Tim Shanahan 

Stepping into the role of Chairman in 2021, Tim Shanahan has helped steer the organisation as it takes on an expanded role in Australian science with the help of Setonix. He speaks about the role high-performance computing plays in developing Australia’s skills and capacity to manage advanced science. 

"I've really enjoyed my time as chair of Pawsey. It's been very satisfying and very exciting, and in many ways, I've been able to build on the work of others who have come before me. The commissioning of the Setonix machine, which is really exciting and has increased the capability and capacity of the centre, is really a watermark event for the Pawsey supercomputing centre.  

And of course, that work started many years before I became chair. I think a facility, like Pawsey, creates a halo. It creates a halo of reputation. It creates a halo of attraction for people who want to use the facilities. So, you not only get very high-end experts on computing and data, but you also get a lot of people in various disciplines who want to be able to use the equipment.  

So it does act as a very significant attractor of talent, not just here in Perth, but nationally and internationally as well. And I think that's really illustrated by the role that the Pawsey supercomputing centre plays with the Square Kilometre Array, where that is attracting scientists, researchers and technical people from all over the world.  

So, there's no doubt in my mind that there's this very significant halo that comes from having the supercomputing centre based here in Western Australia. 

 No doubt that the Pawsey supercomputing centre is really the product of a lot of vision and a lot of hard work and a lot of partnerships between a range of different governments and organisations. So, there was alignment between the state and federal government, as well as the four public universities and the CSI road to make the Pawsey supercomputer what it is today.  

And that's not just a one-off effort. That's been a continuous effort over the last few decades, for that to have occurred, and there's a range of investments that we're expecting that will be brought to bear over the next several years that will both support Pawsey’s compute capacity, its data capacity, but also improve its energy efficiency, which is so important in contemporary society.  

At the same time, I think we'll consolidate our role as a foundational element of the Square Kilometre Array, and that as that is now being constructed in the Murchison, at least the Australian part of it.  

The role that Pawsey plays in that will become increasingly important at the same time, the new machine that we have, and what we see as the projected expansion of that machine will allow us to broaden the use cases that we're currently working on. 

It will include more biomedical, human health, and climate modelling, areas that we're already active in, but I think that will become mainstream, for want of a better phrase. So Pawsey is a national facility. It's based in Western Australia.  

We have national and international collaborations, and I see, I really see that increasing over the next decade, that we'll have more of a role on a national stage than maybe we have had in the past. And I suppose the work that I have and the work of the team at Pawsey is to have that role recognised as a national and international role. 

 I think there's a range of opportunities that we have across the next decade that will really build on the work of those that have gone before us and really put Pawsey and keep Pawsey at the centre of the national stage for high-performance computing and data."

Amanda Barnard 

One of Australia's most highly awarded computational scientists, Amanda Barnard has worked closely with Pawsey in several roles and is a strategic advisor to the board. Here, she looks at the possibilities of leveraging Pawsey’s computational power and data centre for Australia’s adoption of AI.   

"In 2020, just at the beginning of COVID, Pawsey started up the Pacer scheme, which was investing more in human capital than just the infrastructure.  

It was a unique program, and I joined the steering committee. We evaluated things and gave out some wonderful grants and cash to support researchers, not just research, which is what makes that scheme so interesting. It up-skills scientists to become more ready to use new infrastructure when it becomes available.  

The pace of change in the infrastructure has outstripped our ability to keep up naturally. The way we would naturally keep up before was a generation of PhD students would do the work. And I mean, I've got there's problems with that model as well, because they go off and leave, and then you have got to start all over again. 

You're left with legacy code, which is slowly keeping pace with technology. We've lost that technology is moving too fast for the normal cadence of research to adapt, so there needs to be something to arrest that situation.  

And this is what Pacer did. It was a circuit breaker that came in and said, We need a dedicated effort. It's not going to have to happen naturally and organically, that researchers will keep pace with research infrastructure.  

So we need to make a deliberate effort to upskill them and prepare them for what infrastructure we know is on the horizon. Research is being embedded in their research.  

