Initial Design Work Completed On SKA Radio Telescope Supercomputer

Engineers have completed design work on one of the two supercomputers planned to power the world’s biggest public science data project – the Square Kilometre Array (SKA), a radio telescope collecting information from thousands of antennas spread across the world.

The University of Cambridge led five years of design work that brought together nearly 40 institutions in 11 countries, according to the SKA’s Science Data Processor (SDP) consortium.

Engineers created plans for one of the two massive supercomputers intended to process the data generated by the SKA’s telescopes.

The supercomputer, called SKA-low, is to be based in remote Western Australia, and will be joined by a unit called SKA-mid in Cape Town, South Africa.

Compute power

Together the two units form the SDP, which the consortium estimates will have a compute power of around 250 petaflops, some 25 percent faster than IBM’s Summit, currently the world’s fastest supercomputer.

Maurizio Miccolis, SDP’s project manager for the SKA Organisation, said the SDP would process and distribute some 600 petabytes of data each year, enough to fill more than one million average laptops.

Construction of SKA’s first phase is set to begin this year.

SDP is the second stage of processing data from the telescope’s receivers, after the raw data is processed by the SKA’s Central Signal Processor (CSP), comprised of hardware and software that converts astronomical signals into data.

The SDP is expected to process about 5 terabytes per second of information, requiring it to make near real-time decisions about what to retain for processing and what to discard as noise.

Data to information

“SDP is where data becomes information,” said Rosie Bolton, data centre scientist for the SKA Organisation.  “This is where we start making sense of the data and produce detailed astronomical images of the sky.”

Miccolis said the project was developing new software and hardware for its needs, which would create opportunities for applications in other fields.

After it’s processed by the SDP the telescope’s data is to be sent to regional computing centres around the world where it can be accessed by scientists.

Bolton said the SDP architecture must be able to adapt over the 50-year lifetime of the SKA project, meaning it must be highly scalable and extensible.

The telescope is expected to be able to detect very faint radio signals emitted by cosmic sources in the universe’s first billion years of existence.