there are a limited number of OSX workers to schedule that shards work on, so you end up waiting longer for them to execute