• DocumentCode
    2875730
  • Title

    Fast Way-Prediction Instruction Cache for Energy Efficiency and High Performance

  • Author

    Xu, Cuiping ; Zhang, Ge ; Hao, Shouqing

  • Author_Institution
    Inst. of Comput. Technol., Chinese Acad. of Sci., Beijing, China
  • fYear
    2009
  • fDate
    9-11 July 2009
  • Firstpage
    235
  • Lastpage
    238
  • Abstract
    This paper introduces a new way-prediction scheme for achieving low energy consumption and high performance for set-associative instruction cache. The proposed scheme is based on the different cases of how program execution proceeds. By predicting for the sequential instruction flow and non-sequential one respectively, we can achieve a high way prediction hit rate. Since way-prediction doesnpsilat increase the cache access time, we call it fast way-prediction cache (FWPC). Then we implement FWPC for instruction cache on GS232 processor. The experimental results came from FPGA shows that it has very high way prediction hit rate, which is 97.932% on average. The energy dissipation of instruction cache is reduced by 64.83% compared to conventional cache at the cost of 0.2% performance penalty, in which the way prediction miss only cost 0.38% power.
  • Keywords
    cache storage; content-addressable storage; field programmable gate arrays; microprocessor chips; FPGA; GS232 processor; energy consumption; energy dissipation; energy efficiency; fast way-prediction instruction cache; program execution; sequential instruction flow; set-associative instruction cache; way prediction hit rate; Cache memory; Cache storage; Computer architecture; Costs; Energy consumption; Energy dissipation; Energy efficiency; Energy storage; Field programmable gate arrays; System performance; Architecture; Cache; GS232-IP; Godson; Low Power; Way Prediction;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Networking, Architecture, and Storage, 2009. NAS 2009. IEEE International Conference on
  • Conference_Location
    Hunan
  • Print_ISBN
    978-0-7695-3741-2
  • Type

    conf

  • DOI
    10.1109/NAS.2009.54
  • Filename
    5197330