Overview
Blue Waters is a joint effort of the University of Illinois at Urbana-Champaign, its National Center for Supercomputing Applications, IBM, and the Great Lakes Consortium for Petascale Computation to create the most powerful supercomputer in the world. When it comes online in 2011, the system will sustain one petaflop performance—a thousand trillion floating point operations per second—on a wide range of applications. This extreme high-performance system is expected to open new opportunities for a range of science and engineering research, from understanding how the cosmos evolved after the Big Bang to predicting the behavior of hurricanes and tornadoes.
RENCI is working to enhance performance monitoring and analysis for Blue Waters. This task is particularly challenging given the scale and complexity of the Blue Waters system. RENCI’s approach uses low-overhead agents on each processing element to collect data and perform local pre-analyses, including data compression and reduction, to limit the system-level impact of potential problems. The local data is then subjected to a variety of hierarchical data-reductions and analyses within a communication infrastructure.
Overview
Blue Waters is a joint effort of the University of Illinois at Urbana-Champaign, its National Center for Supercomputing Applications, IBM, and the Great Lakes Consortium for Petascale Computation to create the most powerful supercomputer in the world. When it comes online in 2011, the system will sustain one petaflop performance—a thousand trillion floating point operations per second—on a wide range of applications. This extreme high-performance system is expected to open new opportunities for a range of science and engineering research, from understanding how the cosmos evolved after the Big Bang to predicting the behavior of hurricanes and tornadoes.
RENCI is working to enhance performance monitoring and analysis for Blue Waters. This task is particularly challenging given the scale and complexity of the Blue Waters system. RENCI’s approach uses low-overhead agents on each processing element to collect data and perform local pre-analyses, including data compression and reduction, to limit the system-level impact of potential problems. The local data is then subjected to a variety of hierarchical data-reductions and analyses within a communication infrastructure.
Funding
National Science Foundation
University of Illinois
RENCI Project Team
Rob Fowler, project leader
Partners
University of Illinois
IBM
Great Lakes Consortium for Petascale Computation
More Information
Blue Waters
