Earlier this Summer, President Obama signed an executive order that established the National Strategic Computing Initiative (NSCI), a project that aims to build the world’s fastest computer by 2025. This computer, an ‘exascale machine’, would be capable of making one quintillion (a billion billion) calculations per second - a figure known as one exaFLOPS (floating point operations per second).
The executive order stated: ‘Over the past six decades, US computing capabilities have been maintained through continuous research and the development and deployment of new computing systems with rapidly increasing performance on applications of major significance to government, industry and academia. Maximizing the benefits of HPC in the coming decades will require an effective national response to increasing demands for computing power, emerging technological challenges and opportunities, and growing economic dependency on, and competition with, other nations.’
There has been steady progress in computing power over the last two decades, broadly following ‘Moore's law’ - the observation that, since the dawn of computing hardware, the number of transistors in a dense integrated circuit has doubled around every two years. The United States Department of Energy (DOE) funded supercomputer research which led to the first teraFLOPS (1 trillion FLOPS) computer in 1997 and a petaFLOPS computer (1 quadrillion FLOPS) in 2008. The best US machine at the moment can achieve 17.59 petaFLOPS, while China's current supercomputer is the Tianhe 2, which is capable of 54.9 petaFLOPS of computational power at its peak. This is being upgraded in 2016 to process 100 petaFLOPS.
The exaFLOPS computer represents a whole new challenge, with the main obstacle being the energy demands of running it. The computer components have to be far more power efficient than they currently are, and even then, the electricity demands would be massive - roughly 60 megawatts, costing a minimum of $60 million a year.
There are, however, a number of potentially game changing benefits. An exascale machine could solve some of the greatest challenges facing the world today, aiding scientific research and national security projects, as well as weather modeling and medical applications. The computing power of an exaFLOPS computer could more accurately model the Himalaya watershed, determine how genes work on the molecular level and provide a better understanding of how brain synapses work. Henry Markram’s Blue Brain Project, for instance, estimates that a full simulation of the human brain would require about an exaFLOPS. This could lead to huge breakthroughs in AI.
NASA scientists also claim that it would allow them to model turbulence, which could lead to more streamlined aircrafts being designed - without the need for extensive wind tunnel testing.
The forming of the NSCI is the latest move in the race to build an exascale machine, with China out in front at the moment. Startup company, Optalysis, is also in the pack, claiming that they will be able to create an optical solver supercomputer with an astonishing 17 exaflops machine by 2020.
President Obama’s executive order did not set out how much funding the NSCI project would receive, but it likely that the investment required will be eye watering. Estimates as to when an exaFLOPS computer will be possible vary from 2018 to 2025, whether this is the US or another project. Whoever manages it, the benefits will be world changing.