Nvidia, an American technology company, has reportedly unveiled Perlmutter, billed as the world’s fastest supercomputer, at the NERSC (National Energy Research Scientific Computer Center), California.
Perlmutter was named after Saul Perlmutter, an astrophysicist, and is expected to help users overcome AI workloads. This supercomputer will initially be tasked with developing the most ambitious 3D universe map, leveraging the 6,000+ Nvidia A100 Tensor Core GPUs. It is apparently capable of offering over 4 exaFLOPS of AI performance, which will make it the fastest system on 16- & 32-bit mixed-precision math AI applications. Additionally, its performance is likely to be bolstered following the introduction of the 2nd tranche of CPU cores.
Following the completion of the system development, Perlmutter will become one of the top 5 supercomputers available across the globe, which assesses the overall performance using the HPL (High Performance Linpack) benchmark. For the record, above 7,000 researchers will put Nvidia’s Perlmutter system to work, while several applications have been already readied with an aim to advance astrophysics and climate science.
As per the statement made by Nvidia’s HPC and AI Product Marketing Lead, Dion Harris, Perlmutter will aid a project of assembling the largest 3D universe map to date. The supercomputer will also process data from the DESI (Dark Energy Spectroscopic Instrument), which is a type of cosmic camera capable of capturing up to 5,000 galaxies in a single exposure, he added.
The universe map, once complete, will help researchers gain more knowledge about dark energy, the force behind the universe’s ever-increasing expansion. Similar other projects are also developed to run on the new Perlmutter machine, many of which will adopt the unique qualities of Tensor Cores in A100 to boost interactions between atoms.
The projects are expected to prove revelatory across the field of materials science as they will uncover ways to develop efficient biofuels, batteries, and the like.