What is it? 

High-performance computing (HPC) applies to any technology that provides a fast execution environment. These days, this almost exclusively involves parallelism, although in the past a variety of techniques could deliver high performance. Modern HPC machines (also referred to as supercomputers) are usually built as clusters of commodity hardware, connected with high-speed, low-latency switching technology. 

Currently, it is common to have processor nodes with multiple sockets, each of which can contain a multi-core processor. This means that processor nodes can implement shared memory parallelism, typically at the thread level. Environments such as OpenMP allow a program to exploit this type of concurrency.  

At the node level, parallelism can goa step further with modern accelerator technologies, such as graphics processing units (GPUs), designed to speed up execution even further, with simplified instruction sets, variable precision and application-specific instruction optimisation. 

Larger programs can be distributed across the machine using message passing libraries such as MPI. It is not uncommon to exploit both parallelism forms, and hybrid programs execute multiple threads (typically as unrolled loops) within a node, and send messages between nodes. 


What is happening at UQ? 

UQ researchers have access to a broad range of cutting-edge HPC resources, both locally and nationally.

The Research Computing Centre (RCC) manages central high-performance computing resources at UQ.

Central HPC facilities at UQ currently comprise the Bunya and Wiener supercomputers. Note that UQ HPCs Awoonga, FlashLite and Tinaroo were decommissioned in March 2022, January 2023, and March 2023, respectively, as they had reached end-of-life status. If you were previously using these HPCs, please check our Bunya and Wiener webpages about how to access those HPCs or email the RCC Support Desk for guidance: rcc-support@uq.edu.au.
 

UQ's Advanced Computing Strategy 

Following almost a decade of success and a growing HPC user community at UQ, RCC refined its advanced computing strategy in 2023, focusing on the creation of a heterogenous high-performance computing platform to serve the widest possible range of scientific domains.   

This strategy involved the consolidation of partitioned application space and dedicated niche HPCs into one large platform — Bunya — capable of dealing with many different research domains, including very large tightly-coupled parallel jobs, loosely-coupled high-throughput jobs and accelerated computing.  

In line with the UQ Technology Master Plan, RCC is providing a new range of technology options to facilitate ease of access, lower barriers to entry and smooth delivery of HPC resources to a continually widening HPC user community. 
 

Which HPC should I use? 

Wiener 

Wiener is specifically for imaging-intensive science, generated by UQ’s world-leading microscopy facilities. It provides near real-time outputs of deconvolved, tagged and appropriately characterised imaging data.   

Wiener, launched in early 2018, will support UQ’s new world-class IMB-based Lattice Light Sheet Microscope

Wiener was developed with strategic funding from UQ and a consortium of the University’s various cutting-edge microscopy facilities housed within the Centre for Microscopy and Microanalysis (CMM), Institute for Molecular Bioscience (IMB), and Queensland Brain Institute (QBI). 

Bunya 

Bunya, launched in early 2021, was developed with strategic funding from UQ and was designed to consolidate and aggregate all UQ’s and its partner institutions’ needs under one roof.  

Bunya is the largest and most powerful supercomputer in UQ history. It uses a phased deployment ‘evergreen’ model, putting new tranches of technology into production every year, delivering cutting-edge technologies for researchers in predictable timeframes.  

Bunya is a heterogeneous supercomputer, designed to cater to almost every type of workload UQ researchers may have for it. Bunya was built with a balance of very high throughput, very high floating point, tightly coupled, loosely coupled and accelerated computing principles in mind. 

Bunya also features interactive visual computing facilities via onBunya, providing users who prefer a visual desktop experience a very usable, familiar graphical environment, with all the power and capabilities of a supercomputer.
  

HPC Accounts 

UQ staff and higher-degree research (HDR) students are entitled to free HPC accounts on Bunya or Wiener but need to apply.  

To open an HPC account for research, UQ staff and HDR students should do the following: 

  • Register for a QRIScloud account if they do not already have one (register at QRIScloud).  

  • Request a cluster-specific service, i.e., Bunya or Wiener. See the Bunya and Wiener webpages for further information about getting an account. 

Once your account has been created, your access details for HPC will be confirmed via email. 

For any other support requests, please contact the RCC Support Desk: rcc-support@uq.edu.au
 

Training 

UQ staff and HDR students may want to attend a training session to familiarise themselves with UQ's HPC environment. Training is provided monthly, and researchers are welcome to consult with RCC’s HPC staff about their projects — please email: rcc-support@uq.edu.au
 

Other HPCs that UQ researchers may use 

Very large tightly-coupled parallel jobs that support NCG-funded research 

National Computational Infrastructure (NCI) at the Australian National University (ANU) best serves these jobs. NCI operates a 3,200-node supercomputer called Gadi, that entered production in late 2019. 

UQ researchers may access Gadi via QCIF’s share. See our National Facilities webpage for more information. 
 

Loosely-coupled high-throughput  jobs 

The ARDC Nectar Research Cloud is an ideal platform for running loosely-coupled jobs.  

QCIF operates the Queensland node of the Nectar Research Cloud. The Queensland node is called QRIScloud, and its main data centre is located at the Polaris Data Centre in Springfield, Brisbane.  

See our Cloud Computing webpage for more information.