Parallel Query for Amazon Aurora

Sounds like they can do predicate pushdown to the storage nodes now. And...mainly beneficial for medium-to-highly selective scans of mostly uncached tables?