Fugaku Could Be First Exaflop Supercomputer in 2021 If US and China Delay Beyond 2020

Fujitsu PRIMEHPC technology could build exaflop Fugaku supercomputer in 2021. Japan’s new supercomputer Fugaku is set to begin operations around 2021 with the country aiming to regain the title of building the world’s fastest computer, replacing its current supercomputer K, government-backed research institute Riken. The Fugaku, a nickname for Mount Fuji, aims to be about 40 to 120 times faster than the K, the first supercomputer in the world to achieve a speed of over 10 quadrillion computations per second.

PRIMEHPC supports the same Tofu interconnect D as Fugaku for maximum scalability in large configurations, and ultra-dense implementation of up to 384 nodes per rack, FX1000 can efficiently build large systems that deliver theoretical computing performance in excess of 1.3 exaflops. This model uses a water-cooling system, aiming for high performance and reduced total energy consumption including in cooling.

The United States and China have both revealed plans to release supercomputers with equal computing ability to the Fugaku in 2020 or 2021.

The Department of Energy’s Argonne National Laboratory in Chicago expects to deliver the exaflop Aurora supercomputer by 2021.

Fujitsu announced that it will begin global sales of the PRIMEHPC FX1000 and PRIMEHPC FX700 models from the “Fujitsu Supercomputer PRIMEHPC” series, which utilize the technology of supercomputer Fugaku jointly developed by RIKEN and Fujitsu. Shipment is scheduled for March 2020.

2 thoughts on “Fugaku Could Be First Exaflop Supercomputer in 2021 If US and China Delay Beyond 2020”

  1. They run physics simulation , to study anti-matter, how particle collide with each other. Also when one government trying to brute force other nation server , etc. you need all the power you can get

  2. These chips sound pretty pathetic actually. The new AMD Rome server chips sound a lot more impressive. They have very high benchmarks, and good performance per Watt. https://www.tomshardware.com/news/amd-epyc-radeon-frontier-exascale-supercomputer,39275.html
    Well, not Rome exactly, it will be a customized version. No idea what the modifications are, but I gotta think it has to worth the extra time and money, or they would just buy off the shelf Epyc Rome chips. Perhaps modified to put more than two chips on a board? Perhaps 4 or 8? Or maybe they want 7nm IO, and something else in there using the saved space. Perhaps something to help with code breaking.

    To tell the truth, I can’t see any need for these massive supercomputers. These are more powerful than you need to do…whatever. There was a time when we really needed them because computers were not that powerful…but now, just one or two chips or a couple of powerful GPUs can do a lot of heavy lifting. Why you need 12,000 eludes me. The only answer I come up with is national pride. And that just seems pathetic.

    Makes more sense to me to have lots of smaller supercomputers than these massive ones.

Comments are closed.