Cycle Computing is supporting Purdue University in harnessing idle compute power to create a dynamic Condor pool at the upcoming SuperComputing 2009 conference, Nov. 14-20, in Portland, Ore. The on-site project will be a demonstration of high-performance computing (HPC) and open source solutions in the cloud.

As demand for computational power for compute-intensive, mission-critical tasks rapidly outpaces IT budgets, companies are looking for innovative solutions to data center space, power and resource challenges. Cycle and Purdue will demonstrate how harnessing the power of idle, unused compute cycles can help companies cost-effectively perform computationally intensive tasks while optimizing existing IT investments and saving energy.

The Condor pool created at the event will be a smaller-scale version of Purdue's DiaGrid, which utilizes Cycle Computing's CycleServer management tools to aggregate the idle compute power of 28,000 processors across the university and on other campuses in Indiana, Kentucky and Wisconsin. Generating 177 teraflops of compute power, DiaGrid contributes more than two million hours to research projects every month -- the equivalent of a $3 million supercomputer. The research clusters within the DiaGrid pool also average a 1-2 percent idle time, providing one of the highest compute-utilization levels ever. The project benefits faculty, researchers and students who are able to perform demanding science, bioinformatics and social science research and computationally intensive tasks like rendering high-resolution pictures and video. The benefits of DiaGrid are available internationally as well through the NSF TeraGrid and Open Science Grid. Additionally, Campus Technology Magazine selected DiaGrid for a 2009 International Campus Technology Innovators Award.

To learn more about DiaGrid, visit www.itap.purdue.edu/diagrid. An animated video about DiaGrid is also available at www.youtube.com/watch?v=CH_YHGYQl2g.

"With DiaGrid, we were able to live by the motto of 'No Cycle Left Behind' by building a grid of idle campus resources to provide computational capacity to researchers throughout the world," said John Campbell, associate vice president of information technology, Purdue. "This recreation at SuperComputing 2009 will demonstrate how other organizations can easily share idle cycles, avoid purchasing additional computational capacity, maximize existing data center space and squeeze every bit of performance out of each hardware dollar spent."

In its 21st year, SuperComputing 2009 is the premier annual event for the technical community, bringing together a diverse group of researchers, scientists, IT and data center managers, application developers, computer manufacturers, programmers, journalist and analysts. At the event, Cycle is deploying the equivalent of a Linux Server image on Windows/Mac machines, enabling server HPC workloads to run on idle infrastructure. Thanks to virtualization, Condor and CycleServer, unused capacity can be harvested for useful calculations.