Free Fix for Tesla P100 Performance in llama.cpp via turboquant v0.3.0

Core summary

One sentence to understand this update

A simple three-line fix, now available in turboquant v0.3.0, significantly improves Tesla P100 performance in llama.cpp by correctly utilizing its fp16 capabilities.

Impact & opportunity

What this could mean

Developers with Tesla P100 GPUs can now achieve better performance in llama.cpp by applying this free fix, making older hardware more efficient for local LLM inference.