What advantages is this style of fast memory register supposed to have over the work request style found in IB 1.2 and iWARP? How is this style of fast memory register supposed to be utilized from DAPL or ITAPI when they both assume work requests?