Having big facilities like Pawsey supporting not just the research we do now, but also forward thinking to the research we'll do in the future is so important. So, Pacer invested in the people, the people who can get up and walk away from Australia if we don't support them; supercomputers don't get up and walk away. There's definitely an opportunity.  

Firstly, these large AI language models, for example, that can help write code are very good at system one type thinking. So, they're very reactive.  

They can automate low-level processes, which can help with doing the kind of thing that I did 10 years ago in automating my workflows, where I had to write all my own code to be able to do that, where it would monitor itself, check if it had completed properly, and rerun if it needed to.  

AI can take care of that for all the researchers, and they can have the capability that I had very easily, but there's a risk associated with that.  

There's a risk that becoming too dependent on AI to automate our workflows means that if there are errors or problems in there where it's much more difficult to identify them, and it puts a different emphasis on the role of the scientist, which not everybody is ready for or can accommodate this sort of supervisory role. 

I would say that HPC workflows automation is probably the low-hanging fruit, and I'd be surprised if most researchers are not doing that now, having support from AI to write the kind of code that will enable them to do more with less.  

However, there's the future. The expectation in the future is that AI will be a bigger participant in our scientific workflows with actual hypothesis testing. Our hypothesis generation as well as helping to analyse our results. This is probably an area which would be referred to now as AI for science, and there's a lot of focus on this around the world, but we're not yet at the stage where I think that we can rely on the AIS for more than being in a supporting role.  

From the infrastructure point of view, there's an opportunity to have an AI, develop an AI that becomes an embedded systems monitor, but also think about load balancing more effectively, so that when users log their jobs to the queue, it can intelligently decide how to distribute them and then when to run them in.  

Researchers always want their results right now, but most of them, when they go to bed at night, want them when they wake up in the morning. 

So, using the cadence of behaviour of researchers on the system. We could intelligently think about when we run different kinds of workflows on different parts of a heterogeneous number of processes, whether it's a GPU or a CPU related job, or, in the future, even more sophisticated processing units. 

 So this is where we move away from that system one thinking to more of a systems two thinking, where instead of it just being an AI advantage, it can do it faster than we can, to an AI supremacy, it's actually doing something we can't do.  

Pawsey, in particular, is very internationally focused and internationally connected. So, in a way, it's more of our international centre than our National Centre.  

I see the intrinsic value of supercomputing to Australian science, not our ability to digitally replicate what we can do in the lab or verify what we can do in the lab. The real value comes in doing all the things we can't do in the lab. Work on dangerous things.  

We can work on things that are in extreme environments. We can work on things with a level of precision and perfection that's impossible to generate in the real world. We can do different kinds of environments, like zero gravity.  

We can do all kinds of different scenarios that are completely impossible for our experimental colleagues. So that's the true value of this scientific instrument. It's not just high-performance computing, it's high-performance science." 

Sarah Pearce 

As the SKA-Low Telescope Director at the SKA Observatory in Australia, Sarah Pearce has seen the Pawsey Centre evolve in parallel to the world’s biggest multinational science project. Here, she describes the importance of supercomputing in helping the SKA achieve its potential.   

"I was first involved with Pawsey when I was director of CSIRO space and astronomy. So Pawsey was at its early stages of being the supercomputing centre for the ASCAP project, which was just in construction at the time.  

This was one of the first times, I think, that a new radio telescope had been born at the same time as a new Supercomputing Centre, which was needed in order to be able to process the data from the telescope. SKA has always been kind of fundamental part of Pawsey; both of the host countries for SKA, both Australia and South Africa, need supercomputing centres. Pawsey sort of fulfills that requirement for Australia.  

So it's been absolutely critical for us that we have a functioning and strong partner supercomputing centre in order to be able to live deliver SKA and SK-Low in Australia. And we're a couple of years into construction now. We have built most of the infrastructure that we're building on site, and we're kind of starting on our way to installing what will in the end be 131,000 antennas. So we've got about 3000 of those antennas started now, but perhaps most importantly for us, we've got 1000 of them working as a small first version of the telescope.  

