Visit the CADathlon 03 Web Pages


SIGDA Image

SIGDA's CADAthlon at ICCAD

Sunday, Nov. 10, 2002
7:30 a.m.-6 p.m.
Double Tree Hotel
San Jose, CA



 

Home


The Winners!!

The Results!!

Great Pictures!!

Problems

Platform

Participants

Contest
Guidelines


Participation
Request



Important
Dates


Organizing
Committee



 

SIGDA Sister Forums

 
  • Ph.D. Forum at DAC
     
  • Design Automation Summer School
     

      Sponsored by SIGDA
     
    SIGDA Contact
    Dr. Soha Hassoun

  • To download the tar.gz file for all problems and solutions, click here. You will also need the GTL library and the systemC libraries - Check out platform information.


    Problem 1: Analysis & Circuit Design

    Submitted by Frank Liu, IBM Research Labs in Austin

    Overview: Estimating delays of on-chip interconnect plays a major role in the design and verification of digital circuits. The students should focus on understanding how the CRC pi-segment model approximation (equations 14-16 in the reference paper) can be computed for a given generalized RC tree.
    "Modeling the Driving-Point Characteristic of Resistive Interconnect for Accurate Delay Estimation," Peter R. O'Brien & Thomas L. Savarino, ICCAD, 1989, pp. 512-515.

    Problem 2: Physical Design

    Submitted by Patrick Madden, Binghamton, NY

    Overview: The paper below gives an overview of a popular placement algorithm. Students should become familiar with the optimization objectives and also the techniques outlined in this paper.
    "Algorithms for Large-Scale Placement," Jens Vygen, DAC, 1997.

    Problem 3: Logic & High-Level Synthesis

    Submitted by Geert Janssen, IBM, Watson

    Overview: Although the DAGON paper is rather classical, its principles are still in use today. The CADathlon related problem will focus on structural technology mapping of a normalized (or canonical) circuit to a prescribed library of cells. The solution should be optimal with respect to a certain cost function.
    "DAGON: Technology Binding and Local Optimization in DAG Matching," Kurt Keutzer, Design Automation Conference, 1987, pp. 341-347

    Problem 4: System Design and Analysis

    Submitted by Sandeep Shukla, Virginia Tech

    Overview: Most real-time models operate concurrently. These models can be implemented in SystemC as clocked threads. Simulation efficiency of the SystemC model generally decreases with increase in the number of threads. Thus, as the complexity of the models increase, simulation efficiency might be poor.

    To solve the relevant CADathlon problem, the student should gain some knowledge of the structure of a SystemC program and understand how to use sc_methods in SystemC. A good introductory SystemC example is in Chapter 2 in the SystemC Version 2.0.1 Users guide available on the www.systemc.org web site. Furthermore, the students should focus on *broadly* understanding the concurrency re-assignment strategy outlined in the paper, and on learning how to use SystemC threads.

    "Automated Concurrency Re-assignment in High Level System Models for Efficient System-Level Simulation," Nick Savoiu, Sandeep Shukla, Rajesh Gupta, DATE, 2002
    SystemC Users Guide http://www.systemc.org/projects/sitedocs/document/v201_Users_guide/en/1

    Problem 5: Functional Verification

    Submitted by Geert Janssen, IBM, Watson

    Overview: Sequential circuit verification is a broad area of research. Make sure that you understand the theory behind reachability analysis and the algorithms involved. Also consider how to apply BDDs in this context.
    "A Unified Framework for the Formal Verification of Sequential Circuits," Olivier Coudert, Jean Christophe Madre, ICCAD, 1990, Santa Clara, CA, pp. 126-129

    Problem 6: Timing, Test, and Manufacturing

    Submitted by Soha Hassoun, Tufts University

    Overview: The reference paper deals with clock schedule verification for circuits that contain level-sensitive latches. The related CADathlon problem requires students to understand the SMO clocking model, the relationship between arrival and departure times while considering the phase shift operations, and the underlying circuit graph model. Also, consider how the proposed algorithm will be simplified if no physical loops exist within the circuit graph.
    "Verifying Clock Schedules," Thomas G. Szymanski, Narendra Shenoy, ICCAD 1992, pp. 124-131