Will High Performance Compute See the Stars?

Located in Australia and South Africa and headquartered in the UK, the Square Kilometer Array is a large multi radio telescope project scheduled for construction in Australia, New Zealand, and South Africa.

In the news
29 Apr 18Author: FutureSight

Located in Australia and South Africa and headquartered in the UK, the Square Kilometer Array is a large multi radio telescope project scheduled for construction in AustraliaNew Zealand, and South Africa. It promises a total collecting area of approximately one square kilometre and will be used for advanced radio astronomy projects and to solve some of the biggest questions in the field of astronomy. It will enable scientists to look far back in time, and see more recent events at a much higher resolution than ever before.

This project will push the limits of high-performance computing and management of huge data sets.

The SKA will process information generated by 350 dishes and 256,000 dipole antennae on two continents. It will operate globally, coordinating across a large radio telescope array in Australia, elements in the UK and New Zealand and eight African nations.

The SKA will need compute greater than the fastest computer currently in existence. It will drive innovation and change in high-performance computing (HPC) that will alter the technology landscape.

A key challenge will be process the data coming in from these widespread telescopes, converting this into usable pictures of the universe by reconciling each receiver’s signals. To achieve this SKA needs clock stabilities in picoseconds (10 to the negative 12) and processing power up to 100 petaFLOPS, filling a 6 terabyte drive every few seconds.

The risks and benefits in SKA are extreme:

  • This project will teach us new things about HPC and Exascale computing, advancing our understanding when it comes to managing data sets at huge scale across globally distributed facilities
  • To enable SKA the project will likely need to employ artificial intelligence-based predictive maintenance at the very least and other AI elements to process and make sense of the vast inputs
  • The two primary astronomical sites in South Africa and Australia will need the most balanced, energy efficient compute in the world, with intensive requirements in power and cooling in a highly challenging desert environment.

Read more about meeting the compute and real-world challenges of the SKA here.

The future of flash is here

29 Mar 18

AI for planes, trains and automobiles!

15 Mar 18

Taming Big Data with AI

03 Apr 18

Will we stay in control as AI evolves?