• DocumentCode
    1834835
  • Title

    PRAM and NAND flash hybrid architecture based on hot data detection

  • Author

    Park, Lung Sik ; Chung, Ki-Seok ; Kim, Hi-Seok ; Han, Tae Hee

  • Author_Institution
    Embedded Software Eng., Sungkyunkwan Univ., Suwon, South Korea
  • Volume
    1
  • fYear
    2010
  • fDate
    1-3 Aug. 2010
  • Abstract
    NAND flash memory is widely used in embedded systems, as it has various advantages, including low power consumption, a small footprint, and non-volatility. However, it has some limitations, for example, the units of its read/write operation and erase operation do not match, which lowers its performance. Furthermore, it cannot perform the in-place overwrite operation and has a limited number of erase operations. Thus, additional software techniques are required to use NAND flash memory, and numerous studies were recently conducted on hybrid architecture using NAND flash and the next-generation non-volatile memory. In this paper, it is suggested that a hot data detection layer be added to a software layer based on the NAND flash hybrid architecture with PRAM (Phase-changed RAM). The hot data detection layer is used to detect hot data and then place this data on PRAM. The simulation results showed that the read performance of the suggested method improved by 12.74%, and its write performance by 17.78% compared to existing methods. The number of erase operations was also reduced by 8.18%.
  • Keywords
    flash memories; random-access storage; software architecture; NAND flash memory; PRAM; embedded systems; hot data detection layer; phase-changed random access memory; software layer; software techniques; Nonvolatile memory; Phase change random access memory; Hot data detection; Hybrid architecture; NAND flash; PRAM; SW architecture;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Mechanical and Electronics Engineering (ICMEE), 2010 2nd International Conference on
  • Conference_Location
    Kyoto
  • Print_ISBN
    978-1-4244-7479-0
  • Electronic_ISBN
    978-1-4244-7481-3
  • Type

    conf

  • DOI
    10.1109/ICMEE.2010.5558587
  • Filename
    5558587