HPC and artificial intelligence are no longer limited to a few domains but are having an impact in all fields, from physics to linguistics.
Andrew Petersen
Research Computing Specialist and Principal Investigator
Researchers from the NC State Department of Physics published work in March on a quantum chromodynamics model that had not yet been studied numerically — a model that could give us a better understanding of the universe’s origins.
This work was made possible by powerful computing resources added to the OIT High-Performance Computing (HPC) infrastructure in the spring of 2024.
Funded by a National Science Foundation Campus Cyberinfrastructure (CC*) grant, awarded in June 2023, OIT purchased six Lenovo SR670 nodes with 24 top-of-the-line graphics processing units (GPUs).
“Our method for solving these equations relies completely on GPU,” said Josh Ott, co-author on the quantum chromodynamics paper and a physics student. “These simulations are very computationally intensive, and without the parallel capabilities of the resources on HPC, it would have taken us months rather than weeks to get the data we needed.”
OIT identified the need to expand its GPU resources to better support researchers on campus about four years ago, said Eric Sills, assistant vice chancellor for Shared Services. The growth of artificial intelligence (AI) applications was a big contributing factor.
“On top of their usual computing, they’re also adding AI,” Sills said of researchers. “These GPUs are a specialized resource that is especially useful for artificial intelligence uses.”
The HPC cluster supports more than 200 research projects, totaling 7.5 million individual jobs during the 2023-2024 fiscal year.
Because HPC services are available to all NC State faculty members, researchers across disciplines have equal access to these new resources, said Andrew Petersen, research computing specialist and principal investigator on the CC* grant.
“HPC and artificial intelligence are no longer limited to a few domains but are having an impact in all fields, from physics to linguistics,” said Petersen.
In addition to AI, the new GPUs provide computing power for a dozen scientific applications, including molecular dynamics, neural network training, genome assembly and fluid simulation.
The CC* grant, which has provided $467,607 in funding, requires 20% of the resource to be shared with the broader research community. This is being facilitated by making a portion of the GPUs available through the Open Science Grid, a nationwide consortium that provides researchers with shared high-throughput computing services.
According to Marc Hoit, NC State chief information officer and vice chancellor for information technology, these expanded resources are another stepping stone for the university to continue leading the way in the research community.
“Research is a critical component of NC State’s success. Increasingly, that research involves AI and machine learning,” said Hoit. “GPUs are the lifeblood for these efforts, and grants like this help grow our capabilities across campus.”
