25/06/2021

Licensing Consultant

Not just any technology

Pawsey brings new $2m GPU cluster online – Hardware

MWA antennas. Resource: Pawsey Supercomputing Centre

The Pawsey Supercomputing Centre has set reside a new $2 million cluster of graphics processing units (GPUs) to ability analysis working with the Murchison Widefield Array (MWA) telescope.

The cluster was procured from HPE before this calendar year as element of Pawsey’s $70 million money refresh job funded by the federal governing administration.

Dubbed Garrawarla, that means ‘spider’ in the language of the Wajarri folks whose land the telescope sits on, the committed seventy eight-node cluster procedures the petabytes of details streamed from the MWA.

GPUs have been selected to provide further ability to the astronomy analysis that makes use of compute-intensive artificial intelligence and equipment mastering workflows for details analytics.

Researchers who had early obtain to the cluster when it was in pre-production claimed software package undertaking two to eight times more rapidly than on its predecessor.

MWA director, Professor Melanie Johnston-Hollitt, said scientists from twenty organisations primarily based in 5 unique nations are benefitting from the Garrawarla installation.

“The MWA collects a petabyte of details a month, and we have now archived around thirty petabytes at Pawsey,” Johnston-Hollit said.

“Having a committed program suggests that we can perform with Pawsey to high-quality-tune the program and maximise the performance of our workflows.

“Garrawarla and Pawsey’s knowledge will speed up our investigation and processing of this details, helping us to reveal a lot more about the origins and composition of the universe.”

Pawsey workers are at the moment setting up the software package surroundings to help a lot more MWA analysis to migrate workflows on to Garrawarla.

ASKAP

Pawsey has also signed a different $one million deal with HPE to procure ingest nodes to help scientists working with CSIRO’s Australian Square Kilometre Array Pathfinder (ASKAP) telescope.

The supercomputing centre described the ingest nodes as “one of the most significant components” of the pipeline between the telescope and the Pawsey details retail store, which homes ASKAP’s closing details goods.

The nodes acquire the details in actual time from the correlators at the Murchison Radio-astronomy Observatory (MRO) and creating the details to disk to be processed on Pawsey’s Galaxy supercomputer.

The new ingest nodes will element the most recent AMD processors, offering scientists twice as a great deal bandwidth to and from the CPU as the latest technology of nodes, as nicely as a lot more memory channels.

Pawsey’s executive director Mark Stickells said the new infrastructure will help the centre’s methods keep up with the deluge of details that are expected to be developed when ASKAP begins total sky surveys in the coming months.

“The ASKAP ingest nodes will maximize the potential to transfer details from the MRO into Pawsey’s storage methods, and will help to unlock the worth of the astronomical details produced by the telescope,” he said.