b8354
📦 llama-cppView on GitHub →
✨ 1 features🐛 1 fixes🔧 1 symbols
Summary
This release optimizes Vulkan performance on AMD hardware by switching to the graphics queue and disabling the async transfer queue. It also includes a wide array of pre-compiled binaries for different platforms.
✨ New Features
- Vulkan backend now uses the graphics queue on AMD hardware for slightly improved performance.
🐛 Bug Fixes
- Disabled async transfer queue usage on AMD Vulkan implementations.