FRIDAY, NOVEMBER 16

Full Day
8:30am-5:00pm
Room A110/112
  • APART 2001: Automatic Performance Analysis: Resources and Tools

    This full-day workshop, held in conjunction with SC2001, will be a forum for discussing automation of performance analysis for performance-critical distributed and parallel applications. The presentations will focus on performance modeling, measurement and> analysis techniques/tools for parallel, cluster and GRID architectures. Attendance is free for all SC01 participants, who are invited to listen to the presentations and participate in the discussions. The workshop is organized by the European IST Working Group APART (http://www.fz-juelich.de/apart), which includes 10 partners from Europe and the United States.

    Apart Information:
    Michael Gerndt, Technical University of Munich, Germany
    gerndt@in.tum.de
    +49 (89) 289 22386 phone
    +49 (89) 289 28232 fax

    Agenda:

      9-10:30am

    • The IST Working Group APART
      Michael Gerndt, Technical University Munich, Germany

    • Sensibility and Reliability of Parallel Performance Data
      Bernd Mohr, Research Centre Juelich, Germany and Luiz DeRose, IBM
      ACTC, USA

    • Toward a Prototype Implementation of a Performance Bottleneck Analysis System
      Thomas Fahringer, University of Vienna

      Coffee Break

      11:00am-noon

    • Automatic Performance Analysis for SMP Cluster Applications
      Felix Wolf and Bernd Mohr, Research Centre, Juelich, Germany

    • First Steps Towards Using GRM/PROVE for GRID Monitoring
      Peter Kacsuk, Hungarian Academy of Sciences, Hungary

      Lunch Break

      1:30-3:00pm

    • On-line Performance Forecasting Using the Network Weather Service
      Rich Wolski, University of Tennessee, USA

    • Tools and Techniques for Measuring and Improving Grid Performance
      Rupak Biswas, NASA Ames Research Center

      Coffee Break

      3:30-5:00pm

    • Structuring, Locating, and Delivering Grid Performance Information
      Dan Gunter, Lawrence Berkeley National Laboratory

    • Prophesy: An Infrastructure for Analyzing and Modeling the Performance of Parallel and Distributed Applications
      Jonathan Geisler, Northwestern University