A couple of months ago, we released the first image from that first very small version of SKAO. 

I think most people, in particular, the first time they go to the centre, what we call the core of the telescope, which is an area of nearly a kilometre by a kilometre, which will, you know, when we finished, be sort of full of antennas, and they see the scale of the kind of land there. 

 It's amazing. But the other scale is, you know, the distance between the antennas. So we have, you know, we have, as I say, a few 1000 installed so far, but we're going to start in the next now, sort of six months or so, really spreading out across the 74km of land where we'll be installing antennas on what are called the spiral arms. 

So we've got 131,000 antennas at full scale, and each of those antennas sends a stream of data. And what we need to do is we have to process that data together to make an image, to make a picture, so that all the antennas are operating like one big telescope. And it's the supercomputing facility that allows us to do that.  

We have what's called a correlator, that's the high-speed digital signal processing that kind of adds together all the different antennas, and then we'll also have a supercomputer which does things like calibrate the antennas so work out what's actual signal from the sky and what is maybe interference or the response, for example, and so SKA is really a big data telescope.  

You know, the challenge of deploying the antennas on site is considerable as is operating the telescope, but the real challenge will come with managing the data, the massive data streams that will flow from the telescope, and getting the best science out of them.  

It has also been fantastic to see the vision come to reality, because I've been involved in SKA for nearly 15 years now, and to be on site when the sort of first camp goes down, and then when the first roads and the trenches start to get dug and to really see what people have been working on — some people much longer than me — is really actually happening now.   

It's happening, you know, at such an amazing scale. It's been a really exciting few years."  

John Langoulant 

As former Pawsey Chair John Langoulant recalls, there is both a political and economic imperative to fund high-performance computing in Australia, and Pawsey offered an opportunity for governments to work closely with university partners and CSIRO to advance Australian research.  

"It's like so many things in my non-executive career, I was asked out of the blue by people out of Curtin University whether I would be willing to consider chairing the board of Pawsey. 

 It was at a time when I was developing a portfolio of activities in the non-executive world. And like so many of my other boards, I didn't know much about supercomputing, but I was at least sharp enough to know that it was going to be really important for WA if we could build it or continue to build it.  

So, I gave it a crack, and it turned out to be, you know, one of the more enjoyable tasks that I had over that period of time.  

Astronomy was a large part of what I played around in my other roles, particularly in my roles in Treasury, because that was an emerging area of activity. So, I could see that developing and Pawsey supporting the Square Kilometre Array was critically important.  

And the other major factor, I think, as to why people came knocking on my door was because of my Treasury background, and everybody could see that Pawsey needed money and how we were going to secure the money from the government. 

I'm alert enough to know that that was probably the major attraction of getting me as a chair, because I had those connections back into government, both at the Commonwealth and state level.  

One of the areas which we tried really hard with at Pawsey was to get more engagement out of business, there was this constant engagement with the universities and the universities had quite a mixed track record about how they interact interacted with business, but we needed a person who could get out and actually promote the importance of having a supercomputing capability in Western Australia.  

And then, of course, the next step was to have the capacity in our machine where people could come and access that capacity.  

And that's the sequence of events, that is, getting the money for the new investment, getting that new investment in place, and then being able to sell that capability into the marketplace was a critical series of steps.  

You can take success up to a certain level and accept it up to a certain level. And I think Pawsey was extremely good because we had promoted into both the federal bureaucracy and across key ministers in that government.  

So, this was a Turnbull government of the need to enhance the machine capability of Pawsey and to get into a sequence of investments between us and a news computer NCI, a series of investments which will provide sustained growth in the capacity of supercomputing across Australia, particularly, at the teaching level, at the science and teaching level, because there was a bit of other stuff going on in the community.  

You know, others were investing and seeking to get into the commercial space. We were more into the research and teaching capability space, and we wanted to secure that. And as things often happen when you're well prepared, well, communicated, as we were, along came the end of June one year, and Malcolm Turnbull found he had a bit of money that he needed to spend before he moved into an election campaign.  

