College’s Super Computer Cluster Awaits a New Name

The college’s new super computer has already enabled dozens of faculty, students, and staff to conduct research at lightning speed. Next week, it will receive its official name, as decided by a community vote.

College’s Super Computer Cluster Awaits a New Name
The Super Computer boasts 1024 CPU cores and 126,336 GPU cores. Photo Courtesy of Amherst College IT.

The college’s high-performance computer (HPC) was unveiled in 2022 and has 1024 CPU cores and 126,336 GPU cores (hundreds of times more than the typical laptop), but it does not yet have a name.

Following a community vote, the mathematics and statistics department and the computer science department are hosting a “name drop” on Wednesday, April 10, complete with student poster sessions, talks, and refreshments. The computer, currently referred to as “the cluster,” has five potential names: BitMammoth, FrostByte after Robert Frost, Kleene Cluster after mathematician Stephen Kleen ’30, Mammoth Matrix, and Romer Cluster after Betty Steele Romer, former director of computing at the college.

Professor Lee Spector, of the computer science department, and Professor Amy Wagaman, of the mathematics and statistics department, received a grant from the National Science Foundation to buy the computer in August 2021. The supercomputer, which resides at the Massachusetts Green High Performance Computing Center, a LEED-Platinum certified facility in Holyoke shared by seven Massachusetts institutions, allows members of the Five College Consortium to crunch numbers at speeds that would enable previously impossible research.

Spector contrasted the cluster with the everyday laptop.

“Laptops are optimized for your user experience; you’ve got a nice screen, it’s lightweight, and so on. A supercomputer is where you just care about the computation. And you’re going to try and pack as much computational power as you can into a unit, and you’re going to care about the speed,” he said.

Users began onboarding in January 2023. Currently, 44 faculty members, 236 students, and nine staff members have active accounts on the cluster.

“Some of our astronomers deal with huge amounts of picture data … that are in the terabytes or petabytes. You can’t do that on a computer with a single processor,” Wagaman said … “The clusters have some ridiculous number of CPUs and GPUs … so that it can handle the bigger jobs because it’s distributing the workload over more things.”

Students have been eager to explore research using the supercomputer, covering topics such as Bayesian statistics, molecular dynamics, and neural networks. Some computer science courses also utilize the cluster for class projects.

“Our research was on just developing this general algorithm that you can use to solve a bunch of problems … it’s inspired by concepts from evolution,” said Andrew Ni ’24, who used the cluster for computer science research. He explained how his research used high-speed computing to test thousands of random algorithm iterations to produce the best problem-solving methods.

“This research quite literally would have been impossible on just my [laptop] … if I was to run one problem [without the cluster], it would probably have taken 100 hours, but […] I had to run seven problems and 100 times each. I don’t have 170,000 hours,” Ni said.

He added that he only utilized a fraction of its full computing power during his research and that the maximum capacity of the entire computing system would be even more impressive.

STEM students, however, aren’t the only ones benefiting from the cluster.

“These days, there are all sorts of uses of AI modeling in social sciences. In the arts, people can make generative AI systems to do all sorts of things in many fields,” Spector said. “There are so many things that are now enabled by high-performance computing.”

For instance, the UVM Art + AI Research Group at the University of Vermont used supercomputing to create an art exhibition titled “Aberrant Creativity: Unusual Partnerships Between Humans and Machines” to explore the relationship between humans and machines.

“We [the IT department] have been developing solutions that are no code or low code … if a barrier to students or users outside of STEM is coding, it should be as simple as using the technology tools on your laptop,” said Doug Hall, a research computing specialist at Amherst Information Technology.

As Amherst anticipates the supercomputer’s new name, those already in the know are optimistic about its fit. “The [name] that’s most prominent in my memory is the one that won,” Spector said. “There were some good suggestions [that] all came from the community … and then there was a vote, and I think [the results] are good news.”

Correction, April 3, 2024: A previous version of this article did not include statistics in reference to the mathematics and statistics department.