Pawsey said in a statement on Thursday that the new compute cloud was part of the $70 million capital refresh project to improve the rate of scientific discovery.
On 14 November, Pawsey invited tenders for a new system to replace existing Magnus and Galaxy supercomputers.
Pawsey is a joint venture of CSIRO, Curtin University, Edith Cowan University, Murdoch University and the University of Western Australia and provides services in supercomputing, data management and analysis, and visualisation.
Acting head of Data, Mark Gray, said: "You can cluster containers, maybe you need to spin up 10 machines, to database services, a web server, five computational nodes, and get them all talking to each other and other HPC facilities at Pawsey.
Pawsey data workflow. Supplied
"With this expansion, you will be able to do it, and automate it – this is a system where researchers can run their applications wherever they want and whenever they need."
The new system is built on Dell EMC Power Edge servers and has 58 computer nodes using second-generation AMD EPYC processors supporting up to 14,800 virtual cores, 9 petabytes of Ceph storage, 58 terabytes of RAM (up to 8GB per core) and 100GB ethernet networking.
“In advancing human progress researchers are often only limited by advances in technology,” said Andrew Underwood, field chief technology officer, HPC and Artificial Intelligence, Dell Technologies, Asia Pacific and Japan.
“We have created a technology solution that will deliver significantly higher computing power, in a flexible and modular design, allowing researchers at Pawsey to push the limits of compute and data-intensive research workloads and delivery of faster research breakthroughs.”
The new cloud will be tested during the first quarter of 2020.