And those various forces came together, as they so often do, and we were fortunate from that point of view. And you know, the investment we got from the federal government was so critically important.  

When I left in late November ‘21, we had grown Pawsey to the point where we were seen as being the most progressive, I think, and interesting supercomputing facility in Australia. 

With all modesty, we had, moved past NCI, and people were seeing Pawsey as being the progressive facility which could provide real capability across Australia, a reach into the rest of Australia, including the initiatives we took to expand the capability of the board and also Mark took to expand the capability of the senior executives in Pawsey were two very significant developments that gave us, the foundations for what is now a very successful entity.  

The other extremely important thing was, not only did we get the funding from the Commonwealth, but we were very successful in society. Sustaining the funding from the state government, which was critically important because the funding from the state government gave us operating revenue, and we needed that operating revenue to both develop the capability of the Pawsey executive group and also to support the introduction of the new machines.  

So those things didn't just happen. There was a lot of work put into that, and we were successful in getting all that into play." 

Cathy Foley 

A former Chief Scientist of Australia, as well as Chief Scientist at CSIRO, Cathy Foley was instrumental in supporting the development of the Australian quantum industry and helping articulate the transformative potential of supercomputing. Here, she considers the importance of Pawsey in demonstrating the link between these emerging technologies. 

"I was in CSIRO for a long time, and my engagement with Pawsey was because I'm a physicist. 

Pawsey was set up as sort of the add-on, I suppose, of managing the whole data concept that was coming out of the Square Kilometre Array and the ASCAP radio telescope.  

Big Data is not something that is taken for granted in everyday day. But several years ago, when Pawsey was first set up, it was new and people just didn't have the experience. 

My engagement has been where Brendan Dalton, who was the head of it in CSIRO at the time, asked if we should invest in supporting Pawsey in bringing quantum brilliance in to set up a hybrid quantum high-performance computer together and be a world first.  

It was quite some years ago now that that started, and it's now gone from strength to strength, so we're seeing quantum capability being a major part of Australia's ambition based on amazing research, but it's also the role of Pawsey in really taking on that initiative.  

And it was a small investment, but it was a critical one, where there was high risk, but I think we are already seeing a high return.  So, the whole concept of, ‘do you have your own HPC computer or not’, is always one of the biggest challenges.  

And I think Pawsey has actually created a lot of capability in the people who've been running it. The idea of constant upgrades, how to manage these big computer systems which are now becoming part of everyday life, servers and the whole sort of digital world that we live in now, is something which we all take for granted.  

But being able to have some foundational sovereign capability there is important, because they're not cheap. I mean, you've got one of the top high-performance computers in the world, and then, a couple months later, it goes down the list, so it's a constant investment, but it's almost like anything of infrastructure for any modern society is that the infrastructure changes and constantly has to be upgraded and improved. The importance of Pawsey has been a foundational aspect to this.  

I think it's having partners managed by CSIRO. It's a collaboration of different universities and organisations coming together, but it's got a level of prominence which has meant that it's almost like a showcase in some ways. 

And I think, you know, the role Mark Stickells has played, for example, as the director, has been pivotal and really getting out there and making sure people understand why it's there.  

And it isn't just something which is a black box hidden away behind closed doors that people don't know anything about. Australia's got an amazing track record in research. Our challenges are that it is more than translation of research and getting the system linked up so that we can pass things down to add to sort of having a knowledge-based economy.  

We need to turn that into new industries for Australia. And I think you could say we're very efficient with our research dollars and the infrastructure that we have, but maintaining it and keeping it up to scratch needs to be a priority for the country if we're going to use that research and move it its way through, so that we do have the not just new knowledge, which is always good to have. 

Humanity needs to understand the world we're in and ourselves, but also then turning those smarts into something which will lead to economic prosperity is our next challenge."

Megan Clark

As the former director of the CSIRO and former head of the Australian Space Agency Megan Clark remembers the birth of the Pawsey centre as an opportunity to create an innovative, sustainable supercomputer centre cooled by a unique groundwater system.  

