The MegaRun


In November 2013, Cycle Computing, along with AWS, and Schrödinger, completed a record-setting cluster cloud computing run. The program was designed and performed to not only demonstrate the capability to achieve this type of performance using Cloud cluster computing, but to also advance science. While the data points of this amazing run are very exciting to consider, and appreciate – perhaps the most important is that Real World science was achieved. Below is a brief description of the scientific impact to this run.

Cloud Computing For Organic Photovoltaics Sets A Record High-performance computing accelerates efforts to find low-cost, efficient organic semiconductors for LED displays and solar cells. The advent of faster computers, the ability to tap into unlimited numbers of those computers via the Internet, and more sophisticated data analysis software has led to vastly improved methods for high-throughput screening of molecules for drug discovery and materials science. In a record-breaking example, a team at Cycle Computing, based in Stamford, Conn., in collaboration with organic electronics expert Mark E. Thompson of the University of Southern California, has announced the results of a computing run to search for new organic-based semiconducting materials as replacements for silicon-based materials.
On Nov. 3, the team used Cycle Computing’s CycleCloud system on more than 156,000 interconnected computers on five continents to collectively screen some 205,000 candidate molecules. The run chewed up 2.3 million computing hours in an elapsed time of only 18 hours, tapping into $68 million worth of equipment at a cost of only $33,000, according to Cycle Computing’s CEO, Jason Stowe. The molecules screened included simple organic compounds such as pentacene in which CH units were systematically replaced by nitrogen, benzyl, or other groups, Thompson says. Thompson’s group is now mining the data to find optimal molecules for various semiconducting applications, with an eye toward molecules that will be easy to synthesize and test in prototype photovoltaic devices such as LED displays and solar cells.

- Professor Mark Thompson, University of Southern California

Below are three great interviews where Cycle Computing CEO Jason Stowe discuss MegaRun at length.



SiliconAngle & The Cube Video Interview with Cycle Computing’s Jason Stowe (recorded live)


insideHPC & Rich Report Podcast Interview with Cycle Computing’s Jason Stowe