FDA panel recommends Johnson & Johnson COVID booster for 18 and up Xbox Series X mini fridge Red flag emoji Apple's Oct. 18 event: How to watch Registration to buy a PS5 PS5 Pro

Good news for future Macs: The fastest supercomputer uses Arm chips

The same processor family that powers phones and likely the next generation of Apple computers is at the heart of the world's fastest machine.

The Fugaku supercomputer at Japan's RIKEN center, named the world's fastest machine in 2020, uses Fujitsu-designed Arm processors.

The Fugaku supercomputer at Japan's Riken center, named the world's fastest machine in 2020, uses Fujitsu-designed Arm processors.

Riken

In a development that oddly enough bodes well for Apple Macs, a Fujitsu machine has been named fastest in the world. The system, called Fugaku and installed at the Riken Center for Computational Science in Japan, features processors that are members of the Arm family that Apple will use in its Mac personal computers later this year.

Fugaku uses Fujitsu A64FX processors that are bigger and more power-hungry than the various Arm processors that are inside smartphones today and will soon power Macs. But the supercomputer helps show that Arm-powered machines can reach the loftiest of performance heights.

Now playing: Watch this: Apple announces its transition to ARM-based chips with...
6:14

That's good news for those who worry that high-end Macs won't be competitive without the Intel chips they've relied on since 2006. Apple announced Arm-based Macs on Monday at its WWDC event for developers, an expansion of its use of A-series chips in iPhones and iPads. Apple didn't comment for this story but at WWDC promised high performance from the chips.

Fugaku topped the Top500 list of supercomputers with sustained performance of 416 quadrillion mathematical calculations per second, or 416 petaflops. That's nearly triple the IBM Summit supercomputer it bumped down into second place. But it also uses nearly three times the power -- 28.3 megawatts instead of 10.1 megawatts for Summit. That's enough power for more than 11,000 homes, according to federal energy consumption rates.

Supercomputers are mammoth machines whose fastest models can occupy rooms the size of basketball courts and cost hundreds of millions of dollars. They tackle the hardest computing challenges people are willing to pay for, like simulating nuclear weapons explosions, predicting the Earth's future climate, modeling aircraft aerodynamics, training artificial intelligence systems and more recently, seeking drugs to fight the coronavirus.

Fugaku is a big achievement in the race to build so-called exascale supercomputers, models that will surpass 1 quintillion calculations per second, or 1 exaflops -- about 2.5 times Fugaku's speed. The US Energy Department is funding Hewlett Packard Enterprise and AMD to develop and build a 2-exaflops machine called El Capitan that's expected to be the world's fastest supercomputer when it arrives in 2023.

The Riken research institute isn't the only one interested in the Fujitsu chips. Cray, one of the oldest names in supercomputing and now part of HPE, offers supercomputers using the Fujitsu A64X chips. Customers include Los Alamos National Laboratory, Oak Ridge National Laboratory, Stony Brook University and the University of Bristol.

The Fugaku supercomputer uses 48-core Fujitsu A64FX processors. They're much more powerful cousins to the Arm chips found in iPhones and Android devices.

The Fugaku supercomputer uses 48-core Fujitsu A64FX processors. They're much more powerful cousins to the Arm chips found in iPhones and Android devices.

Fujitsu

Arm, a UK company, licenses chip designs to companies like Qualcomm. But it also licenses the chip instruction set -- the commands software uses to control it -- to companies like Apple that create their own designs. Fujitsu designed the A64FX in cooperation with Arm.

Each A64FX has 48 separate processing cores to churn through calculations in parallel -- many more than the handful on typical mobile gadget processors. Everyday computing benefits from having fewer, faster cores, while supercomputer jobs are often more amenable to being broken down into parallel jobs.

Researchers update Top500 list twice annually in conjunction with the Supercomputing conference.