This is very interesting... I've never thought to divide the buffer into multiple blocks. Are they requiring some low-level OS abstraction to pull this off or a

This is very interesting... I've never thought to divide the buffer into multiple blocks. Are they requiring some low-level OS abstraction to pull this off or are we looking at a novel technique at a higher level (i.e. could I patch an existing implementation)?