Yesterday European commissioner and VP of the European commission Neelie Kroes announced the new initiative to get Europe back on track in the High Performance Computer world (HPC)
1,000,000,000,000,000,000 operations per second. Yes, my editor told me to never start a sentence with a number, but that´s what exa-scale means. A staggering 10 to the 18th flops/s. To put this number into perspective: the fastest super computer in the world, Japan´s K computer, leads the official list with 10 Petaflops/sec. So the new European computers would be a 100 times as fast!
Sounds futuristic and impossible, but the catch is that the project aims to achieve that performance level before 2020. So the European industry has a maximum of eight years to reach this very ambitious goal. Considering that between 2008 and 2012 the performance leader in the super computer field got from 1 to 10 Petaflops, it could be achievable if Moore´s law still holds.
Moore, the Intel co-founder, predicted that computer speed would about double every two years. So theoretically it´s possible to get a hundred fold speed increase in 8 years. However the problem with scaling super computers to these extremely high levels is another limit: the power consumption. To be realistic the maximum budget is about 20MW to keep cooling under control, making energy efficiency key to achieving this result. This will be an important part of the project and stimulate spin-offs.
What can you do with that amount of computing power at your disposal? Well typical high performance computer (HPC) areas are weather prediction, nuclear research, and in the US: intelligence services. The American NSA is the world´s largest eavesdropper and to do so, they specialize in code breaking. Decrypting messages is extremely challenging so it´s rumored that the NSA wants to get hold of an exa-scale machine around 2018.
The European commission has different objectives. Europe doesn´t have the large defense industry funding so the goals are civilian. The purpose is to stimulate economic growth and create novel HPC appliances to stay competitive. The European car industry saved 40 billion Euro by simulating new car platforms, instead of building and testing early prototypes. Even better, the time to market went down from five years to only two years by using HPC. A crucial advantage in a competitive market.
Medical research is another area where large progress can be made through the use of extremely advanced super computers. Examples are hospitals in Germany quotes as being able to avoid last-minute decisions during childbirth. Analysis of 3D brain imaging makes early diagnosis of for instance Alzheimer much easier.
To achieve these ambitions the EU commission wants to stimulate the industry with a package worth 1.3 billion Euro doubling the current efforts. The non profit organization PRACE will be the center point of the efforts as they already bundle and coordinate the European efforts in HPC.
This new ambition makes clear to the world that Europe is not about to give up in the race to technological supremacy. Over the decades there was a shift from the United States as the leading nation in super computing, to Japan which currently has the fastest computer. Rapidly upcoming China owns the no. 2 place and the no 4 on the top 500 list. The US holds the third place at the moment while the fastest European computer can be found at the ninth spot.
If Europe would succeed in achieving this goal with such a relatively small budget compared to the military financed projects in the US and China it would be a great accomplishment.
What´s your opinion: good to use European money to help the industry achieve these ambitions and create high tech jobs?
Author: Max Huijgen
Further reading:
_http://www.prace-ri.eu/_
_http://www.top500.org/lists/2011/11/press-release_
Picture attached is the K computer system at Riken’s laboratory in Japan.
