Oak Ridge National Laboratories' Frontier is the world's fastest computer
A few days ago, my partner Darkcrizt tand counted on the list of fastest computers in the world. In this article I am going to tell you about the characteristics of the first computer on the list, with which the United States regains leadership in the computing industry.
I am referring to Frontier, which was officially built by Oak Ridge National Laboratories for the US Department of Energy. I say officially only because it would be rare for them to waste the potential of that equipment for military uses, but don't listen to me. Today I put on the aluminum hat
Learn more about Frontier
To understand what follows in the article, let's start by defining some terms:
FLOPS: English acronym for Floating Point Operations Per Second.
Name Abbreviation Value
kiloFLOPS kFLOPS 103
MegaFLOPS MFLOPS 106
GigaFLOPS GFLOPS 109
TeraFLOPS TFLOPS 1012
PetaFLOPS PFLOPS 1015
ExaFLOPS EFLOPS 1018
ZettaFLOPS ZFLOPS 1021
YottaFLOPS YFLOPS 1024
As we can see, an exaflop is equivalent to a trillion operations per second.
Which led to the Department of Energy's Oak Ridge National Laboratory's Frontier supercomputer being ranked highest as the world's fastest computer in the 500th edition of the TOPXNUMX list was its performance of 1,1 exaflops. The Frontier system goes down in history as the first to reach a hitherto unattainable level of computing performance known as exascale, we're talking a threshold of quintillion calculations per second.
However, its developers go further. Frontier features a theoretical maximum throughput of 2 exaflops, or two quintillion calculations per second, which means ten times more computing power than the Summit system also developed by the Oak Ridge National Laboratories. The system will enable scientists to develop technologies applicable to the country's energy, economic and national security issues, helping researchers tackle problems that were impossible to solve just five years ago.
Speaking to the press, Thomas Zacharia, director of the ORNL, was not exactly modest:
Frontier is ushering in a new era of exascale computing to solve the world's biggest scientific challenges.
This milestone offers just a preview of Frontier's unmatched ability as a tool for scientific discovery. It is the result of more than a decade of collaboration between national laboratories, academia, and private industry, including the DOE (United States Department of Energy) Exascale Computing Project, which is implementing the applications, software technologies , hardware, and integration needed to ensure impact at exascale.
But Frontier's accomplishments aren't limited to performance. It was also ranked number one on the Green500 list, which rates the energy use and efficiency of commercially available supercomputing systems, with a performance of 62,68 gigaflops per watt. Frontier rounded out the biannual rankings with the top spot in a newer category, mixed-precision computing, which rates performance in formats commonly used for artificial intelligence, with a performance of 6,88 exaflops.
Start up
Frontier's delivery, installation and testing work began during the COVID-19 pandemic. It required more than 100 people from the public and private sector who had to work 24 hours a day on tasks ranging from obtaining millions of components to guaranteeing delivery of system parts on time, including careful installation and testing. 74 HPE Cray EX supercomputer cabinets, including more than 9400 AMD-powered nodes and 90 miles of network cables.
Frontier Components
- Frontier has 74 supercomputer cabinets HPE Cray EX, specifically designed to support the performance and scale of next-generation supercomputing.
- Each node contains an optimized EPYC™ processor and four AMD Instinct™ accelerators, for a total of over 9400 CPUs and over 37 GPUs across the system.
- HPE Slingshot, the world's only high-performance Ethernet fabric designed for AI and HPC solutions technology, including larger and data-intensive workloads, to address demands for higher speeds and congestion control to keep applications running smoothly and improve performance.
- An HPE I/O subsystem. The I/O subsystem features an in-system storage layer and Orion, an enhanced Luster-based core file system that is also the world's largest and fastest single parallel file system.
The same as always, every time a new supercomputer comes out, it seems that it is going to revolutionize the world and then nothing, how many decades has the world already had supercomputers and why? I am not against them, but the cancer and many other things continue and I don't see that these maquinones really contribute, what they are supposed to contribute, but since they are supposed to contribute a lot, well, they are welcome.