May 27, 2021 was the day that the world’s fastest AI supercomputer was launched. Tech giant Nvidia and National Energy Research Scientific Computing (NERSC) were behind the making of this supercomputer. This supercomputer is the fastest supercomputer ever. This supercomputer is named Perlmutter. This super computer is named after Saul Perlmutter ji.
Saul Perlmutter, who is also an astrophysicist at Lawrence Berkley National Laboratory. Talking about “Perlmutter”, this super computer is said to be the fastest in performing artificial intelligence (AI) related tasks.
Nvidia’s product marketing lead revealed Perlmutter’s details via an official blog post. Harris stated that Perlmutter is “the fastest system on the planet” to process workloads, including 16-bit and 32-bit mixed-precision arithmetic, which are used in various artificial intelligence applications.
However, in the beginning, Perlmutter would be tasked to create the largest 3D map of our universe so far before moving on to other AI-based projects in the future. In addition, researchers will add even more AI computing power to the system later this year as part of its second phase of development.
“In one project, the supercomputer will help assemble the largest 3D map of the visual universe ever. It will process data from the Dark Energy Spectroscopic Instrument (DESI), a one-of-a-kind cosmic camera that can capture 5,000 galaxies in a single exposure,” wrote Dion Harris in a blog post.
Now, coming to the computer’s internals, Perlmutter packs 6,144 Nvidia A100 Tensor Core GPUs.
So, over six thousand Nvidia GPUs will power the system to process some of the most complex AI-based tasks. It is expected to give more than 7,000 NERSC researchers access to four exaflops of mixed-precision computing performance for AI-based scientific research.
“Traditional supercomputers can barely handle the mathematics required to generate simulations of a few atoms in a few nanoseconds with programs like Quantum Espresso. But by combining their highly accurate simulations with machine learning, scientists can study more atoms over a longer period of time,” said Wahid Bhimji, acting head of NERSC’s data and analytics services team.