The Maryland CPU-GPU Cluster is a unique computational infrastructure that leverages the synergistic cluster coupling of CPUs, GPUs, displays, and storage.

Project Goals

• Build upon the synergies afforded by coupling CPUs, GPUs (graphics processing units), displays, and storage.
• Driving Force: Evolution of GPUs into powerful, programmable, and inexpensive stream processors.
• Explore interleaved computation and display using multiple networks on the same cluster.
• Explore mapping several applications to this heterogenous and unique computational infrastructure.

Driving Applications

• High-performance Computing - Querying and Visualization of Large Scientific Datasets
• Scientific Computing - Multi-particle Dynamics, Fast Multipole Method, Linear Programming
• Scientific Visualization - Visualization-assisted Computational Steering for Protein Studies
• Virtual 3D Audio - Real-time Soundscape Rendering
• Computer Vision - Modeling and Visualization of Humans and their Activities

Participants

Principal Investigators

A. Varshney, R. Chellappa, R. Duraiswami, J. JaJa, D. O'Leary

Senior Investigators

S. Bhattacharyya, M. Cummings, L. Davis, H. Elman, L. De Floriani, F. Guimbretiere, D.Jacobs, D. Mount, H. Samet, A. Sussman

Systems Staff

F. McCall, B. Erdman

Acknowledgements

NSF CISE RI 04-03313