BATAGE-BFNP: A High-Performance Hybrid Branch Predictor with Data-Dependent Branches Speculative Pre-execution for RISC-V Processors

被引：2
作者：
Dang, Nhat Minh [1 ,2 ]
Cao, Hai Xuan [1 ,2 ]
Tran, Linh [1 ,2 ]
机构：
[1] Ho Chi Minh City Univ Technol HCMUT, Dept Elect, Ho Chi Minh City, Vietnam
[2] Vietnam Natl Univ Ho Chi Minh City VNU HCM, Ho Chi Minh City, Vietnam
来源：
ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING | 2023年 / 48卷 / 08期
关键词：
Branch predictor; RISC-V; Speculative pre-execution; Data-dependent branch; TAGE predictor; Neural predictor;
D O I：
10.1007/s13369-022-07593-9
中图分类号：
O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];
学科分类号：
07 ; 0710 ; 09 ;
摘要：
Handling control dependence in a pipeline CPU has always been a big problem in computer architecture design. Since the result of a branch instruction becomes available in the execution stage, and the program counter always requires the instruction address after each cycle, it is necessary to predict the address with the available information instead of stalling the pipeline. In this work, we propose BATAGE-BFNP-a hybrid branch predictor that combines two of the most high-performance branch predictors in the literature (Bayesian tagged geometric length and bias-free neural predictor). Our design uses a speculative pre-execution mechanism to predict the outcome of some particular data-dependent branches. Our implementation reduces Misprediction-per-kilo-instructions (MPKI) by 55.5% compared to BATAGE predictors and 68.2% compared to bias-free neural predictor in Embench benchmarks.
引用
收藏
页码：10299 / 10312
页数：14
相关论文
共 36 条
  • [1] Low-Power, High-Performance Analog Neural Branch Prediction
    Amant, Renee St.
    Jimenez, Daniel A.
    Burger, Doug
    [J]. 2008 PROCEEDINGS OF THE 41ST ANNUAL IEEE/ACM INTERNATIONAL SYMPOSIUM ON MICROARCHITECTURE: MICRO-41, 2008, : 447 - +
  • [2] [Anonymous], SYMBIYOSYS SBY DOCUM
  • [3] Chen L, 2003, NINTH INTERNATIONAL SYMPOSIUM ON HIGH-PERFORMANCE COMPUTER ARCHITECTURE, PROCEEDINGS, P65, DOI 10.1109/HPCA.2003.1183525
  • [4] Embench, MOD EMB BENCHM SUIT
  • [5] Falcón A, 2004, CONF PROC INT SYMP C, P250
  • [6] Farooq MU, 2013, INT S HIGH PERF COMP, P59, DOI 10.1109/HPCA.2013.6522307
  • [7] Gochman S., 2003, INTEL TECHNOLOGY J, V07, P21
  • [8] Gope D., 2014, CHAMPIONSHIP BRANCH
  • [9] Bias-Free Branch Predictor
    Gope, Dibakar
    Lipasti, Mikko H.
    [J]. 2014 47TH ANNUAL IEEE/ACM INTERNATIONAL SYMPOSIUM ON MICROARCHITECTURE (MICRO), 2014, : 521 - 532
  • [10] Hu J, 2018, PROC CVPR IEEE, P7132, DOI [10.1109/CVPR.2018.00745, 10.1109/TPAMI.2019.2913372]
1234