Exploring branch predictability limits with the MTAGE+SC predictor *

In the previous championship CBP-4, the winner of the unlimited storage track [5], poTAGE-SC was combining several TAGE based predictors using different forms of histories (local, global, and frequency), a COLT inspired [3] prediction combiner and a statistical corrector (SC) predictor [8, 10] fed with various forms of branch histories. With MTAGE-SC, we improve this predictor in two ways. First through incorporating new forms of branch histories, adding a new TAGE component and incorporating other forms of histories in the statistical corrector predictor. Second in conveying more information from the TAGE predictors stage to the statistical corrector and to the final prediction computation stage. On the CBP-4 traces, the proposed MTAGE-SC predic-tor achieves 1.600 mispredictions per thousand instructions (MPKI), while the winner of CBP-4 was achieving 1.691 MPKI. On CBP-5 train traces, the MTAGE-SC pre-dictor achieves 2.575 MPKI, 4.7% lower than the winner of CBP-4 (2.717).