ggml/llama.cpp releases: b10816

TL;DR

The ggml/llama.cpp project released a new version including tuning updates for the M3 model and additional precision settings, addressing formatting issues. These changes are significant for developers working with the M3 model to optimize performance on metal GPUs.

Detailed Summary

The ggml/llama.cpp project has updated its codebase to include fine-tuned vector settings (fa-vec tunings) for the M3 model, specifically adding q4_0, q4_1, q5_0, and q5_1 configurations. This update is part of the metal tuning enhancements and involves modifications in the ggml-metal-tuning and ggml-metal-tuning.cpp files. The broader impact includes improved performance and flexibility for users working with the M3 model on Metal-based hardware.

Key Points

• • addition of m3 in fa_vec_tuned_table
• • adding q4_0,q4_1,q5_0,q5_1 in ggml-metal-tuning
• • Fix formatting in ggml-metal-tuning.cpp