100 PetaFLOPS Supercomputers to Emerge in 2017 - Expert

ExaFLOPS-Class Machines on Track for 2019 - 2020

by Anton Shilov
02/09/2011 | 07:12 PM

Nowadays machines capable of performing a quadrillion (1015) floating point operations per second (FLOPS) are only beginning to emerge. But already in six years from now supercomputers with a hundred of times higher performance will take the stage thanks to continuous development of microprocessors.

 

"Everything is moving along according to Moore's law, so things are doubling every 18 months, roughly. Computer makers in 2017 will have 100PFLOPS  systems and [ExaFLOPS-class] systems between 2018 and 2020. All of these things are dependent on funding. This will happen if the funding is in place for those machines," said Jack Dongarra, a computing expert at the University of Tennessee, reports InformationWeek.

This is not the first time when Mr. Dongarra makes no doubts that supercomputers capable of performing a quintillion (1018) FLOPS will emerge towards the end of the decade thanks to evolution of silicon process technologies as well as processors. Earlier he also predicted mobile chips with TeraFLOPS-class performance sometimes in 2020.

The governments of many countries now understand the importance of high-performance computing as well as benefits its brings to many industries. As a result, various HPC projects now gain help from government organizations from around the world, which indirectly speeds up development of new technologies necessary to make PetaFLOPS-class supercomputers more affordable and ExaFLOPS-class systems a reality by the end of the decade.

Many experts claim that exascale supercomputers will be heterogeneous and will utilize both traditional central processing units as well as special highly-parallel accelerators, such as AMD FireStream, Nvidia Tesla, Intel's forthcoming MIC or other architectures that will deliver maximum horsepower amid moderate consumption of energy. In fact, heat and power consumption are currently believed to be the main challenges for exascale computing.