Donor funds to upgrade Watson College computing cluster for AI research

To solve today's complex problems, faculty and students at Binghamton University's Thomas J. Watson College of Engineering and Applied Science need strong and reliable computing power.

The High-Performance Computing Cluster has been ready to help since it was established at Binghamton University's Innovative Technologies Complex in 2011. Nicknamed the Spiedie cluster (pronounced "speedy," after the Binghamton region's popular marinated chicken sandwich), it offers the super-fast compute nodes and graphics processing units (GPUs) needed for machine-learning algorithms and artificial intelligence.

Now, thanks to generous gifts given during the Binghamton Fund EXCELERATOR Challenge in April, the HPC has received a significant upgrade. The EXCELERATOR Challenge was a first-of-its-kind initiative to boost the successful multi-year EXCELERATE campaign, which concluded on June 30 with alumni and friends giving more than $261 million to advance Binghamton's competitive excellence as a premier public university.

As part of the two-day challenge, gifts designated for Watson College joined an unrestricted fund for special initiatives like improving the Spiedie cluster.

"We are very grateful to our generous alumni and friends who help Watson students achieve their goals and help elevate their academic experience," Dean Atul Kelkar said. "The money raised through the EXCELERATE campaign and other channels also will fuel our excellent researchers as they discover solutions that will benefit the world. The HPC is a true differentiator as our faculty pursue innovative research and compete with top researchers from around the globe."

Phil Valenta, Watson's interim director of information technologies, likes to call the HPC upgrade "Spiedie 2.0." It includes an ultra-low latency InfiniBand network that offers 800 gigabits per second of throughput, as well as nodes with top-of-the-line GPUs. The Watson IT staff also has made it easier to add new components later.

"We've laid the groundwork with this revamp where researchers can simply buy the hardware, and everything's already there for them - power, networking and cooling - so we can get it integrated quickly," Valenta says. "It was a little bit tougher to go through that process before the upgrade."

Mengen Wang, an assistant professor in the Department of Electrical and Computer Engineering, and PhD student Kejia Li use the HPC for semiconductor materials research. Through theoretical modeling, she and her students analyze surface and defect properties that could affect how semiconductor materials perform.

When she interviewed for a faculty position at Binghamton in 2022, Wang knew that access to computing power from the Spiedie cluster would be essential to pursuing her work.

"The cluster is a very important resource for our group," she says. "It's really accessible, and the IT teams are very supportive. No matter what we run on the cluster, they have the critical experience for debugging the code and helping us to run smoothly."

Li adds: "As a PhD student, having access to Spiedie has been incredibly important for my research. It allows me to run density functional theory computations efficiently, which is essential to my projects. With the recent upgrade, I can now take advantage of the improved GPUs for my machine-learning algorithm training."

Upgrading the Spiedie cluster helps not only Watson College - which is attracting more faculty and students each year - but all of Binghamton University. Researchers across campus can request access or pay a yearly subscription, depending on the level of computing power needed.

"Any student, undergrad or graduate working with faculty members who use the cluster will gain experience running code on the latest and greatest," Valenta says. "They can take that experience and tell employers that they've worked on this hardware to run their language models, AI or machine-learning processes."