Japan To Build ‘Fugaku Next’ Zetta-Class Supercomputer,1000X Faster Than Current Supercomputers

Sarfraz Khan
Japan To Build 'Fugaku Next' Zetta-Class Supercomputer,1000X Faster Than Current Supercomputers 1

It will take $750M to build Japan's Fugaku Next supercomputer but it will be the fastest in the world with Zetta-Class performance.

Japan's Fugaku Next supercomputer will reach speeds in zettaFLOPS and will be up by 2030

The battle of supercomputers is about to escalate as we now have Japan setting on a journey to build the first Zetta-class supercomputer. Surprisingly, Oracle may develop the first Zetta-class supercomputer before Japan as in the news yesterday. Regardless of who comes first, Japan is reportedly preparing its next supercomputer named 'Fugaku Next'.

Related Story Aurora Supercomputer With Intel Ponte Vecchio Fails To Beat All-AMD Frontier In Exascale Race
Japan's Fugaku Supercomputer is powered by Fujitsu's 48-core microprocessor that is based on Arm's architecture for the chips. The Fugaku is the world's fastest supercomputer and the A64FX is manufactured on Taiwan Semiconductor Manufacturing Company (TSMC)'s 7nm process node.

Fugaku Next will succeed the current fastest supercomputer in Japan 'Fuguaku' and will be over 2000 times faster. Fugaku can achieve up to 442 petaFLOPS of performance, which is equivalent to 442 quadrillion floating-point operations per second. This enabled the Fugaku to enter the TOP500 supercomputers list and is currently ranked number four. However, Fugaku Next will reach up to a zettaFLOPS in performance, which is equivalent to one sextillion floating-point operations per second.

For comparison, the current fastest supercomputer, Frontier can reach up to 1.206 exaFLOPS or 1.206 quintillion floating-point operations per second. The Fugaku Next will achieve a performance of around 1000 times higher than Frontier. This is currently only imaginable in theory but is possible as (MEXT)Japan's Ministry of Education, Culture, Sports, Science and Technology is heading this project, and is ready to spend over 750 million dollars.

In the first year, MEXT will invest around 29 million dollars but will increase the spending as the development progresses over the next years. The project is estimated to be complete by 2030 and will make Fugaku Next probably the first supercomputer to reach a zettaFLOPS in performance.

Oracle NVIDIA

However, as Oracle revealed yesterday, it's possible that the OCI Supercluster will be the first to achieve this feat and will be even faster, offering up to 2.4 zettaFLOPS in performance or 2.4 sextillion floating-point operations per second. For this, Oracle is going to deploy 131,072 upcoming NVIDIA Blackwell datacenter GPUs, which is almost 3 times the GPUs Frontier uses at the moment.

This, however, requires unprecedented amounts of power to operate smoothly. The Fugaku Next will need energy equivalent to the output of 21 nuclear power plants and is one of the biggest challenges Japan has to face to run the supercomputer efficiently. Fugaku Next will probably be cross-compatible with Fugaku due to the usage of components from the same company, Fujitsu. With such a big project,  Japan is planning to advance in several fields like AI, Machine Learning, Astrophysics, Energy Research, Medical Research, Climate Modeling, and much more.

News Source: Techspot

Sarfraz Khan Photo

About the author: Sarfraz Khan is a hardware reporter with a focus on PC components and the builder community. With years of experience writing about PC hardware and laptops, his work has been featured on several reputable technology publications. Sarfraz's hands-on experience is demonstrated through his first-person accounts of using and comparing different hardware configurations, providing practical and relatable insights for everyday users. His technical analysis is respected by peers in the enthusiast community and has been cited by specialized hardware sites such as Germany's Igor's Lab.

Follow Wccftech on Google to get more of our news coverage in your feeds.

Button