I'm sure this has been addressed already, but it seems to me that higher clock speeds could be ascertained through the use of modulation of the cores. By setting three cores with a full on/off modulation and alternating a single core to push data through the bus cache allowing data to stream unabated through the cores. In theory temperatures should remained in check through the use of proper modulation in much the same way high powered diodes are made to keep from burning. I'm sure through more optimized prefetching and possibly a background running defrag script, data transfer could made even more efficient.