Renaissance Computing Institute

Focus Areas Main

Biosciences & Medicine

Disaster & Environmental Research

Economic Development

Education & Outreach

Faculty Fellows

High Performance Computing

MAESTRO

Humanities, Arts & Social Science

Visualization & Collaborative Environments

Completed Projects

Multi-core Adaptive Execution System and Threaded Runtime Organization

Overview
Multi-core Adaptive Execution System and Threaded Runtime Organization (MAESTRO) is designed to manage application power on high core count microprocessors. In addition, the system will present a simpler programming interface to users by virtualizing the processor cores. Once the base runtime is operational, MAESTRO will be a viable experimental platform for potential research in areas from hybrid programming models, to application resiliency to hardware faults, and integrating debugging/performance tools into production execution.

RENCI will implement a multi-core runtime that uses excess computational power to better utilize hardware resources when shared resources are saturated. The initial implementation will focus on reducing power utilization when memory is saturated, but will be capable of support for other research areas with the prototype.

Funding
U.S. Department of Defense

RENCI Team
Allan Porterfield, project leader
Rob Fowler
Nassib Nassar
Mark Neyer, student

Publications
Allan Porterfield, "MAESTRO: Program Thread and Synchronization Interface (version 0.1)," 2008.