The latest upgrade of The University of Queensland’s main supercomputer, featuring the first GPU technology deployment of its nature in the Australian research sector, is due to go live this July.
Phase 4.0 of high-performance computer Bunya includes only the third deployment of AMD’s Instinct MI355X GPU technology in a university worldwide.
Bunya’s expansion combines some of the world’s most powerful GPUs, mixed with the latest CPUs, in a solution designed for efficiency and sustainability.
Implementation of Bunya Phase 4.0 is almost complete, meaning the next-generation technology will likely be available for research use in July 2026.
Once online, the new infrastructure will be available to all UQ researchers with a Bunya HPC account. Visit RCC’s Bunya webpage to learn how to get an account on the supercomputer.
Research Computing Centre (RCC) Director Jake Carroll said: “Bunya Phase 4.0 is the most significant supercomputing investment UQ has made, both in terms of the leap in technology from year to year and in terms of the total amount of computing capability in the phase.”
Care has been taken to balance the needs of simulation, data processing, exploratory work, machine learning, and the newly popular field of generative artificial intelligence.
RCC in conjunction with its delivery partner, Hewlett-Packard Australia, have installed all the hardware UQ planned last year to receive for Bunya Phase 4.0 (see a list of technical specifications in RCC’s December 2025 article about the phase).
The delay to deployment of Bunya Phase 4.0, originally expected last December, was due to unprecedented global demand for HPC components driven by rapidly increasing AI usage.
“RCC needed to work closely with its strategic technology supply partners to navigate this demand crisis in order to deliver cutting-edge systems in this complex global market,” said Jake.
Bunya Phase 4.0 is configured with the latest direct liquid cooling technologies, known as DLC (Direct Liquid to Chip), which provide a more sustainable and efficient way to cool the increasingly energy-intensive AI engines of modern research infrastructure.
DLC technologies reduce the energy required to cool modern chips, improving environmental outcomes.
“Liquid cooling is more efficient in terms of operations per unit of energy consumed. Whilst computing more efficiently, we are also advancing the operating environment with our data centre partner, leading to more sustainable infrastructure, long term,” said Jake.
Bunya is housed at the Polaris Data Centre in Springfield, Queensland, about 30km from the UQ St Lucia campus. Polaris is committed to delivering sustainable and environmentally friendly services, and uses closed loop refrigerative cooling, which recycles water with negligible water consumption, minimising the environmental impact.
“The use of Direct Liquid Cooling in Bunya Phase 4.0 (and following upgrades) allows UQ to scale its deployments in less physical floor space, use less energy on a per FLOP basis and use a new generation of much higher performance AI technologies that are increasingly unavailable to facilities that do not have the ability to use advanced liquid cooling technologies,” said Jake.
“It saves UQ money in energy consumption for the performance it delivers; recaptures heat more efficiently and gives UQ a performance advantage for its most challenging AI and HPC accelerated workloads.” *
Meanwhile, procurement for the next upgrade of Bunya is already underway to ensure timely delivery later this year.
RCC will reveal more details about this next upgrade in the coming months once the hardware is confirmed. “But one thing is clear; it will involve even more GPUs to meet UQ’s demand,” said Jake.
*Click on the following links for more information about the direct liquid cooling technologies used in Bunya Phase 4.0:
The AMD Instinct MI355X: Very large GPU capability for the AI-HPC era
By RCC Director Jake Carroll
The arrival of AMD Instinct MI355X accelerators within Bunya Phase 4.0 represents far more than a routine hardware refresh for UQ’s Research Computing Centre.
The MI355X is one of the most capable AI and scientific computing accelerators currently available anywhere in the world, combining 288 GB of ultra-high-bandwidth HBM3E memory with more than five petaflops of FP8 performance and substantially improved double-precision capability for traditional simulation workloads.
This combination is particularly significant for UQ because RCC supports an unusually diverse research community spanning artificial intelligence, computational and structural biology, engineering, climate science, astronomy, materials science, and broadly data-intensive research.
The exceptionally large memory capacity of the MI355X allows many modern foundation models, large-scale simulations, and data analysis workflows to execute on fewer GPUs, reducing complexity and making advanced computing more accessible to researchers who may not be specialists in distributed AI systems.
At the same time, the accelerators’ strong FP64 and HPC performance ensures that established scientific workloads continue to benefit alongside emerging generative AI applications.
For UQ researchers, the significance of the MI355X lies in enabling larger models, faster experimentation, shorter time-to-discovery, and new forms of interdisciplinary research that combine simulation, data analytics and artificial intelligence within a single computational platform.
As Bunya Phase 4.0 comes online, these accelerators position RCC to support both the next generation of AI-driven research and the computational science challenges that will define the coming decade.
