Once more, with feeling. An A9 and A8 fabricated on the same process technology: A8 runs at a higher clock frequency. Depending on the difference in clock frequency, it may or may not consume more power (it would certainly consume less than a multicore). A9 has a higher IPC which means it can get equivalent performance for less clock speed (and hence less switching power, albeit more leakage power). It has many more wires switching, which negates some of that advantage.
The real advantage of A9 is that it will generally be fabbed on smaller process nodes than A8, which makes up for the power-sucking additional hardware. But, of course, if an A8 is fabbed on that same process node, A9 loses that advantage.
