SRNL Acquires High-Performance Computing Machine to Tackle Big Data Problems
A new HPE Apollo 6500 Gen 10 Plus System, dubbed “Raptor,” has a new home at the Savannah River National Laboratory. The technology is equipped with premier graphics processing units (GPUs) that will provide a significant upgrade to the scientific computing capabilities at SRNL.
“When I came [to the lab], we had a 2006 GPU that was available, and that’s almost two decades old,” Cyber Strategy Director Glenn Fink, Ph.D., said. “The applications of AI have been growing so fast, it’s on an exponential curve. So, getting this machine is an important investment for us to move our research to the next level.”
“Raptor” features eight of the latest type of GPU, the H100, which can deliver industry-leading conversational AI. The lab will be using this new technology to support researchers first in the Laboratory Directed Research & Development program and later externally sponsored projects. Fink also said this type of power gives the lab the capability to assist in providing data science solutions to big data problems coming directly from lab sponsors. One of those issues being advanced cyberattacks.
SRNL has acquired “Raptor,” a new technology that will provide significant upgrades to the capabilities within the lab’s research programs. Photo from Hewlett Packard Enterprise (HPE)
“We want to help our [cyber] defenders use AI, in a positive way, to detect advanced attacks that are happening,” Fink said. “AI has become very good at emulating humans, and it would be good to know whether an attacker is human or not, so we needed updated hardware to do these things.”
SRNL Chief Information Officer Randy Coleman and his team, Director of Innovation and University Engagement Liz Hoffman, and the LDRD team at SRNL all played a role in the acquisition of the “Raptor” machine. Fink noted that the LDRD program will offer training for scientists and engineers because of the great difference between programming central processing unit (CPU) simulations and big data intensive GPU applications. Armed with this training, Fink says it will be time to let them be proactive and imagine the future, problem-solving together with their sponsors.