16 * speed depends on source allignment
17 * destination allignment is secondary
18 * byte-at-a-time foreward copy to
19 * get source (A1) alligned.
30 * quad-long-at-a-time forward copy
42 * cleanup byte-at-a-time
58 * everything the same, but
66 * byte-at-a-time backward copy to
67 * get source (A1) alligned.
78 * quad-long-at-a-time backward copy
90 * cleanup byte-at-a-time backward