Team of Purdue and Colombian students share a common language - supercomputing

October 31, 2014

Student supercomputing competitions have taken Kurt Kroeger to Salt Lake City, Leipzig, Germany, and now New Orleans, but they also have him headed for an even more intriguing destination — his future.

Kroeger is part of the six-member combined Purdue and Universidad EAFIT, Colombia, team competing in the 2014 student supercomputing competition at SC14, the world’s largest supercomputing conference. The conference is in New Orleans Nov. 17-21. HP and Intel are sponsoring the Purdue and EAFIT team.

“I’ve found the competitions to be great learning experiences and an awesome opportunity to connect with people from the high-performance computing industry,” says Kroeger, a Purdue senior in computer science from Nashville, Tennessee. “It led to me interning in the field of distributed systems and I plan on working in that field after I graduate.”

The Purdue and Columbian team, one of 12 selected, will be competing against teams from Australia, China, Germany, Singapore, Taiwan and around the U.S. Purdue’s entry is the only international partnership and the only team with students from the Big Ten.

In addition to Kroeger, the Purdue part of the team includes Fangning “Alfa” Cheng, a sophomore in computer science from Jiangsu, China, and Matt Molo, a sophomore in computer science from Joliet, Illinois. ITaP Research Computing (RCAC) staff members Stephen Harrell and John-Michael Mulesa are advising the team. Michael Baldwin, associate professor of atmospheric science, has served as faculty advisor during the team’s preparations for the competition.

The Colombian team members include Alejandro Gómez-Londoño, Mateo Gómez Zuluaga and Pablo Restrepo Henao, all seniors in computer science at Universidad EAFIT in Medellin. Juan Guillermo Lalinde Pulido, a computer science professor and head of the supercomputing program at EAFIT, is faculty advisor.

Purdue and Columbia began a research, education and economic development partnership in 2010 and furthered it in January during a visit by Purdue President Mitch Daniels. EAFIT’s Apolo supercomputer is built, in part, of components Purdue donated from its decommissioned Steele cluster research supercomputer.

That all made collaborating on a student supercomputing team a natural fit, Harrell says.

“They wanted their students to have more experience,” Harrell says. “I think it’s good for our students to meet and work with people from other cultures. It can only benefit them.”

Given the combined, international makeup of the SC14 team, preparations have been a bit different than in past student supercomputing competitions, in which the team included only Purdue students. The team still meets regularly to fine-tune the hardware and software for its supercomputer and the Purdue students still do that in a classroom at Purdue, as does Alejandro Gómez, who is on the Purdue campus for an internship this semester. The other Colombian students are “there,” too, through a Google Hangout online collaboration space with video and audio, as well as Internet chat and email. Harrell calls it telecommuting.

“Working with the Purdue team has helped me improve my knowledge about high-performance computing and learn about new tools that I did not know before,” Restrepo says.

The student supercomputer is like a mini version of Purdue’s Community Cluster Program supercomputers, with the latest hardware from HP and Intel. ITaP has built six of these research supercomputers for its faculty partners since 2008. This gives Purdue the best collection of supercomputing power for use by faculty on a single campus in the U.S. Purdue researchers in dozens of disciplines all over campus — from engineering and physics to biology and pharmacy — use the community clusters to drive world-changing discoveries and innovations that can benefit the economies of Indiana and the nation.

Limited to undergraduates, the student teams have to prepare their machines to run an assigned variety of real research software and to crunch voluminous sets of sample data as quickly and efficiently as possible over two days. The 2014 applications are used for predicting storms and flooding, studying the actions and interactions of chemical molecules and examining systems as diverse as automobiles and wind turbines. Most of this year’s Purdue team members help manage Purdue’s community clusters as student workers for ITaP Research Computing (RCAC). The EAFIT students likewise work on the Apolo cluster there.

“The team from Colombia brings a new perspective into the competition,” Kroeger says “Their knowledge, combined with the knowledge of the Purdue team, gives us an edge we have previously not had.”

Originally posted: October 31, 2014  3:50pm