The llama.cpp project has released version b10057, which includes several bug fixes for the SYCL backend alongside standard binary distributions across multiple platforms.

  • Fixes incorrect row calculation when K_QUANTS_PER_ITERATION is 1 in SYCL.
  • Updates the non-reordered Q5_K kernel to use K_QUANTS_PER_ITERATION.
  • Adds missing second half processing to the reordered q5_k kernel.
  • Resolves a potential off-by-one error and a missing row > nrows check in SYCL code.

This release provides updated binaries for macOS, Linux, Windows, Android, and iOS, supporting various hardware backends including CPU, CUDA, Vulkan, ROCm, OpenVINO, and SYCL.