PATOMA (PArtitioning TO Optimize Module Arrangement) Home Page


PATOMA is a software tool based on recursive bipartitioning to perform fixed-outline floorplanning on a combination of fixed and variable-dimension blocks for wirelength minimization. PATOMA uses a look-ahead framework to predict if partitioning can continue, or an end-case has been reached. Two look-ahead predictors are used, ZDS when all the blocks are soft, and ROB when at least one block is hard. If cutsize-driven partitioning is not predicted to be feasible in a subregion, then another partitioning is applied to the blocks of the parent subregion. This partitioning is based on the preconsstructed ZDS or ROB layouts that are guaranteed to be feasible. PATOMA does not use simulated annealing at all in its process, making it a very scalable floorplanner.
People
Project Director
 Professor Jason Cong
Student
  • Michail Romesis
    • Publications
    1. J. Cong, M. Romesis, and J. Shinnerl"Fast Floorplanning by Look-Ahead Enabled Recursive Bipartitioning ", UCLA Computer Science Department Technical Report TR040043, December 2004.
    2. J. Cong, M. Romesis, and J. Shinnerl"Fast Floorplanning by Look-Ahead Enabled Recursive Bipartitioning ", Proc. of the Asia-South Pacific Design Automation Conference, Shanghai, China, January 2005.
    Software
    Benchmarks

    Copyright  2004.  The Regents of the University of California.