The LANL Parallel Computing Summer Research Internship program started four years ago. This internship is an intense 10-week program aimed at providing students with a solid foundation in modern high performance computing (HPC) topics integrated with research on real problems encountered in large-scale scientific codes. This 10-week summer program combines lectures and a research project to develop expertise in HPC applications development
The New Mexico Consortium supports this program by sponsoring the participation of three students from the three New Mexico research universities for all four years. Bob Robey, one of the original founders of the parallel computing school, was pleased to have this offer of sponsorship of New Mexico university students because it fit within the programs plan to develop a strong pipeline of skilled professionals to fill in a critical need for Los Alamos National Laboratory (LANL). This last year, the three students, Brian Romero from University of New Mexico and Jeff Keithley and Ben Mastripolito from New Mexico Tech were selected by a panel of research scientists to fill the positions. Besides doing excellent work on their projects this summer, all of them may return for future work with LANL.
“One of the strengths of the student sponsorships from the New Mexico Consortium is that students are more likely to continue their student research at LANL and eventually become staff members. We also get the benefit of strengthening the education and research program at New Mexico Universities, thereby developing a steady stream of well-qualified prospective staff. Just in the four years of the program, we already have a high return rate from students in the program” – Bob Robey
Brian Romero, one of the participants this year, worked on porting the Higrad atmospheric simulation code to new architectures. He was able to get the code running on the graphics processors (GPUs) and get it to speed-up by a factor of 35. This means that a month-long calculation can now be done in one day. Higrad is the underlying computational kernel in the Firetec wildland fire modeling application. The ability to run wildland fire scenarios fast enough to inform fire management has long been a dream of the Firetec team. With Brian’s work, a large step has been made to achieving this goal.
Jeff Keithley and Ben Mastripolito worked on two different projects that were exploring how to get the most performance out of current computing hardware. With the advent of the Exascale era, computing has hit limits on how small computer circuits can be miniaturized and how much power can be used. This means that scientists can no longer just throw more computer processors at the problem to make it run faster. They must understand how to get every bit of performance out of the existing hardware. This requires a fundamental understanding of the factors that impact the performance of real applications. Jeff and Ben made great strides in getting performance measurements of application performance and how to optimize them.
According to Jeff, “I really enjoyed collaborating with the other students in the program on various issues we had in common between our projects. This program enabled me to learn how to perform real research and present results.” Ben states that, “Being a Parallel Computing Summer Research intern has helped me understand what a research position entails, and how I might go about finding one. This experience has also given me a lot more insight to how a national lab runs and what kind of work is available.” Ben also says he enjoyed working with other interns, learning about research at the labs, and the learning experience of attending talks.
To learn more about this program and how to apply see the LANL Information Science and Technology Institute’s (ISTI) Parallel Computing Summer Research Internship webpage.
Photo by Carrie Talus