High performance computing (HPC) has been a mainstay of research institutions worldwide
for five decades, turbocharging scientific invention and innovation in numerous fields.
From groundbreaking medical research for new cancer treatments to the search for the first
stars and galaxies, HPC has consistently played an instrumental role in many home-grown
Discoveries.
Yet, as the demand for well-supported HPC resources continues to grow, Australia’s existing
infrastructure is significantly constraining researchers. Serious issues that are impeding
progress include hardware limitations and reliability, as well as a lack of support with respect
to software development, maintenance and optimisation.
Currently, researchers must compete for time on Australia’s national facilities: the Pawsey
Supercomputing Research Centre and the National Computational Infrastructure.
Unfortunately, requests for time far exceed the time available. Put simply, these facilities are
oversubscribed which is impeding research.
Also, the constant and rapid advance of computing technology, combined with an increasing
diversity of researcher needs, means more and more end-users either lack or cannot access
the expertise required to effectively leverage HPC. This is being compounded by the rise of
computationally hungry artificial intelligence (AI), which is increasingly becoming a standard
tool in research. Many early-career researchers, and those unfamiliar with the niche skillset
required to use HPC effectively, are disadvantaged.
In addition, researchers often face prolonged wait times from application to allocation. This is
another impediment to the timely publication of results in a fast-paced world of research
Races.
When researchers cannot access the HPC resources they need, the flow-on effects are
extensive, including a much longer runway for commercialisation or simply missed
opportunities for innovation.
To address these challenges, the model which governs how HPC resources are allocated
could be refined. Providing researchers with more autonomy and flexibility to select their
HPC provider from a group that includes certified commercial facilities would ensure they get
the fit-for-purpose resources they need, when they need them.
Globally, precedents have been set with respect to HPC funding for both national and
commercial providers. For instance, the UK Meteorological Office announced a move to the
cloud with a £1.2 billion investment to capitalise on the power of commercial HPC.
As the proliferation of AI intensifies the demand for agile HPC solutions is more crucial than
ever. Commercial HPC providers procure hardware just in time, bypassing the limitations of
multi-year cycles. This approach ensures that researchers receive timely access to the latest
Resources.
Moreover, commercial HPC companies can prioritise tailored solutions, with HPC experts
dedicated to addressing each researcher’s distinct needs and challenges, empowering them
to focus solely on their scientific endeavours.
The Albanese Government’s agreement to all 10 recommendations of the Review of the
Australian Research Council Act 2001 is welcome. It’s also encouraging to see that Federal
Education Minister Jason Clare is likewise focussed on strengthening Australia’s research
landscape, ensuring it remains responsive to contemporary challenges.
The HPC sector is looking forward to the review and is committed to collaborating with the
Government to support scientific advancement in diverse sectors such as health, defence,
space exploration, life sciences, and environmental research.
Written by Stuart Strickland, Chief Operating Officer, DUG Technology