Not exactly what you’re looking fir, but perhaps you’ll find it useful - llama benchmarked on all M-series chips, and in comments there are comparisons with nvi

Not exactly what you’re looking fir, but perhaps you’ll find it useful - llama benchmarked on all M-series chips, and in comments there are comparisons with nvidia. https://github.com/ggerganov/llama.cpp/discussions/4167 https://github.com/ggerganov/llama.cpp/discussions/4167