• DocumentCode
    3179250
  • Title

    Fast power delivery network analyzer

  • Author

    Hwang, Bosun ; Koo, Jongeun ; Hwang, Chanseok ; Cheon, Younghoi ; Ahn, Sooyoung ; Lee, Jongbae ; Yoo, Moonhyun

  • fYear
    2011
  • fDate
    14-16 March 2011
  • Firstpage
    1
  • Lastpage
    4
  • Abstract
    With the increase in circuit frequency and supply voltage Scaling, a robust power network design is essential to ensure that the circuits on a chip operate reliably at the guaranteed level of performance. Traditionally the power network analysis has its main focus on IR-drop effects. However, IR drop analysis approaches have strong dependence on the input vectors and may require a tremendously long execution time. In this paper, we propose a novel and fast power network analysis method which calculates the effective resistance between all power pads and power grids. This method explores huge parasitic power networks and detects hot spots with an abnormal effective resistance value resulted from gross errors in the post-layout power network. We currently use the proposed method for our memory and DDI circuits to validate the post-layout power network quickly. We developed our method by using multi-thread and multi-process techniques, resulting in up to 50 times speed improvement.
  • Keywords
    distribution networks; network analysis; power aware computing; power grids; IR drop analysis; circuit frequency; multiprocess technique; multithread technique; parasitic power network; power delivery network analyzer; power grid; power network analysis; power pad; robust power network design; supply voltage scaling; Equations; Mathematical model; Matrix decomposition; Power grids; Resistance; Resistors; Sparse matrices; Power distribution network; effective resistance; linear solver; multi-process; multi-thread;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Quality Electronic Design (ISQED), 2011 12th International Symposium on
  • Conference_Location
    Santa Clara, CA
  • ISSN
    1948-3287
  • Print_ISBN
    978-1-61284-913-3
  • Type

    conf

  • DOI
    10.1109/ISQED.2011.5770799
  • Filename
    5770799