Sorry, that is a little misleading. That's not the time for the inner loop. That's the time for each cycle of the whole loop, minus the inner loop. The inner loop (the part that does the actual mirroring) is 45 t-states a cycle, 355 in total. But since it's exactly the same for both, I didn't think it mattered much :) > ; 34 bytes / 119 t-states for inner loop