The latest supercomputer may help solve a few of the most significant questions of astrophysics. As per VentureBeats, the National Energy Research Scientific Computing Center officially has dedicated Perlmutter supercomputer to help build the biggest 3-dimensional map of visible universe for studying the dark energy that’s boosting the expansion of cosmos.
Perlmutter, dubbed as one of the fastest AI supercomputers, will process information from the DESI (Data Energy Spectroscopic Instrument) for guiding observations. The machine would process several galaxies from one night’s research for dictating where DESI focuses next.
As per AMD, the initial Perlmutter version includes around 1,536 nodes. Each of these nodes have four NVIDIA A100 GPUs and one 64-core Epyc 7736 chipsets. The end-result is one system that comprises of nearly four exaflops while handling AI related tasks and lower calculation times. Perlmutter could even process one year DESI data in only a couple of days rather than months or weeks, claimed NVIDIA.
One second phase that’s due later this year will be adding 3,072 CPU nodes, with each having dual of the Epyc chipsets.
Notably, the map is not the sole project at hand. Perlmutter will be studying the atomic interactions as well for fostering clan energy technology such as biofuels, for instance.