b8697
📦 llama-cppView on GitHub →
🐛 1 fixes🔧 1 symbols
Summary
This release introduces a safety check in CUDA operations to prevent buffer overlap during fusion and provides numerous updated pre-built binaries across different operating systems and hardware configurations.
Migration Steps
- If using CUDA fusion, be aware that a check for buffer overlap is now performed.
🐛 Bug Fixes
- CUDA: Implemented a check for buffer overlap before fusing operations.