b9771
📦 llama-cppView on GitHub →
✨ 1 features🔧 1 symbols
Summary
This release introduces an optimization in the Vulkan backend by making mul_mm ALIGNED a spec constant, leading to smaller binaries and reduced shader variants. Various pre-built binaries for different platforms are provided.
✨ New Features
- Vulkan implementation now makes mul_mm ALIGNED a spec constant to reduce shader variant explosion and binary size.