The University of Pennsylvania has taken a major leap in AI research with the launch of ‘Betty’, an off-campus supercomputer built to handle large-scale datasets and deliver refined, adaptive results.

The University of Pennsylvania has taken a major leap in AI research with the launch of ‘Betty’, an off-campus supercomputer built to handle large-scale datasets and deliver refined, adaptive results. Powered by NVIDIA’s SuperPOD reference architecture and equipped with 31 eight-way GPU nodes connected via NDR400 InfiniBand, Betty enables researchers to scale a single experiment across the entire system. “The needs for modern AI research have grown to a scale… where it is no longer feasible to be maintained by any one school,” said Kenneth Chaney, Associate Director of AI and Technology at Penn’s Advanced Research Computing Centre (PARCC).

Moving away from costly, individual lab servers, Penn now offers a centralised high-performance computing resource via PARCC. “In the future, we will be leveraging AI in all areas of research… I think we will see it adopted in many/most fields,” noted Marylyn Ritchie, Vice Dean of AI and Computing at the Perelman School of Medicine. She also highlighted the energy demands — with the current version requiring one megawatt of power — making its off-campus location necessary.

Named after Frances Elizabeth ‘Betty’ Snyder Holberton, one of the first six programmers of the ENIAC computer built at Penn in the 1940s, the system honours the university’s legacy in computing innovation. Fully accessible to researchers regardless of location, Betty marks a pivotal shift toward collaborative, next-gen AI-powered research at Penn.

The University of Pennsylvania has taken a major leap in AI research with the launch of ‘Betty’, an off-campus supercomputer built to handle large-scale datasets and deliver refined, adaptive results.

Tags: