IBM and the Netherlands Institute for Radio Astronomy, better known as ASTRON, have announced that they are collaborating on energy efficient, exascale computer systems that will be able to process the enormous amounts of data produced by the forthcoming Square Kilometre Array (SKA) radio telescope. The SKA will use millions of antennas across an area more than 3,000km (around 1,850 miles) wide, with a combined area of around one square kilometer, and is intended to explore the history of space.
This vast array will result in an enormous amount of data being collected — scientists expect that the telescope will deliver a few exabytes of data per day, meaning between 300 and 1,500 petabytes will need to be stored each year. This is more than double the daily traffic of the internet worldwide, which means that that a powerful, low power computing system will be needed to process it all.
This is where the €32.9 million (about $44 million) DOME project comes in: the team plans to develop an energy-efficient exascale computer system, working with data provided by the existing Low-Frequency Array (LOFAR) in the Netherlands for testing. The researchers say that they will need to explore architectures beyond current state-of-the-art systems in order to handle this information, including advanced accelerators and 3D stacked chips designed to boost energy efficiency, optical interconnect technologies, and new experimental storage methods.