Previous benchmarks have established the performance of sync*. Additionally, why allocate a list when you don't need to? Since these APIs are private, they can be re-arranged a bit to avoid creating as many lists. Will probably not have a large effect on benchmarks