The llama.cpp project has released version b12536, which includes enhancements for Vulkan support. Specifically, this update enables Mixed-Precision Quantization (MMQ) with mixed signedness integer dot products. The changes involve specializing MMQ shaders and tracking integer dot product support through signed, mixed, and unsigned flags, with some checks moved to MMVQ call sites. AI
IMPACT Improves performance for certain AI model operations on hardware supporting Vulkan.
RANK_REASON This is a software release for a specific project, not a frontier model release or significant industry event.
Read on llama.cpp — Releases →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →