If I were Intel I would have done things differantly. I would have left the L2cache at 4mbs. Used the saved transitors to double the L1 cache size. maybe that would have lowered the clock speeds I don't know. But increaseing L1 it seems to me would be a greater performance leap.
