b9939
📦 llama-cppView on GitHub →
✨ 1 features🐛 1 fixes🔧 1 symbols
Summary
This release introduces hardware acceleration support for Depthwise Convolution (CONV_2D_DW) on the Metal backend and includes corresponding bug fixes for F16 kernels on CPU.
✨ New Features
- Added support for Depthwise Convolution (CONV_2D_DW) on Metal backend.
🐛 Bug Fixes
- Fixed F16 kernel support for CONV_2D_DW on the CPU backend.