 |
| NVIDIA helps create the largest 3D map in the universe |
NVIDIA and NERSC run Perlmutter.
Perlmutter described it as the world's fastest supercomputer for artificial intelligence workloads.
The supercomputer includes 6144 NVIDIA A100 Tensor Core GPU.
One of his missions is to create the largest 3D map of the visible universe.
Nvidia said, Mother of Pearl is the fastest system for processing workloads that includes 16-bit and 32-bit mixed-precision arithmetic in AI applications.
Later that year, Perlmutter added more AI supercomputing capabilities to Phase 2.
Nvidia writes: In a project, the supercomputer is helping to assemble the largest 3D map of the universe currently visible.
The computer processes data from the DESI dark energy spectroscopy. It is a cosmic camera capable of capturing up to 5,000 galaxies in a single exposure.
For example, researchers need the speed of the Mother of Pearl's GPU to capture dozens of recordings in one night so they know where to take DESI the next night.
It may take weeks or months for the data to be released for an old system. But the mother-of-pearl should help them accomplish their mission in less than a few days.
NVIDIA, artificial intelligence and machine learning:
People are exploring ever larger neural network models and need access to more robust resources.
Perlmutter is responding to this demand for artificial intelligence with an A100 GPU, a complete flash file system, and data streaming capabilities.
With Perlmutter, nearly 7,000 NERSC-supported researchers can access four exaFLOPS hybrid precision accounts for science projects using AI.
In addition to the DESI mapping project, researchers are working a lot on supercomputers in fields such as climate science.
Mother of pearl helps study subatomic reactions to discover green energy.
NVIDIA said: This project requires a special combination of artificial intelligence and high-performance computing provided by Perlmutter.
Traditional supercomputers can hardly handle the mathematical operations required to create simulations of a few atoms in a nanosecond using programs like Quantum Espresso.
By combining high-fidelity simulation with machine learning, scientists can study more atoms over a longer period of time.
Researchers are optimistic about the ability to use AI in high-performance computing.
The project also aims to uncover the dark energy and mysterious physics behind the accelerating expansion of the universe.