Hi Keith,
easiness for changing sample rates can indeed be a plus. Although, depending on the system (for me: Windows 7) you don't even have to stop playback to release/change/re-enslave a USB driver when changing
sample rate. But, some systems and versions need to do otherwise, I suppose.
For the maths, I don't see which maths are there involved in creating then sending a buffer. I won't get into details of driver calls and parameter handling, but believe me, it's my job: there's no math.
And the HD to RAM trick ... : if this is could be true, then we could start thinking that different processors, motherboards or memories (2x512 ? 2x1Gb ? 2x2Gb) can produce a different sound. Even with Amarra.
So... ?. How can a software master EM emissions and DC tensions ? A mystery to me.
Once more, as it's my job (at least as much as the Amarra guys), I can tell you that it's only hype, based on peoples' ignorance.
Whatever happens up to the point the data is received by the USB driver is of no importance as far as the integrity of these data is concerned. Fortunately, otherwise, we would never use computers...