b9087
📦 llama-cppView on GitHub →
✨ 2 features🐛 1 fixes🔧 1 symbols
Summary
This release introduces performance improvements to the SYCL backend by reordering MMVQ/dequant paths for Q5_K and Q8_0 quantization, alongside removing duplicate definitions.
✨ New Features
- SYCL backend updated Q5_K quantization to reorder MMVQ/dequant path.
- SYCL backend updated Q8_0 quantization to reorder MMVQ path.
🐛 Bug Fixes
- Removed duplicate definitions in the SYCL backend implementation.