b9329
📦 llama-cppView on GitHub →
✨ 2 features🔧 2 symbols
Summary
This release introduces a fast Walsh-Hadamard transform for CUDA and updates internal kernel logic, including setting the warp size to 64.
Migration Steps
- For CUDA builds, internal code was updated to change size_t to int in some review contexts.
✨ New Features
- Added fast Walsh-Hadamard transform support for CUDA.
- Added support for warp size 64 in CUDA kernels.