WASAPI apparently offers little advantage over kernal streaming in terms of speed. It is not so much designed for users as it is for developers who do not have to write all the code themselves in order to get that low level access. One limitation is that it is designed to handle multiple data streams simultaneously. That is potentially an advantage for users who want to use multiple audio devices, but understandably, the performance of WASAPI (CPU load) takes a serious hit when it is asked to do so. The weakness of ASIO (one driver at a time please) is also a strength (don't dump your crummy data in my stream).