@CliffsEsport@mastodon.social @knowprose@mastodon.social That is a Ryzen AI system (Strix Halo) so it will load a fat 76GB GGUF model, but run slow AF, maybe 3tps. My quad P100 does 46 tps. It is because my quad P100 has ~2,928 GB/s memory bandwidth (~732 GB/s per GPU) and Strix Halo is ~215 – 256 GB/s total.