qcmadness

4-issue core (好明顯啦)
決心改善integer performance (1 INT decode => 2 INT decode)
成個core都optimize去做4-issue (4 INT / 4 FP / 4 SSE instruction / cycle)

Ref.: K10 block diagram