• DocumentCode
    2775393
  • Title

    A Timing Driven Congestion Aware Global Router

  • Author

    Erzin, A.I. ; Zalyubovskiy, V.V. ; Shamardin, Y.V. ; Takhonov, I.I. ; Samanta, Radhamanjari ; Raha, Soumyendu

  • Author_Institution
    Sobolev Inst. of Math., Russian Acad. of Sci., Novosibirsk, Russia
  • fYear
    2011
  • fDate
    19-20 Feb. 2011
  • Firstpage
    375
  • Lastpage
    378
  • Abstract
    The multi-net Global Routing Problem (GRP) is a problem of routing a set of nets subject to resources and delay constraints. Many modern routers use FLUTE (Fast Lookup Table Based Rectilinear Steiner Minimal Tree Algorithm for VLSI Design) as the basic method to construct Steiner trees. Here we first give another method MAD (Modified Algorithm Dijkstra) to construct timing-driven Steiner trees for each net. The Elmore delays of the trees constructed by MAD and FLUTE are almost same. Next we describe the Gradient algorithm which is used to reduce the congestion. We implemented our algorithm on ISPD 98 benchmarks and compared the wire length and maximum overflow of our router with Labyrinth Predictable Router. Our router performs better in terms of wire length and is much faster.
  • Keywords
    delays; gradient methods; table lookup; trees (mathematics); Elmore delays; FLUTE; ISPD 98 benchmark; Labyrinth predictable router; MAD method; VLSI design; delay constraint; fast lookup table based rectilinear Steiner minimal tree algorithm; gradient algorithm; modified algorithm Dijkstra; multinet global routing problem; timing driven Steiner tree; timing driven congestion aware global router; Algorithm design and analysis; Benchmark testing; Capacitance; Complexity theory; Delay; Routing; Steiner trees; Congestion(Overflow); Elmore delay; Global Routing; Gradient Method; Steiner tree;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Emerging Applications of Information Technology (EAIT), 2011 Second International Conference on
  • Conference_Location
    Kolkata
  • Print_ISBN
    978-1-4244-9683-9
  • Type

    conf

  • DOI
    10.1109/EAIT.2011.33
  • Filename
    5734962