a couple questions for people who have been using it. Where does this fit between a typical budget arm cortex and a gpu? and what are the practical sized mode

a couple questions for people who have been using it. Where does this fit between a typical budget arm cortex and a gpu? and what are the practical sized models you could run on one of these?