Change8

b8610

📦 llama-cppView on GitHub →
🐛 2 fixes🔧 3 symbols

Summary

This release includes bug fixes and refactoring within ggml-cpu, specifically addressing RVV kernel fallbacks and setting zvfbfwma to default off.

Migration Steps

  1. The default setting for zvfbfwma in RVV kernels has been turned off.

🐛 Bug Fixes

  • Fixed fallback for RVV kernels lacking zvfh support in ggml-cpu.
  • Refactored sgemm and fixed RVV checks in ggml-cpu.

Affected Symbols