"My first interaction with Pawsey was back in April 2009 when Pawsey was IVEC, and yet it was really important that we had a supercomputer facility to support the Square Kilometre Array. At the time, I was head of CSIRO. So, this was a major project for us. We sought the $80 million funding for that from the federal government. I remember that as a personal conversation with the Prime Minister at the time, and I said: ‘Well, you know, you'd better take an announcement in with you, like that would be a good place to start’. And, I said: "You know, this is how we're going to use this.” 

And he did. So that was fantastic.  And then, as CEO of CSIRO, we looked at the cost, the annual electricity cost of cooling the computer, which was around a million dollars.  

And while we'd got the capital for the project, I didn't have a million dollars of operating funding spare to be able to fund this. But I was born in Western Australia. I was very aware of the hot water, very aware of the original sort of hot water that you see under, underneath the old sugar plant, the victim bars with the hot water.  

And I can remember contemplating, how are we going to find this? And I was at Perry Lakes, and there's a hot water upwelling that comes out of Perry Lakes.  

I thought, we've got to use this.  

It hadn't been done. So, this was quite innovative to do this, but we had this unique capacity. So we then sought some funding for that. The early design of this was, we were going to go very deep into the hot water, to bring that up, use heat exchanges and but the team said.  

Actually, technology, moved on a little bit, and we think we can do this with ambient temperature, maybe sort of 20C and above.  

We then went with a simpler project design, which is beautifully described by Thomas Pula in a conference paper.  

My third personal interaction with Pawsey was through my role in the Australian space agency. We had just started the space agency and we had a space infrastructure fund, and we had a series of six priorities in which to spend that.  

Our intention also was to make sure that it was spent in the states and territories, so that it was well distributed in the right place for that infrastructure. It wasn't a big grant, but it was definitely to support earth observation and the data we might need for space situational awareness, and really to support Pawsey as a critical piece of infrastructure for the space sector.  

So, they're my very personal connections, and I suppose I've been fighting for Pawsey for quite a while and trying to make sure that it continues to be supported. It's absolutely vital that if we are to participate in these data-rich sectors, whether it's space, whether it's supercomputing for industry, whether it's even just the combination of quantum and supercomputing, which I think is an underrated but very important combination that will be visible in the future. 

Pawsey, and the teams at Pawsey, and the quality of the people that we've had at Pawsey all the way through, since my engagement in 2009, have been such a hallmark. And the dedication and doing lots with not very much, you know, just a fantastic team."

George Beckett 

Computational mathematician George Beckett was seconded from the EPCC supercomputing centre in Edinburgh to spend three years advancing Pawsey’s evolution as Deputy Director and Head of Supercomputing. He speaks here about the work needed to bring Magnus and Epic online.   

"About mid to late 2011, Andrew Rohl, who was the director of IVEC, contacted several groups around the world, looking for technical staff to help them with the supercomputing expansion, which was the Pawsey project.  

And I worked at the Centre for three years, which saw the introduction of the Magnus supercomputer, the establishment of a strong relationship with the radio astronomy community, and the transformation of IVEC into Pawsey.  

It was a dynamic team. Some people who had worked in IVEC who continued to be part of the team.  

 It was very interesting to see how different groups and different centres had run their services in the past, and how that might reflect on what Pawsey was trying to do.  

Pawsey had a pretty good setup already. They were expanding rapidly and having to deal with many more users, potentially more challenging computational science, potentially more demanding International, or certainly national science community, and my team brought our experience in managing expectations, and prioritising the key elements of the service, empowering scientists to do pretty ambitious things.  

So the radio astronomy community was already well established in Australia. It's a world leader in radio astronomy, but there were other key science communities in the country and, in fact, in Western Australia.  

So one of our key jobs was to show them the new opportunity they would get with a supercomputer, a technology they hadn't had access to before, which allowed them to do science in different ways than how they'd worked before.  

