IBM to study universe with massive telescope
Screenshot by ZDNet
(ZDNET) IBM and Astron, a Netherlands-based astronomy organization, plan to develop an exascale computer system that will form the backbone of the largest and most sensitive radio telescope.
This exascale system will be designed for low-power usage and plug into the international Square Kilometre Array (SKA). SKA will be built out with the help of astronomers from more than 20 countries. The telescope will have millions of antennas and have a collection area of about one square kilometer. SKA will also be able to scan the width of the continental United States.
Under a 5-year, 32.9 million euro ($43.9 million) deal, the two groups will build the Astron & IBM Center for Exascale Technology. The telescope, due in 2024, will need processing horsepower that will be on par with several million of today's fastest computers. The project is funded by the Province of Drenthe, the Netherlands, and the Ministry of Economic Affairs, Agriculture and Innovation (EL&I). Astron (short for Netherlands Institute for Radio Astronomy) and IBM have collaborated before on the Blue Gene supercomputer, which analyzes data from Astron's low-frequency array.
The Astron telescope will aim to spot new galaxies, dark matter, and the origin of the universe. To accomplish that goal, the telescope will need a high-performance computing architecture and fast data transfer. The 5-year collaboration between Astron and IBM, dubbed DOME, will look at emerging technologies for everything from storage to analytics to develop the telescope. Data will be analyzed and collected daily.
That daily data stream will be equivalent to twice the world's Internet traffic each day -- a few exabytes a day for a single beam per one square kilometer. As with most analytics projects, the trick for Astron and IBM will be finding the key signals in that mass of information. IBM also gets to try out its 3D stacked chips and other technologies. As is the custom with IBM's research efforts, the company is looking to collaborate with outside scientists and then commercialize the findings.
IBM said that the location for SKA has to be finalized, but a decision is expected in 2012. Australia/New Zealand and South Africa are the two leading options to install the millions of antennas required.
A few key figures:
- SKA will need to store 300 to 1,500 petabytes of data per year.
- The CERN hadron collider stores 15 petabytes per year on average.
- 18 exabytes is the limit of what today's 64-bit architectures can address.
- SKA will aim to explore the universe dating back 13 billion years.
This story originally appeared at ZDNet's Between the Lines under the headline "IBM, ASTRON eye exascale systems to scan universe."
Popular in SciTech
- Microsoft announces Xbox One
- Microsoft's new Xbox: What to expect, where to watch
- Watch: NASA captures Okla. tornado from space Play Video
- Microsoft announces Xbox One 16 Photos
- Power of Okla. tornado surpassed Hiroshima bomb
- The 7 weirdest things made by 3D printing
- Storms that spawned deadly Oklahoma tornadoes seen from space (video)
- NY official: Airbnb stay illegal; host fined $2,400
The information received from the telescope must be classified in a way that resembles the expanding universe. For example the telescope can be the center (particle) of an expanding sphere that comprises all the celestial objects on the spheres layers (wave). Classifying the data with its time/space coordinates ensures that all objects can be viewed from its center and from any other objects center (particle) also on another spheres surface (wave).
So already by pointing the telescope at a certain point in the universe, the time/space classification structure will ensure that its coordinates are filtered and stored only once in the relational/smart data base.
Physically storing the petabytes can be done by modifying solar panel grids to act like gigantic RAM memory chips.