[emilfihlman]

>Does this also apply for data that's already compressed

Of course not, but there are ways like multiple connections and threading (preventing blocking or slow io) that might increase throughput considerably, but normally? Nah.