The world’s fastest supercomputer is expected to go into production in 2018 and will run on Nvidia’s next generation Pascal graphics processing unit.
The world is expected in 2019 to have the fastest supercomputers in the world, according to a new report from Gartner, the world’s leading computer security and technology research firm.
The new supercomputer, called the P2X, will have a performance of 10 petaflops, or 10 billion calculations per second, according the company.
Gartener said the new supercomputing system will be used in the next generation of big data and machine learning, among other things.
The system is expected by the end of 2020 to run on a Xeon E5 processor, which is the fastest Xeon processor ever.
That processor is currently being used in Intel’s new $4.5 billion Xeon Phi processor.
The report does not say when the new P2Y system will start being used.
Nvidia, the graphics processing company behind the new system, announced in June that it would be making the P3X supercomputer available for private consumption by early 2020.
The company has already made it available to a handful of customers, including Google and Amazon.
Gabor Snedka, chief technology officer of Nvidia, said the company is aiming to offer its P2XX supercomputer to customers by the start of 2021, as well.
“We expect this system to offer supercomputational power in the millions of compute units, while offering superior computing performance to existing supercom systems,” Snedkas said in a statement.
The P2XY system will run at speeds of up to 20 petafls per second.
The price tag of the system is not yet known, but the Gartners report states it will be priced around $30 billion.
The supercomputer will be able to crunch up to 10 petabytes of data a second, Garters said.
It will be powered by the first generation of Nvidia’s Vega graphics processor.
It was announced in May that the supercomputer would be manufactured at Nvidia’s main factory in Nevada, in the Las Vegas area, and be based on a 12nm node.
“This supercomputer was designed to accelerate the design and manufacturing of supercomputed data centers,” Gartening said in the statement.
“It’s designed to make it easier to scale up and scale down large data centers.”