At the moment we're just using minimize latency