Abstract

I will discuss the problem of developing tools and middleware for large scale parallel environments. We are especially interested in systems, both leadership class parallel computers and clusters that have 100,000's or even millions of processors. The infrastructure that we have developed to address this problem is called MRNet, the Multicast/Reduction Network. MRNet's approach to scale is to structure control and data flow in a tree-based overlay network (TBON) that allows for efficient request distribution and flexible data reductions. I will then present an overview of the MRNet design, architecture, and computational model and then discuss several of the applications of MRNet. The applications include scalable automated performance analysis, a vision clustering application and, most recently, an effort to develop our first petascale debugging tool, STAT, a scalable stack trace analyzer running currently on 100,000's of processors on both the Cray XT and IBM BlueGene. In the last part of my talk, I will present a plea for funding programs that will allow a broad collection of universities to provide the training and experience to learn how to operate at these large scales.

Session 2: Applications

2:30-3:00 Netlag: A Performance Evaluation Tool for Massively Multi-User Networked Applications, Alexander Ploss, Dominik Meilaender, Philipp Mollers, Frank Glinka, and Sergei Gorlatch, University of Muenster, Germany

3:00-3:30 BTWorld: Towards Observing the Global BitTorrent File-Sharing Network, Maciej Wojciechowski, Mihai Capotă, Johan Pouwelse, and Alexandru Iosup, Delft University of Technology, the Netherlands

3:30-4:00 Break

Session 3: Models and Systems

4:00-4:30 A Hybrid Markov Chain Model for Workload on Parallel Computers, Anne Krampe, Joachim Lepping, and Wiebke Sieben, Technical University Dortmund, Germany

4:30-5:00 LogGOPSim -- Simulating Large-Scale Applications in the LogGOPS Model, Torsten Hoefler, University of Illinois at Urbana-Champaign, USA, Timo Schneider, Indiana University, USA, and Andrew Lumsdaine, Indiana University, USA (Best paper)

5:00-5:30 Fast and Scalable Simulation of Volunteer Computing Systems Using SimGrid, Bruno Donassolo, Instituto de Informatica Universidade Federal do Rio Grande do Sul, Brazil, Henri Casanova, Information and Computer Sciences Dept. University of Hawai'i at Manoa, USA, Arnaud Legrand, CNRS, University of Grenoble, France, and Pedro Velho, University of Grenoble, France