Science, engineering students developing HPC programming skills through Great Lakes Consortium Virtual School summer course

 

Ohio Supercomputer Centers hosts 40+ researchers from Ohio, nearby states

 

Graduate students – from various disciplines and institutions across the country – are improving their multi-core programming skills this week during a summer school course offered by the Great Lakes Consortium’s Virtual School of Computational Science and Engineering.


High-definition streaming video enables students to participate in the course, “Many-Core Programming for Science and Engineering Applications,” from the National Center for Supercomputing Applications (NCSA) at the University of Illinois, the University of Michigan, the Electronic Visualization Laboratory at the University of Illinois at Chicago, and the Ohio Supercomputer Center (OSC), located on the west campus of The Ohio State University.


“We’re pleased to host more than 40 people from throughout Ohio and nearby states at the Ohio Supercomputer Center location,” said Steve Gordon, senior director of education at OSC and executive director of the center’s Ralph Regula School of Computational Science. “As America’s future science and engineering workforce, these participants will be better prepared to take advantage of high performance supercomputers.” 


Instructors Wen-mei Hwu, from the University of Illinois, and David B. Kirk, from the graphics-processor company NVIDIA, are teaching the course. The lessons provide students with hands-on experience in developing applications software for multi-core processors, such as general-purpose graphics processing units (GP-GPUs). Participating students have access to NCSA’s 32-node cluster of GP-GPUs and are learning to use the software CUDA to write programs for the cluster.


Petascale computer systems are capable of reaching performance in excess of one quadrillion floating-point operations per second. These emerging systems currently represent the fastest high performance computers in operation and are used for such tasks as weather and climate simulation, nuclear simulations, cosmology, quantum chemistry, and fusion science.


In an effort to prepare the next generation of computational researchers for emerging petascale computing systems, the Great Lakes Consortium for Petascale Computation is spearheading education and training programs such as this course. Petascale computing is more complex than previous computing paradigms and requires programmers to establish a solid grounding in computational science and engineering.