Intel Perspectives – Competing Ahead of the Exascale Pack
By Naveed Siraj
It’s no surprise that governments across the globe – including China, Japan, the US, and Europe’s biggest players – are in a race to develop exascale systems. Being able to house the most powerful supercomputers will enable public or large private organisations make them the first port of call when looking to address current and future computing needs. It will create not only significant revenue and employment opportunities, but the chance to become the global hub of IT innovation for various industries. As I’m sure you can imagine, competition is rife.
Currently, Asia is on the way to take ahead of the pack. For instance, the world’s current fastest supercomputer, which is built entirely with Intel chips, was created by the National University of Defence Technology in China, while Australia has just launched its largest supercomputer at the Australian National University. Furthermore, over a quarter of the world’s top 500 HPCs are located in APAC. Without a doubt, Asia is well placed to grab its share of HPC revenue, which is expected to reach US $14 billion by 2016 according to IDC.
However, with the level of global investment in High Performance Computing increasing across the globe, what must Asia do to ensure it goes ahead of the game and is the first to develop exascale systems?
First and foremost, Asian developers must map a true path to exascale success, which I’m sure you can imagines is no easy feat. How can developers anticipate future needs, build platforms that are able to integrate technologies which in some cases may not even have been developed yet, and account for market changes?It seems an impossible task, but if Asia wants to stay at the cutting edge it needs to be able to answer these questions and many more.
At Intel,we are working with organisations in Asia (and across the globe), to provide the answers through our ‘pathfinding’ platform. The process involves identifying the desired product development ‘path’ and setting clear objectives alongside it to keep everything on track.
If countries in Asia are going to be successful in being the first to develop exascale systems, then technological breakthroughs are also going to be vital. An ability to increase processor performance is critical, which means developers must constantly look at ways to enhance memory technology, interconnect and integrate new functions into the processor, reduce power consumption, identify innovative cooling techniques, and identify new technologies delivering increased flexibility to software developers.
We know that memory is a major factor in the mix, and Intel’s commitment to Moore’s Law will ensure developers have access to processors capable of addressing their current needs today and in years to come.Unfortunately, we can’t yet rely on memory and processors scaling at the same rate, but there are some promising memory technologies that could change all that in the future.For example, Intel recently announced that the next generation of Intel Xeon Phi processor will deliver integrated on-package memory allowing leadership compute and memory bandwidth, and we’re researching new technologies includingmemory stacked on memory.
At Intel we’re also working to create world-class fabric solutions to meet the growing need of High Performance Computing (HPC) and scales with workload system requirement. Creating a single processor with integrated fabric controller results in fewer chips and therefore fewer chip crossings. In turn this delivers lower cost and power consumption, as well as higher performance and density, which are all critical hurdles that must be overcome on the journey to exascale success.
Another key consideration for Asian developers on their path to exascale success is power. If we were to reach exascale today, we would need the equivalent of two power plant’s worth of energy to power one exascale machine. This is of course an unrealistic approach, which is why we need to think about how far we can push energy efficiency today, and in the future with the various technological breakthroughs that are expected. While ‘Near-threshold Voltage Technology’, which improves energy efficiency at the cost of performance, is one possible solution, more innovation is needed in this area.
Encouragingly, Intel and European researchers have established three European labs with the goal being to create simulation applications that begin to address the energy efficiency challenges of moving to exascale performance.Similarly, we have in place effective parallel programming models where we work in close collaboration with academics and researchers to tackle the various issues affecting HPC.Through the subsequent findings, Intel will be well positioned to advise developers across Asia and the globe as they look to overcome the challenge of exascale power supplies.
The path to exascale by 2020 is certainly achievable, and Intel is supporting organisations across the globe as they attempt to be first over the line. If Asia is to continue leading the charge, then hurdling the challenges posed by energy constraints and adopting and identifying the latest technologies will make the journey much easier. There’s still some way to go, but the hard work already undertaken by developers in the region will ensure that – if they stay on track – Asia will have high chances in the leading pack coming to the finish line.