HPE to Build New Pawsey Supercomputer at Pawsey Supercomputing Center

by November 14, 2020 0 comments

HPE to deliver Pawsey’s new AU$48m research supercomputer

Last month, Pawsey Supercomputing Center selected Hewlett Packard Enterprise (HPE) to deliver its new supercomputer as part of the biggest upgrade to the Pawsey computing infrastructure, which opened in 2009. As a result, HPE was awarded AU$48 million contract for the same, following a thorough tender process led by Australia’s national science agency, CSIRO (Commonwealth Scientific and Industrial Research Organization), the center agent for Pawsey, based on energy efficiency, cost, and HPE’s integrated hardware and software solution.

This supercomputer is part of AU$70m Pawsey’s Capital Refresh Program funded by the Australian government to invest in next-generation supercomputing to advance the nation’s research. It is expected that this new system will deliver up to 50 petaFLOPs, which is at least ten times more power-efficient than its predecessors, Magnus and Galaxy while providing a 30-fold increase in raw computer power. Magnus, a Cray XC40, was commissioned in 2014 and the Galaxy, a Cray XC30 in 2013, and both were rated at 227 and 1,486 peak teraflops, respectively. Currently supporting the operational requirements of several Australian radio telescopes, Magnus is regarded as one of the most advanced supercomputers in the southern hemisphere. Magnus and Galaxy currently occupy 11 cabinets, whereas the new system will occupy eight.

Most of us are aware that today’s supercomputers are used for a wide range of computationally intensive tasks. These include weather forecasting, the drug discovery process, nuclear warhead simulations and more. This supercomputer, will enable researchers to perform complex modeling and simulation tasks in areas such as discovering new galaxies to developing improved diagnostic tests for coronaviruses and developing AI-enabled ways to help farmers reduce herbicide use.

Pawsey’s new supercomputer will be built using the HPE Cray EX architecture, featuring comparatively increased compute power and with focus on accelerators with future-generation AMD EPYC CPUs and AMD Instinct GPUs. It will also integrate the Cray ClusterStor E1000 system, which uses tailored software and hardware features to meet expanded high-performance storage needs along with expanded data storage capabilities. The Cray EX will boast more than 200,000 cores across 1,600 nodes, over 750 GPUs, and more than 200,000 AMD CPU cores, all backed by 548 terabytes of memory. The GPUs and CPUs will be connected using the HPE Slingshot interconnect for moving large volumes of data. The interconnect will deliver at least 200Gbps into every compute node and 800Gbps into every GPU node.  Supercomputer manufacturer Cray Inc has been part of Hewlett Packard, since September last year after the company bought Cray for around $1.8 billion.

The new system will be delivered in two stages. Phase 1, will be available by Q3 2021, and shall provide a 45% increase in raw compute power in one-fifth of the size compared with the Magnus and Galaxy systems. The full commissioning of the system will occur in the second quarter of 2022. Further, it shall be cooled by a groundwater cooling system specially developed by CSIRO for the supercomputing center, which itself will leverage a 118 kW photovoltaic solar array.

Pawsey is also preparing to transition its existing users over to the new system through a variety of initiatives. The first of these is the Pawsey Supercomputing Centre for Extreme-scale Readiness (PACER) program, which aims to boost researchers’ optimization and workflow readiness by running a grand challenge problem at previously unavailable scales. The Pawsey Supercomputing Centre was declared a Centre of Expertise in High-Performance Computing by the Australian Computer Society in 2010.

Recently, Pawsey’s supercomputers were used by researchers to develop Computational fluid dynamics models and to simulate fish swimming styles, including how they increase speed and change direction, and applied these capabilities in machine learning algorithms to train neural networks on underwater robotics. They were also used to study grapevine leaf-roll viruses (GLRV), which infect grapevines and affect up to 70% of production across Australia.

No Comments so far

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.

Your data will be safe!Your e-mail address will not be published. Also other data will not be shared with third person.