Nvidia has unveiled the Perlmutter supercomputer, which it says is the world’s fastest for AI operations. The computer will be involved in many fields to advance science.
Nvidia unveiled Perlmutter, which it describes as the world’s most powerful AI supercomputer, at the National Center for Energy Research Scientific Computing (NERSC) in California. Perlmutter will be used to advance science in many areas, such as creating a 3D map of the universe and researching subatomic interactions for green energy sources, according to the statements.
The supercomputer is named after astrophysicist Saul Perlmutter. The system will consist of two stages and currently, only the first stage is available, Nvidia said, adding that the supercomputer will offer about four exaflops ai performance for more than 7,000 researchers, even without the inclusion of the second phase.
The system, which will be completed later this year with the inclusion of the second phase, is expected to rank among the top five supercomputers in the TOP500 ranking of High-Performance Linpack (HPL) comparisons.
Featuring 6,159 Nvidia A100 Tensor Core GPUs, the supercomputer will create the largest 3D map of the universe ever. To create the 3D map, the computer will process data from the Dark Energy Spectroscopic Instrument (DESI), which can capture up to 5,000 galaxies in a single exposure.
The process of preparing annual data for publication, which will take weeks or months on previous systems, can be carried out in as little as a few days thanks to Perlmutter.
“Perlmutter’s ability to combine artificial intelligence and high-performance computing will lead to breakthroughs in a variety of fields, from materials science and quantum physics to climate modeling, biological research, and much more,” said Nvidia CEO Jensen Huang.