LocalAI supports NVIDIA, AMD, and now Intel, any idea of performance comparisons?

[etlweather]

I am wondering if anyone has any kind of inference speed comparison between the various NVIDIA, AMD and Intel cards that would help modeling the price-to-speed to help make a decision on hardware to get for small projects that don't require a whole data center.