Nvidia and the National Energy Research Scientific Computing Center (NERSC) on Thursday flipped the “on” switch for Perlmutter, billed as the world’s fastest supercomputer for AI workloads. Named for astrophysicist Saul Perlmutter, the new supercomputer boasts 6,144 NVIDIA A100 Tensor Core GPUs and will be tasked with stitching together the largest ever 3D map of the visible universe, among other projects.

Perlmutter is “the fastest system on the planet” at processing workloads with the 16-bit and 32-bit mixed-precision math used in artificial intelligence (AI) applications, said Nvidia global HPC/AI product marketing lead Dion Harris during a press briefing earlier this week. Later this year, a second phase will add even more AI supercomputing power to Perlmutter, which is housed at NERSC at the Lawrence Berkeley National Laboratory.

“In one project, the supercomputer will help assemble the largest 3D map of the visible universe to date. It will process data from the Dark Energy Spectroscopic Instrument (DESI), a kind of cosmic camera that can capture as many as 5,000 galaxies in a single exposure,” Harris wrote in a blog post announcing the news.

“Researchers need the speed of Perlmutter’s GPUs to capture dozens of exposures from one night to know where to point DESI the next night. Preparing a year’s worth of the data for publication would take weeks or months on prior systems, but Perlmutter should help them accomplish the task in as little as a few days,” he wrote.

Supercharging HPC with AI and machine learning

Firing up an AI-optimized supercomputer “represents a very real milestone,” said Wahid Bhimji, acting lead for NERSC’s data and analytics services group.

“AI for science is a growth area at the U.S. Department of Energy, where proof of concepts are moving into production use cases in areas like particle physics, materials science, and bioenergy,” he said.

“People are exploring larger and larger neural-network models and there’s a demand for access to more powerful resources, so Perlmutter with its A100 GPUs, all-flash file system, and streaming data capabilities is well timed to meet this need for AI,” Bhimji added.

Perlmutter will give NERSC’s approximately 7,000 supported researchers access to four exaflops of mixed-precision computing performance for AI-assisted scientific projects. In addition to the DESI mapping project, researchers are teeing up time with the supercomputer for work in fields like climate science, where Perlmutter will assist in probing subatomic interactions to discover green energy sources.

That project, which will generate simulations of atoms interacting, requires the special blend of AI and high-performance computing (HPC) that Perlmutter delivers, Harris said.

“Traditional supercomputers can barely handle the math required to generate simulations of a few atoms over a few nanoseconds with programs such as Quantum Espresso. But by combining their highly accurate simulations with machine learning, scientists can study more atoms over longer stretches of time,” he said.

The ability to leverage AI in supercomputing also has researchers optimistic about the DESI project. In addition to mapping the known universe, the project “aims to shed light on dark energy, the mysterious physics behind the accelerating expansion of the universe,” NERSC data architect Rollin Thomas said. System namesake Saul Perlmutter, who remains a working astrophysicist at Berkeley Lab, was awarded the 2011 Nobel Prize for Physics for his contributions to the discovery of dark energy.

“To me, Saul is an example of what people can do with the right combination of insatiable curiosity and a commitment to optimism,” Thomas said.

He added that in preparatory work with researchers to get code ready for Perlmutter supercomputer workloads, NERSC was already seeing 20x faster GPU processing performance than in previously available systems.