Vector math on raw floating point that is. If you want to wrap that float in a type to force (physical) dimensional consistency, you are going to have a bad time.