Compound Global And Local Two-Level Adaptive Branch Predictor

armoush's picture
Journal Title, Volume, Page: 
Association for the Advancement of Modelling and Simulation Techniques in Enterprises Journal (AMSE-Modeling A), vol. 77, iss. 5, pp. 49-60
Year of Publication: 
Ashraf Armoush
Department of Computer Science , Jordan University, Amman, Jordan
Current Affiliation: 
Department of Computer Engineering, Faculty of Engineering, An-Najah National University P.O. Box 7 Nablus –Palestine
Sami I. Serhan
Department of Computer Science , Jordan University, Amman, Jordan
Preferred Abstract (Original): 

Branch prediction is considered as the most efficient method to alleviate the branch hazard problem. Many predictors have been proposed in the literature to perform the prediction process. Since this is a critical problem in performance improvement, these predictors need continues improvement. This paper proposes a new branch predictor to achieve improvement in the operation speed of highly pipelined computer systems. The proposed predictor combines the different advantages in a single predictor that uses both global and local information in the prediction process. The simulation results of the new proposed predictor and the old methods show that the new predictor gives higher prediction accuracy than previous methods without increasing the hardware cost.