Students and faculty at the University of Wisconsin-Eau Claire can engage in more high-performance, data-driven research thanks to a new public-private collaboration with Hewlett Packard Enterprise that dramatically increases the university’s supercomputing processing power.
Contributions totaling more than $700,000 from HPE, the National Science Foundation and UW-Eau Claire are enabling the university to purchase a high-performance computing cluster/supercomputing cluster and develop the Blugold Center for High-Performance Computing.
HPE contributed $363,426 in hardware and UW-Eau Claire contributed $20,000 to the project. A $350,000 National Science Foundation grant also supports the center.
“Thanks to Hewlett Packard Enterprise and the National Science Foundation, UW-Eau Claire is the first university in the United States to use HPE’s Apollo servers with the Cray hardware called Slingshot,” UW-Eau Claire Chancellor James Schmidt says. “This investment epitomizes the power of partnerships to drive innovation and provide mutual benefits. Not only do UW-Eau Claire’s students gain the chance to conduct research in areas ranging from deep learning and artificial intelligence to data mining and computational number theories, HPE’s leaders and technicians can now pioneer new technologies and receive expert feedback from our talented faculty and staff.”
The additional 61 computer servers at an off-campus regional data center at Chippewa Valley Technical College can do some computations up to 100 times faster than what is possible with the existing Blugold Supercomputing Cluster that has 25 servers in UW-Eau Claire’s Phillips Science Hall.
”As HPE considered where to locate our Global Center of Excellence for Performance Computer Manufacturing, the advantages of staying in Chippewa Falls became clearer and clearer as we engaged Chancellor Schmidt and UW-Eau Claire faculty and staff,” says Adam Bauer, director of Issues and Policy Communications for Hewlett Packard Enterprise. “As a future-focused organization manufacturing supercomputers just up the road from this campus, the opportunity to partner with the university was a no-brainer. Together, we can advance research in Eau Claire to levels commensurate with any major university in the country. At the same time, students will have the opportunity to learn and work with us as undergraduates, laying a strong foundation for future employment.”
The high-performance computing cluster at the regional data center form the Bose Cluster, named after Indian mathematician and physicist Satyendra Noth Bose. The powerful servers will provide UW-Eau Claire undergraduates with even more high-impact learning opportunities as the university becomes a regional hub of high-performance computing, says Dr. Sudeep Bhattacharyay, associate professor of chemistry and biochemistry.
“One of the major goals of this university is to provide research opportunities to as many students as possible,” Bhattacharyay says. “This new computing facility can offer a large number of students this high-impact practice without any further investment. Our hope is that this increased computational power will also increase the innovative discovery of the undergraduate students on this campus.”
The original cluster on campus cannot handle the growing demand for computer processing power at UW-Eau Claire, say Bhattacharyay and Dr. Ying Ma, associate professor in materials science and biomedical engineering. Bhattacharyay and Ma wrote the NSF proposal and received the $350,000 grant.
The increased capacity provided by the regional data center will meet UW-Eau Claire’s teaching and research needs for the foreseeable future, Ma says.
“HPE really is a big deal for us because the current cluster we have right now really limits what we can do in the cluster,” Ma says. “This new cluster opens many new doors.”
Ma and Chip Eckardt, senior information processing consultant at UW-Eau Claire, say computational science is becoming increasingly important in all branches of science.
“Computational science, which requires high performance computing, is going to be absolutely where everybody wants to be in the next three to five years,” Eckardt says.
Calling the regional data center “a huge deal,” Eckardt says the computing upgrade places the university on a par with doctoral universities that participate in the highest levels of research activity, classified as Research 1 by the Carnegie Commission on Higher Education.
Supercomputing demand exceeds capacity
UW-Eau Claire’s supercomputing history dates to 2008, when the chemistry department purchased a small cluster of servers for $25,000.
The Blugold Supercomputing Cluster started in 2014, but campus demand for high-performance computing quickly outpaced capacity.
Multiple academic departments — biology, chemistry and biochemistry, computer science, geology, mathematics, materials science and biomedical engineering, and physics and astronomy — now use the cluster, but many more departments will have access to it with the additional computing power of the Bose Cluster.
“In the departments of chemistry and biochemistry there are several research groups that are dependent on this high-performance computing, as well as materials science,” Bhattacharyay says. “There are a number of other university departments that depend on simulation, not just quantum mechanics-based computation practices, but other different molecular simulations. All faculty will have a very strong interest in this high-performance computing.”
The additional computer processing capacity is “really a game-changer” for UW-Eau Claire, says Dr. Rahul Gomes, an assistant professor of computer science.
“I look at this as a beginning of a new era where UW-Eau Claire emerges as the hub of deep learning and image processing in the entire Midwest,” Gomes says.
“Deep learning” is a subset of machine learning that derives its working logic from the functioning of the brain. Due to its high accuracy, deep learning finds application in self-driving cars that process high-resolution images on the fly.
The Bose Cluster will provide the computing power undergraduate researchers at UW-Eau Claire need as they learn how to apply deep learning to process big data, Gomes says. By logging into the cluster, students can experience how supercomputers operate, submit jobs and process images in batches.
“One of the problems in deep learning for image classification is that it requires a lot of computing power, especially graphics processing units,” says Gomes, who joined the UW-Eau Claire faculty in 2020. “Most of the time we have to resort to other external resources in order to process those data sets which require significant time and introduce compatibility issues. But if our students get the experience of processing data sets at the university itself, it will definitely be a good experience for them.
