This allows you to serve more clients simultaneously, but not to serve any given client faster i.e. you get higher throughput but not lower latency.