Research On Branch Prediction History Migration In Multi-Core Architectures For Big Data Applications

Branch prediction is an effective way to improve instruction supply and throughput of processor pipeline.However,branch predictors usually rely on the branch history to achieve high prediction accuracy through warming up,which is time consuming.Besides,studies on improving performance of big data workloads show that significant common-ality can be found in instruction streams across different cores running the same big data workload.Based on the above observation,in this paper,we propose a technique for migrating branch history between cores to reduce the branch prediction training process and improve the system performance.We conduct a variety of simulations to verify the effectiveness of our method.Experiments were conducted using a CBP simulator for branch prediction contests to verify the effectiveness of our method.The specific work and innovation are as follows:1.In multi-core architecture,the commonality of instruction between cores is stud-ied,which provides a theoretical basis for branch prediction history migration.2.Based on the most advanced TAGE branch predictor,the migration of branch prediction information is tested.The aspects that may affect the system performance are analyzed theoretically and discussed further.3.Different migration strategies and simulation experiments are designed to study the optimal migration method and the influence of different parameter settings on migra-tion results.Through theoretical analysis and experimental simulation,it is concluded that the branch prediction accuracy of new nuclei can be effectively improved by predicting the branch history information between migrating nuclei.Results show that compared to not migrating,migrating branch history can improve the prediction accuracy in the new core by 24.26%and 34.08%on average for migrating history in TO and migrating history in T0-T4 respectively.Besides,our method performs well across different cores.