High Performance Computing with BlueGene/L and
QCDOC Architectures

Location:
Building 510 Physics Department Seminar Room
at Brookhaven National Laboratory

Date: October 27 & 28 2004

Sponsors - Argonne National Laboratory, Brookhaven National Laboratory, Columbia University, Edinburgh University, IBM Corporation, RIKEN BNL Research Center

BlueGene/L and QCDOC are new computer architectures that are designed to be highly scalable with low latency communications and unprecedented computing performance. With the arrival of hardware based on these architectures at several institutions we have organized a highly topical workshop to investigate the full potential of these machines and how to utilize them toward the widest range of applications including biology, climate modeling and high energy physics.

The goal of this two-day workshop is to:

• Explore and understand applications and algorithms that have already been ported to these machines.
• Expose scientists and the community to these emerging highly scalable architectures.
• Identify areas where collaboration among the sponsoring institutions can be fruitful.
• Provide information on how to identify applications that fit the BlueGene/L or QCDOC Architectural models and helpful tips on how to port and tune codes.

The workshop is divided into the following sessions:

• Hardware and Software status review of both machines,
• Presentations on currently ported and prospective applications, including Linpack benchmarks.
• Helpful tips in identifying appropriate codes and tuning.

Organizing Committee

• Gyan Bhanot, IBM
• Norman Christ, Columbia/RBRC
• Jim Davenport, BNL/CDIC
• Ed Jedlicka, ANL
• Bob Mawhinney, Columbia/RBRC
• Ed McFadden, BNL
• Mike McGuigan, BNL
• Tom Schlagel, BNL
• Bill Pulleyblank, IBM