It was a computationally literate community, but what they were lacking was the capacity in terms of having a significant resource like Magnus to pursue their ambitions.  

So when I arrived, they already had two precursor systems. One was called Epic, which was a cluster installed in a shipping crate out at Murdoch University. And the other was called Fornax, which was a GPU system installed at the data centre in UWA. 

Fornax was a bit more unusual. It was based around GPU technologies, which were reasonably mature, but still most of the community were unfamiliar with how to use them, and not just the Australian community, the international community. 

Epic, on the other hand, was a workhorse machine, and it did a great job. It was a very busy system. It worked very hard and produced a lot of science. And Magnus was probably an order of magnitude larger than epic. It had newer processors. It had faster networking and faster storage. But a lot of the things you would recognise from Epic also worked on Magnus, and so it was a reasonably straightforward transition.  

And when I arrived, there was a discussion about what Magnus would look like, and there were two or three different options. One of them was to have a system which would be heavily based on GPUs. Potentially, that would have been a faster system on paper but the chances are that scientists wouldn't be able to use it as effectively, and so a lot of that additional performance would have, in fact, never been realised.  

But the more recent system, Setonix, which has replaced Magnus, is quite heavily GPU-based, or accelerator-based, and I believe that has been very successful, and it feels like it's come at the right time as the community's matured.  

The trouble with working in the computing, supercomputing centre, is it takes so long to build new systems that you're always looking at the next system, even while you're running the current one.  

So, one of the challenges in communicating with sort of policymakers or funders is the use of AI, or the importance of AI. And it's very easy to take an extreme view one way or the other.  

I think what will happen is that a lot of the traditional HPC users will incorporate AI into their workflows where it gives them an advantage or where it gives them additional capability. But what we also see is lots of new communities coming on board, communities that would have never considered using supercomputing as part of their science. And for those new communities, a little bit like Pawsey back in 2012, it will be quite alien to them but potentially something exciting. 

What I really liked about Pawsey is the fact that there were five different institutions involved, the five universities, as well as the sort of national angle. And I really thought that it was, it was a great example of seeing five universities work together fairly closely and in a sort of fairly joined-up fashion.  

And I think it was actually the combination of those universities that meant that Pawsey was a success. And it meant that I got to travel up and down the Kwinana Freeway and various things like that, periodically, to go and visit different sites, and got to see much more diverse science and sort of research as a result of that. So I really enjoyed that aspect."

Rebecca Hartman-Baker 

Rebecca Hartman-Baker travelled from the US to join the Pawsey team, providing specialist advice on how to set up and run a supercomputing centre at scale. Here, she recalls her journey to Pawsey and the many changes she has seen in HPC over the years.   

"I started my career at Oak Ridge National Laboratory in Oak Ridge, Tennessee. And one day, we were asked to host some people from this place, Western Australia. Never heard of it, really. And from this place called IVEC.  

They were looking at our building and our energy efficiency, but they were also looking at how we served our users and what we did for users.  

And I went to their website a few months later, and I saw they had some job openings. And so to me, that was just a really exciting opportunity to go to a completely new place. And, you know, help, kind of build it from the ground up.  

I started in about July 2012. I was a senior supercomputing specialist, and so I was kind of there to help users, if they had projects, to help them with their codes and optimise them and make them more efficient.  

So we did a lot of training, and taught them how to program for a supercomputer, but also how to use tools to optimise their codes and to find bugs and the like.  

We had to get them to think differently about the resources and about what they could do with them. A lot of people were used to running on their workstations, and then, you know, do whatever they want — run a job for two weeks, no big deal.  

But then you're on a shared resource and it is very different.  

When I started, I worked on one of the most powerful supercomputers in the world, which was at the University of Illinois at the time. That's where I did my PhD. And today, that machine, the power, the whole power of that machine, could fit into basically the power envelope of maybe one coffee maker. 

So, at that time, this was a machine that took up hundreds of square meters. Our first petaflop machine at Oak Ridge was 200 cabinets, and then at Pawsey, our petaflop machine was about eight cabinets.  

