b8589
📦 llama-cppView on GitHub →
✨ 1 features🐛 5 fixes🔧 1 symbols
Summary
This release focuses on OpenCL improvements, adding q4_K kernels for Adreno and fixing several related build and runtime issues. It also provides updated pre-built binaries for numerous platforms.
✨ New Features
- Added q4_K gemm and gemv kernels for OpenCL on Adreno devices.
🐛 Bug Fixes
- Fixed kernel build errors in OpenCL.
- Added workarounds for compiler bugs on older OpenCL devices.
- Handled fp16 denorm on X Elite devices for OpenCL.
- Made q4_K cvt kernels signature consistent in OpenCL.
- Fixed whitespace issues in OpenCL files.