An integrated approach to accelerate data and predicate computations in hyperblocks