b8478
📦 llama-cppView on GitHub →
✨ 2 features
Summary
This release introduces OpenCL support for flattened Q4_K matrix operations (mv and mm) and provides updated pre-compiled binaries across multiple operating systems and hardware accelerators.
✨ New Features
- Added flattened Q4_K matrix-vector multiplication (mv) support for OpenCL.
- Added flattened Q4_K matrix-matrix multiplication (mm) support for OpenCL.