b8355
📦 llama-cppView on GitHub →
✨ 3 features🐛 1 fixes🔧 1 symbols
Summary
This release introduces hardware-specific optimizations for AMD GPUs, including new MMVQ parameter tables for RDNA3 and RDNA4 architectures, while refining the RDNA3 table exclusion.
✨ New Features
- Added RDNA4-specific MMVQ parameter table for bs=1 decode.
- Added RDNA3/RDNA4-specific MMVQ parameter table (nwarps=8, rows=1).
- Added dedicated RDNA3 MMVQ parameter table.
🐛 Bug Fixes
- Excluded RDNA3.5 (gfx1150/1151) from the RDNA3 MMVQ parameter table.