It is just amazing how much progress there is there and how much growth there has been in such a short time.  

We are now kind of stopped by Moore’s Law, Dennard scaling is paused, because we're down to the atomic level, and you have to have a whole number of atoms, and so there's not very many that you can fit in there anymore. 

So, I think we're going to have to think about new and different ways of computing. And I know quantum computing, for example, is very promising. And I know Pawsey has a quantum accelerator there.  

Originally, when I thought about quantum computers, I was worried that I might be out of a job, but now I don't think so.  

I think that people are going to want to take multiple quantum computers and put them all together, and so they're still going to be a supercomputing centre. It just may not use the same kind of processors. Magnus came in a couple of different stages, and we work with users, and tried to figure out, if they ready for these different architectures.  

At the time, GPUs were kind of a novelty in HPC, and now, I mean, they're just everywhere. Everybody has GPUs, but at that time, we felt like our users probably weren't quite ready for that, for a full system that had GPUs, so that's why we went with a full CPU system. It was a lovely system. It's one of my favourites to this day. I kind of miss it."

Paul Nicholls 

Securing Federal funding for Pawsey was a critical milestone in transforming from a small supercomputer centre to one that would go on to house Australia’s fastest and most powerful machines. Here, former iVEC acting Deputy Director Paul Nicholls recalls the time when Pawsey’s future became clear. 

"I was the Director of Science Education Science Partnerships at SciTech, and the WA Government started involving a thing called the SKA. It required us to demonstrate that we had an active astronomy community in WA.   

So as part of that, I set up Astronomy WA, which was a website and a whole resource of information about astronomy activities. We purchased a few hundred thousand dollars of telescopes for community events.  

We’d rock up to a beach and do stuff at the beach at night and all sorts of things. A sort of guerrilla science communication, which was a bit of a thing at the time. No one ever predicted that there was an appetite for astronomy.  

Part of that was that we were awarded a grant to promote science education. The other part was the Indigenous Land User Agreement, so making sure we were delivering support services in PI, which was the community where Murchison is based.  

We were part of a successful bid to host a component of the SKA in WAC. And when I started digging down into what the SKA was and what a radio astronomy telescope was, it became evident that it's not a typical light telescope that you take out and point at the stars.  

It's basically a lot of passive sensors, network, more data flow, and then a computer attached to data storage. 

So when I'm sitting inside Curtin and looking for ICT capability, I came across a range of researchers, probably the most prominent was in computational chemistry.  

There were two people in that — one was Andrew Rohl, already at IVEC, and the other was Julian Gayle.  

Julian was recruited as a Premier’s Fellow in computational chemistry, and as part of his Premier's fellowship, they gave him a lot of cash to buy a computer, which ended up in the basement at the ARC Building at CSIRO in Technology Park.  

Andrew Rohl had effectively decided he wanted to come back to Curtin, and they said they'd agreed on the type of computer they wanted.  

They'd started the construction, and he could see light on the horizon, which meant that 18 months down the track, the computer would be operational.  

And he needed those 18 months to get his research career up and going again. So, I was asked if I would be willing to go over. It turned out to be, I think 11 months, my job to try and transition into a national facility.  

So, there's suddenly a lot more scrutiny over why this happened at the time. Everyone thought it was ridiculous. Why the hell would you put $80 million into WA?  It was quite contentious at the time the commitment was made.  

Firstly, because the SKA was experimental. We didn’t know what was going to happen or where it was going to happen. I think they de-risked it by saying that we had national computing equipment in Canberra.  

However, that piece of equipment needed to be refreshed regularly. When that machine was down, we needed to have a machine that was up. I think I left before the actual machines had been commissioned, but they'd been purchased and pretty much installed, and were not ready for use.  So, I invested, I think, $5,000 in the imagery for the computers.  

Did the universities understand the significance? I think they did, and you know, so I was an advocate. I was basically because I had that 12 months experience, they put me straight in as a board member, which meant I knew how the thing worked.  

