IBM scored a contract to be exploring the feasibility of an exascale computer system that is powerful enough to collect and store the data generated by the Square Kilometre Array (SKA) telescope.
The SKA is currently developed as a project to result in the world's largest telescope to investigate the origins of the universe. According to current plans, the SKA would include 3,000 45-foot wide dish antennas that are spread over a surface area that is more than 3,000 km or 1,860 miles wide and would either reach from South Africa to the Indian Ocean islands or from Australia to New Zealand. The combined data collecting area of all dishes is about 1 square kilometer (0.30 square miles), which gives the system its name.
IBM says that the a single 1 square kilometer beam of the SKA will deliver "a few exabytes" of data every day, which is 10 to 100 times more than the annual data production of the LHC at CERN and more than the capacity of 15 million 64 GB iPods. After processing, about 300 to 1,500 PB of data will need to stored annually.
An exascale computer system will be required to handle the data processing when the system is projected to go online in 2024. A final decision if, when and where the SKA will be built is expected to be made this year. Companies such as Intel and IBM have said that exascale system may be available in the 2020 time frame. The world's fastest supercomputer currently is just above 10 PFlops of performance, while IBM has laid out plans for a 100 PFlops system with more than 8.3 million processing cores. Exascale systems are predicted to integrate more than 100 million processing cores.