Peak Performance

Designed with general purpose users in mind, the Fugaku supercomputer has the performance and energy efficiency of a GPU-accelerated machine but remains easy to program.

Tackling the biggest questions

Even as the pandemic continues, Fugaku has been used to further research in other areas. Scientists from Tokyo Medical and Dental University and Fujitsu Laboratories were able to estimate a network of regulatory relationships among genes in epithelial cancer cells. The work, which was based on analysis of data on 20,000 genes, would have taken several months on university supercomputers but was completed in less than a day on Fugaku.

The supercomputer will be used to crunch numbers related to other social and scientific issues of high priority such as drug discovery, personalized and preventive medicine, earthquake and tsunami simulations, big data weather and climate prediction, energy generation, transport and storage, materials science, next-generation design and manufacturing, and basic science research such as the fundamental laws and evolution of the universe.

“In terms of disasters, Japan gets a truckload every year, and supercomputer-derived results that have been used for disaster mitigation have already paid for the supercomputer in terms of money and lives,” said Matsuoka.

He’s personally intrigued by the potential of Fugaku to create a virtualized smart city, including virtual sensors, IoT terminals and networks, as well as virtual environmental and weather, because it’s difficult to scale experimental parts of a city that have been fitted with instruments. If the experiments yield positive results, it could be applied to real urban settings.

Fugaku could also be used to come up with new technologies or designs for next-generation supercomputers including potential successors to Fugaku and, further down the road, quantum computers. Finally, the supercomputer can be used to simulate cognitive activity in higher mammals including humans—in 2013, the K computer simulated one second of human brain activity.

“It’s quite phenomenal that we can run a human brain-scale simulation, and it may lead to discoveries of brain phenomena that are not well understood,” said Matsuoka. “We can also use it for artificial intelligence research and driving science with hypotheses, such as the next state of a molecular structure, and then using physics to validate them.”

If Fugaku must live up to, and surpass, the research legacy of the K computer—whose achievements included simulating a human heart at the molecular level, a feat that contributed to some 1,300 research papers—then it’s already off to a promising start. There are great expectations riding on this billion-dollar machine.

This article was first published in the January 2021 print version of Supercomputing Asia.
Click here to subscribe to Asian Scientist Magazine in print.


Copyright: Asian Scientist Magazine; Illustrations: Oi Keat Lam.
Disclaimer: This article does not necessarily reflect the views of AsianScientist or its staff.

Tim Hornyak is a Canadian writer based in Tokyo, Japan, who has worked in journalism for more than 20 years. He has written extensively about travel, food, technology, science, culture and business in Japan, as well as Japanese inventors, roboticists and Nobel Prize-winning scientists. He is the author of Loving the Machine: The Art and Science of Japanese Robots and has contributed to several Lonely Planet travel guidebooks. He has lived in Tokyo for more than 15 years.

Related Stories from Asian Scientist