“We choose to go to the Moon! ...We choose to go to the Moon in this decade and do the other things, not because they are easy, but because they are hard; because that goal will serve to organise and measure the best of our energies and skills, because that challenge is one that we are willing to accept, one we are unwilling to postpone, and one we intend to win ...”
President John F. Kennedy made that speech in September 1962. He was reiterating a commitment made a year earlier to put a man on the moon and bring them back safely before the end of the decade.
The crew of Apollo eventually achieved that goal in 1969 although sadly Kennedy was not around to see it. With the Vietnam War raging at the time, there was also some scepticism about whether the race to the moon was really a worthy endeavour.
But despite some criticism (as well as crazy conspiracy theories about whether it really happened) the technical complexity and sheer audacity of Apollo 11 makes it an often-cited analogy for ambitious technology projects. For example, the space race was referred to frequently during a recent session on exascale computing - a system capable of a billion, billion calculations per second - at the Data Center Dynamics Zettastructure conference in London in early November.
In a session entitled ‘The race to Exascale – meeting the IT infrastructure needs of HPC’ a panel of experts discussed the benefits of achieving the next big breakthrough in supercomputing. Peter Hopton, founder of UK-based liquid cooling specialist Iceotope argued that China is focused on being the first to exascale but is less interested in the benefits the breakthrough could bring. “They will be able to achieve the required number of flops to say they have done it. But it’s a bit like Neil Armstrong landing on the moon and then taking off again without stepping out of the door because they didn’t build a door on the capsule,” he said. “It’s kind of pointless as you are doing it to say you landed someone on the moon rather than get out, get some moon rocks and do the science.”
Hopton is the dissemination leader (European Union speak for marketing) for a €20 million EU research project into exascale known as EuroExa. The project objectives include developing and deploying an ARM Cortex based system with FPGA acceleration at petaflop level by 2020. The consortium hopes this will enable the development of an exascale system in 2022/23. Unlike some rival initiatives, the 16 organisations involved in EuroExa are concentrating their efforts on building a useable and sustainable platform said Hopton. “The European approach is very much focused on the application,” he said.
Also on the panel was John Goodacre, director of technology and system at ARM and professor of computer architectures at Manchester University. Also part of EuroExa, he described how the European consortium plans to build a system from the ground up. “We have been working in Europe for a number of years for what comes after multi-core computing,” he said. “There is a view that we can’t get to exascale by simply putting more of today’s technology in or optimising any specific part. So we have been working from the transistor and how you put them together to build something all the way up to the infrastructural density.”
Another panellist, Tate Cantrell from data centre operator Verne Global, agreed that the point of exascale research should be to push scientific thinking forward in a similar way to the Apollo programme. “Why did JFK say we choose to go to the moon? The rest of that paragraph ends with ‘we choose this goal because it will force us to organise and measure our skills and our energies’,” he said. “I think that is what exascale is all about. It is about pushing science forward…it’s about creating the materials that are going to create the next generation of compute.”
Iceotope’s Hopton also emphasised that the real benefits of a concerted effort around exascale would be the many potential spin-off technologies. “Things were developed to get man to the moon but as a result of those things being developed other things spun out into society,” he said. “The same is occurring here. We are being given money by the EU to do research for a crazy big supercomputer but we are then taking the cooling technology that we developed with that money and are able to put that into normal data centres to improve efficiency or capture waste heat to heat homes.”
Exascale could also have benefits for healthcare according to another panellist Dr Tim Cutts, head of scientific computing at genomic research organisation The Wellcome Trust Sanger Institute. “You will have a low power but a very high performing small unit of compute which you can stick in a doctor’s surgery close to a DNA sequencer and they can have an answer for you then and there. That is where we want to get to and the way that technology develops is through the HPC world.”
But although the panellists were in broad agreement about the benefits of exascale research, the Apollo analogy was brought up again in response to questions around whether there should be more international cooperation. “Going back to the moon example, would the Americans have bothered doing all of that if the Russians hadn’t been trying as well? No, they wouldn’t,” argued Cutts.
Hopton agreed that just as with Apollo there was a certain amount of national interest driving the research but also scope for cooperation. He referred to the Apollo–Soyuz Test Project (ASTP) in 1975 that saw a US Apollo command module dock with a Soviet Soyuz craft. “There is vanity, there is nationalism but there is also cooperation. Back with the Apollo programme there was the Apollo-Soyuz dock in space. International cooperation occurred in space between two rival nations,” he said.
Ultimately, analogies between exascale research and the space race may be a bit of a stretch. For example, computer researchers aren’t being asked to put their life on the line in the same way as the Apollo astronauts. But the prediction that exascale will have benefits for society beyond its immediate goals seems likely. And just as with Apollo, it seems making the breakthrough will be worthwhile not because it will be easy, but because it will be hard.