NSF Grant Puts Truman in the Supercomputer Game

Through a recent National Science Foundation award of nearly $700,000, Truman is one of four schools that will soon develop a high-performance computer (HPC), commonly referred to as a “supercomputer.” The consortium, comprised of Missouri Western State University, Webster University and Southeast Missouri State University, are all primarily undergraduate institutions, which was by design. Although the majority of NSF funding tends to go to research universities, schools like those in the consortium have been estimated to produce approximately 40% of STEM bachelor’s degrees.

“With computational resources becoming nearly ubiquitous in both academic research and a wide range of career sectors, we think it is important to provide students with the opportunity to work with HPCs as part of their education,” said Colin DeGraf, assistant professor of physics and co-principal investigator for the NSF grant. “The experience and skills they can gain from this should help them with their next step after graduation – whether going to graduate school or entering the job market – and also make them more productive in any computational field they might enter.”

Beyond career readiness, supercomputer access will expand potential research opportunities for Truman students and faculty members alike. That was a key factor in DeGraf’s involvement with the consortium. His current research examines how galaxies collide, which can involve looking at data from 15 to 20 million galaxies.

“On a very personal level, my research is computational. All of the research that I work on are using what are called cosmological simulations,” he said. “It’s a simulation that attempts to model as much of the universe as possible. Running those really requires a national- or international-level supercomputer.”

To solve the type of computationally intensive problems involved with research such as DeGraf’s requires a machine that can do a lot of calculations in a short amount of time. Central processing units have gotten faster over the years, but there are still limits in areas such as how many transistors can fit on a chip, or how to handle the heat they produce or transmission delays. Similarly, the number of cores a computer has will increase its speed. Modern home computers can have multiple cores, but it still is not enough to process the amount of data in some research. An HPC counters this speed problem by utilizing nodes, which are multiple servers networked together. Each node works almost like its own computer, but they can also work together to tackle bigger and harder problems. A single program can be run across multiple nodes resulting in more power and the ability to perform larger, more computationally expensive jobs. The current plan for the HPC in this project calls for 20 nodes, with each node having 128 computing cores and 512 GB of RAM.

“In addition to single jobs using multiple nodes, each node can be used separately, and you can even have multiple people all using a single node at the same time, depending on how many resources each user requests, which gives a lot more flexibility,” DeGraf said.   

Having access to this kind of computing power will allow for more cutting-edge research at Truman and can enable projects which would otherwise not be feasible. While his own research will benefit immediately, DeGraf foresees students getting the most out of this project.

“The cluster is being designed to have several nodes which are prioritized for educational use,” he said. “For example, if a professor wants to use the HPC for a computing lab, they can reserve nodes for their class time – whether for a single instance, or for an entire semester – so their students have guaranteed resources available to them.”  

The NSF grant includes some funding to send students to a summer workshop at the Linux Cluster Institute so they can then act as student leaders on campus to help others make the most of the supercomputer. This will also provide them with additional hands-on experience with HPC use, administration and construction.

Students and researchers who benefit from the HPC could come from almost any scientific discipline. In their NSF application, DeGraf and his fellow investigators included cases ranging from astrophysical simulations, computational chemistry, data science, cybersecurity and genetics.   

“We wanted to focus on how the HPC would improve research and educational opportunities at primarily undergraduate institutions in as wide a range of subjects as possible, rather than focus on a single research area,” he said.

Truman and its partner schools on the HPC project will operate as the Computational Infusion for Missouri Undergraduate Science and Education (CIMUSE) consortium. Initially, that group will consist of only the four institutions listed in the grant, but eventually more will be invited to participate. All primarily undergraduate institutions across Missouri will be eligible to join, and the CIMUSE consortium will look to expand in order to maximize the use of the HPC and the impact it has on both faculty researchers and undergraduate students.

“The goal of this project is to bring more supercomputer access to students across Missouri,” DeGraf said. “It will be used for faculty research, but also we want the best for all of our students, and the more experience we can give them, the better suited they will be.”

The NSF grant will begin in December. The CIMUSE has already begun planning meetings regarding hardware and software needs and purchases, as well as user policies. The HPC itself will be physically housed at the University of Missouri’s HPC center, but it will be accessible from anywhere in the world.

Tentative plans are for Truman faculty and students to have HPC access as soon as the next academic year. To learn more about this project, listen to DeGraf’s recent appearance on the Missourinet podcast “Show Me Today.”