my reading that it will be some hyper-performant db thanks to some very low level optimization utilizing recent hw advancements and formats/pipelines unification and simplification.