Having preemption throughout the whole kernel helps with everything Low-Latency whether on desktop, mobile or server systems; it's not limited to the traditional real-time use case, although it might obviously help there as well to some extent.