Frontier, a supercomputer built by Oak Ridge National Laboratory, is the first capable of an exaflop – a billion billion operations per second
The world’s first exascale computer, capable of performing a billion billion operations per second, has been built by Oak Ridge National Laboratory in Tennessee.
A typical laptop is only capable of a few teraflops, or a trillion operations per second, which is a million times less. The exaflop machine, called Frontier, could help solve a range of complex scientific problems, such as accurate climate modelling, nuclear fusion simulation and drug discovery.
“Frontier will offer modelling and simulation capabilities at the highest level of computing performance,” says Thomas Zacharia at Oak Ridge National Laboratory.
The Frontier computing system is housed in 74 separate cabinets, comprising 9400 CPUs, or standard computer processors, and 37,000 GPUs, which are processors designed to render 3D graphics but can also be used for a range of other tasks.
This means the machine has a total of 8,730,112 cores capable of parallel computing tasks; a typical laptop has between five and nine. At peak power, the computer generates so much heat that it requires four high-powered pumps to push more than 25,000 litres of water around the machine each minute.
Frontier’s exaflop performance means that not only is it number one in the TOP500, an international collaboration to rank the world’s most powerful supercomputers, but it also represents a quarter of the computing power of the entire list.
“One machine represents 25 per cent of the total performance of the whole list, so it’s a very, very impressive achievement,” says Simon McIntosh-Smith at the University of Bristol, UK.
Frontier has yet to achieve its final form, either. Over the coming months and years, as its software is optimised, it could reach a theoretical peak of 2 exaflops.
In the past, supercomputing milestones have been shortly followed by many more machines with similar capabilities. While there are several exascale machines planned for the next couple of years, it isn’t clear how widespread this technology might become.
“The rate of improvement in electronics has slowed down somewhat, so we don’t expect exascale machines to proliferate through the TOP500 as quickly as it did for petascale, for example,” says McIntosh-Smith, referring to machines with a thousandth of the capabilities.
More on these topics: