UW-Madison becomes newest Intel Parallel Computing Center
The University of Wisconsin–Madison has been selected to join the Intel Parallel Computing Centers program.
The program promotes public-private partnerships to modernize technical computing applications in areas important to society, from the development of personalized medical treatments to the delivery of better weather forecasts.
The university joins a small number of institutions worldwide selected to become Intel Parallel Computing Centers.
“Our selection as an Intel Parallel Computing Center will allow us to showcase the ability of the latest Intel parallel accelerators to run a large existing code base … at a higher speed than CPUs can.”
Parallel computing involves the simultaneous, synchronized operation of multiple processors to run elements of a computer program much faster than the sequential operation typical of single processors.
With access to Intel technologies, partners in the program are committed to developing new applications that take advantage of advances in parallel computing technologies. By harnessing the power of parallel processing, the centers accelerate discovery across all scientific fields, including weather forecasting, moving toward more reliable, accurate forecasts over longer time scales.
The Intel Parallel Computing Center at the UW–Madison is directed by Bormin Huang of the Space Science and Engineering Center. He seeks to boost the performance of the Weather Research and Forecasting (WRF) model by adopting the Intel Many Integrated Core Architecture (MIC).
The WRF is a next-generation mesoscale numerical weather prediction system that serves atmospheric research and operational forecasting needs in more than 150 countries.
Huang explains that processing the complex array of algorithms necessary to produce local and regional forecasts requires substantial computing power. The latest Intel MIC accelerator technology shows promise for the needs of weather prediction, he says.
According to Huang, the WRF consists of dynamic and physics modules, with the physics modules processing information about cloud microphysics, cumulus and land-surface parameters, radiation, turbulence, among others. Many or all of the modules must run in order to produce a forecast. Adding to the complexity, the entire model consists of more than 600,000 lines of code making it time-consuming to run.
In the course of several years, Huang has demonstrated and published about the capacity and benefits of using high performance computing technologies to enhance the capabilities of satellite remote sensing and weather forecasting.
Huang notes, “Our selection as an Intel Parallel Computing Center will allow us to showcase the ability of the latest Intel parallel accelerators to run a large existing code base, like the WRF, at a higher speed than CPUs can.”
“We are applying computing technologies to new branches of our discipline, and, in the process, using that technology for the benefit of society,” Huang explains.
There are many areas of study across the university that could also benefit from the faster computational speeds that parallel computing can deliver, adds Huang. He is optimistic about the possibilities for new cross-disciplinary collaborations.
– By Jean Phillips