“I can tell my students we have a supercomputing cluster where we can process all our data. And students get to do it. It’s pretty exciting to see that here. Being a new professor at this university, it opens up a lot of opportunity for me in collaboration and research. It’s pretty awesome.”
For example, the additional processing power will allow the computer science and geography departments to collaborate to process high-resolution satellite images, Gomes says.
Biology researchers spend considerable time in the field, but also use new technologies to generate genomic sequence data for hundreds of individual organisms with multiple sequences, says Dr. Nora Mitchell, assistant professor of biology. Mitchell expects to have sequence data for her students to analyze in the coming months.
“We might get several thousands of pieces of data per individual,” Mitchell says. “When we go to analyze that data, it requires a lot of computing power to analyze and process all of those samples. For many students, we don’t have the software that they would need. But if we have it on the computing cluster we can run it all together, freeing up myself to do teaching and for students to do their other coursework.”
Mitchell says the supercomputing cluster allows students to experience research opportunities ordinarily only found at major research institutions, including using supercomputing tools to work on their own research.
“Often students think that computer science or coding can be intimidating, but having this computer cluster is an amazing resource and really helps students to scaffold those skills,” Mitchell says. “It shows them that it’s not scary, but really exciting, to be working with your own data and getting results from samples students collected and analyzed. That’s a really big reward for them.”
Dr. Mckenzie West, assistant professor of mathematics, also sees the possibilities in using the supercomputing cluster. For example, she hopes to use the supercomputing cluster in her computational number theory course, allowing her students to compute properties of large numbers and solve difficult equations.
High-resolution images also are a key in the biomedical field, Ma says. The Bose cluster creates yet another opportunity for UW-Eau Claire faculty and students to collaborate with its partner Mayo Clinic Health System physicians and others.
“Medical science has become increasingly data driven,” Ma says. “You need a lot of data to make an informed decision, a diagnosis or treatment. This cluster will be the perfect tool to collect those data, analyze those data and then benefit medical science.”
Seeing UW-Eau Claire and Mayo Clinic Health System working together using the computer cluster may inspire other Chippewa Valley industries to consider how they also could benefit from the technology, Bhattacharyay says. Additional collaborations could help other businesses as well.
“There will be a synergy if some other industries see that high-performance computing is being used for teaching purposes and research purposes in a university like ours,” Bhattacharyay says. “Our cluster has all these different capacities, all these different capabilities, so that will open different industries, different institutions that can come to us and we can satisfy their different needs.”
Students gain experience in supercomputing operations
The supercomputing cluster gave Alyssa Huelsbeck, a senior from Little Chute, an opportunity to perform “incredibly relevant” COVID-19 research during her biophysical chemistry course in fall 2020. Bhattacharyay, the course instructor, allowed students to continue their research with the supercomputer cluster during the spring semester, giving them a deeper scientific understanding of the research, she says.
“The supercomputing cluster has been great because we’ve gotten to do a lot of research on the structure of the virus without having to do the wet-lab work,” Huelsbeck says. “We’ve been able to get a lot of data and it’s been relatively straightforward using the supercomputing cluster, so it’s been a great resource this semester.
It’s incredible to be working on coronavirus research while still an undergraduate student, says Huelsbeck, a biochemistry/molecular biology and Spanish major.
“The world’s top scientists right now are working on coronavirus research and we’re undergraduate students getting firsthand experience with the supercomputing cluster,” Huelsbeck says. “It is an amazing experience.”
Huelsbeck is among the UW-Eau Claire students to present their research this spring at the virtual National Conference on Undergraduate Research.
Gomes says Blugolds who have supercomputing operations skills will have an advantage when looking for their first jobs.
“Once they start applying for positions in companies, they can actually say they did a lot of machine learning and used our Blugold Supercomputing Cluster to do that machine learning,” Gomes says.
“It’s a completely new discipline, and the fact that we have it here in order to test our codes and give our students an opportunity to test our codes, it’s basically expanding the curriculum.”
Bhattacharyay agrees, noting that the supercomputing cluster will create experiences undergraduate students are not likely to find elsewhere.
“We can train students, we can provide the opportunity for them to actually try their ideas on the cluster and even to make some mistakes,” Bhattacharyay says. “That opportunity is unique; it’s not like you can get that experience anywhere.”
A team of UW-Eau Claire students manage the supercomputing cluster. Previous student administrators have received internships and found jobs because of their experience, Bhattacharyay says.
HPE’s involvement in the project creates even more opportunities for students, he says.
“The reason we are so thrilled is because HPE has an assembly unit in Chippewa Falls where they bring all those computing nodes and equipment and assemble them,” Bhattacharyay says. “Having a collaboration with HPE will give our students a lot of opportunity in terms of training and high-performance computing maintenance. Students can learn high-performance computing maintenance and serve the entire campus.”
Bhattacharyay says local schools and their teachers also will use the computer cluster through summer training programs.
Supercomputing’s link to Chippewa Falls
UW-Eau Claire is the first university in the U.S. to run the Cray hardware called Slingshot with HPE’s Apollo servers, which significantly increases processing speeds moving data to and from the servers, Eckardt says.
Like Bhattacharyay, Eckardt believes HPE’s involvement in the project makes the collaboration even more valuable. Eckardt grew up in Chippewa Falls and remembers the impact Cray Research and its founder, Seymour Cray, had on the region. Now that HPE owns Cray, it makes sense for UW-Eau Claire to partner with the company, he says.
“We can help HPE with doing research and testing for them and maybe we end up doing training of their users,” Eckardt says. “The possibilities are amazing and the support people are literally 22 miles down the road. You don’t find that anywhere else.”