When a university invests in supercomputers it’s often expected to be the domain of hard STEM research. This was closer to the case when The University of Queensland (UQ) added a new high-performance computer (HPC) to its Research Computing Centre in 2017.

In fact, when UQ and Dell Technologies built the Wiener supercomputer, it was primarily designed for imaging-intensive science including supporting the University’s Lattice Light Sheet Microscope.

However, last year, when the University commissioned Dell Technologies for a new HPC, Bunya, which replaced three older supercomputers, the brief had evolved. This time, the vision was a HPC built for a wider range of research domains, from the sciences to the humanities.

Introducing supercomputing for researchers

The shift is emblematic of the changing applications of supercomputing in research environments, the company shared in a statement.

The payoff for making eResearch more accessible has been a wide range of projects encompassing but not limited to neuroscience, behavioural psychology, econometrics, languages, literature, hypersonics, materials science, genomics, molecular dynamics for drug design, high energy physics, quantum simulation and artificial intelligence research.

These days, supercomputing processing is used for everything from the development of biofuels to analysing national survey data to study the links between work habits and wellbeing.

Accelerating the democratisation has been the explosion in demand for AI, especially large language models (LLMs) which took off with the release of ChatGPT last year, the company states.

Building HPCs for the future

When Wiener came online, neither COVID-19 nor the AI explosion were on the radar. For instance, Bunya began as a traditional central processing unit (CPU)-based supercomputer with several very high-memory capacity nodes for special research use-cases such as bioinformatics pipelines.

Subsequent upgrades have added different node types, including various types of graphics processing unit (GPUs), with each GPU type added optimised for a different group of workloads.

The HPCs also use novel methods of software deployment and management to allow for flexibility and ease of operation for researchers to quickly adapt the HPCs to emerging use cases.

Balancing power and sustainability

Sustainability was also front of mind for UQ as it developed its supercomputing capability. The energy use of each component is factored in, so every kilowatt of power used is accounted for and compared to the research outcomes they generate.

This is not easy to achieve, but it’s doable in part due to the data and telemetry Dell PowerEdge servers can provide, according to the company. By extending upon this, UQ can obtain actionable sustainability intelligence out of the supercomputing platforms.

Next steps

For the next phase of supercomputing, UQ will use advanced GPU technologies to provide a whole new capability of visual compute inside the supercomputer. This will provide a further democratisation, improving usability for researchers who traditionally don’t work a terminal window and source code.

Jake Carroll, CTO of The University of Queensland’s Research Computing Centre, comments, “For us, it is hard to know exactly what path a supercomputer will take. An important part of being a Tier-2 supercomputing facility in Australia is that we must move (rapidly) with the interests of our researchers as their focus changes and evolves with each phase of their research. We have months to do this. Not years."

"Ultimately, if you build it sensibly, and you are considered in your architecture, it can be flexible and malleable. Our researchers will find ways to best use the technology we provide for them as long as we’ve given them that flexible platform to work with."

Andrew Underwood APJ CTO at Dell Technologies, says, “Collaborating with The University of Queensland to build supercomputers has been immensely satisfying as it gives us the opportunity to see what the parallel processing power of our Dell PowerEdge platforms can unlock in a range of disciplines, from science and medicine to literature and the arts."

"The velocity at which supercomputing uses emerge in the AI-era are accelerating and Dell Technologies is proud to provide our customers with the support they need to excel in it.”