binary builds for various platforms including macOS, Linux, Android, and Windows, with expanded support for GPU acceleration technologies like Vulkan, ROCm, OpenVINO, SYCL, HIP, and CUDA. It also includes a new UI component.

✨ New Features

  • Added support for Vulkan on Linux (Ubuntu x64, arm64).
  • Added support for ROCm 7.2 on Linux (Ubuntu x64).
  • Added support for OpenVINO on Linux (Ubuntu x64) and Windows x64.
  • Added support for SYCL FP32 and FP16 on Linux (Ubuntu x64).
  • Added support for SYCL on Windows x64.
  • Added support for HIP on Windows x64 (Radeon).
  • Added support for OpenCL Adreno on Windows arm64.
  • Added support for CUDA 12.4 and 13.3 on Windows x64.
  • Added UI component.

⚡ Deprecations

  • macOS Apple Silicon (arm64, KleidiAI enabled) has been disabled.
  • openEuler builds have been disabled.