I knew what the potential was, and I made sure that Curtin got its value for money. Some of the other universities, I think, were too far away from the Deputy Vice Chancellors of Research to recognise that, or didn't have the researcher capability user, and they came back to Curtin.  

You can see the timeline of the things I set up, which were at the end of that. So, there's a range of initiatives and partnerships, and then cooperative research centres and industry engagement initiatives that are all flowing. 

We generated, probably non-SKA, about $200 million worth of initiatives in corporate and research centres.  

One of the things that I feel that we're still weak at is integration of high-performance computing into the talent development pipelines, inside universities.  

It's still seen as something that's only for those really smart kids and the computer scientists, and we'll add it as a bolt-on course, or we'll have it added as a master's degree.  

I think Western Australia has had this opportunity, and we're probably 10 years behind where we should be in terms of training high-performance computing specialists that could be all over the world by now, but aren't.  

Instead, we continue to import that talent. As we look back, you can see the significant impact this has had on the state far beyond high-performance computing into a whole range of applied areas.  

The opportunity now is for the federal government to stabilise the funding patterns for Tier-1 high-performance computing and, by doing so, secure Australia as a leader in high technology development and applications in major industries in the region. 

To do that, I think there's also got to be an associated investment by the higher education sector and industry in the next stage of high-performance computing."

Geoffrey Harben 

“It was a wonderful feeling when it all came together.” 

Geoff Harben joined as a Non-Executive Director of iVEC and has served more than 13 years in this role at Pawsey. Here, he speaks about the influence of former Deputy Premier and iVEC Chairman Mal Bryce on the Pawsey project — and the ongoing importance of funding Australia’s Tier 1 supercomputing capacity. 

"Mal Bryce was the chairman of IVEC, and Mal and I had worked together on another board, for the state government around technology. And he rang me and said: ‘Look, I feel there's a need for more emotionally oriented people’. And I said: ‘I'd love to’. 

We didn't have the centre that exists today. All of the supercomputing technology f was in the universities, and these were fairly disparate systems.  

So it was really Mel's vision and a huge amount of work that got us to the early days of building our own significant data centre and commissioning our first ever supercomputing facility. 

Mal was probably Australia’s first Minister for Technology. He was also the Deputy Premier of Western Australia. He was ahead of his time. 

He was also a fascinating guy, a dinky-di Aussie, and one of the world's great raconteurs. He could spin yarns that were just extraordinary.  

But he had a vision, and he had done work all around Australia trying to intercept where he thought technology would go. He had a clear understanding of the value of technology to society, and he got his teeth into driving this whole supercomputing thing.  

Without him, it would not have happened. There's been dialogue in the technology press about how incredibly important technology and technology infrastructure is to globalisation, to stability in the world, and how little thought is given to the fact that technology infrastructure has to get refreshed far quicker than most other infrastructure.  

We also want to move to the next stage of computing technology, Exascale. The federal government's looking at funding this again. 

They have put together a working group with senior people looking across the whole spectrum of computing infrastructure to support the research community, to recommend where to invest.  

Universities are doing fantastic research, including research that affects other countries. 

And they are starting to work with the private sector. We are now also engaging with government agencies, including the health department.  

The West Australian government has introduced both privacy legislation and data sharing legislation. Previously, government departments weren't able to share data. The West Australian government has an enormous wealth of information. We're in a position to be able to help them with that.  

This Setonix system and what comes after is going to be able to go through that way better than anything any government department is possibly going to have, or could ever afford to have.  

I think it's important that there are two tiers. Tier one centres, on either side of the country, can support each other and back each other up. I think the public probably doesn't know just how much use is being made of these facilities, PR and in Canberra, so we should be very proud of it."

Lyn Beazley

“It is seen as something that's owned by the community and in which the community has pride.” 

Professor Lyn Beazley was Chief Scientist of Western Australia from 2006 to 2013 advising the WA Government on science, innovation and technology. She speaks about her time supporting the case for the establishment of the Pawsey Centre.  

"I remember thinking what a wonderful opportunity it was. And Andrew Rohl was such an influential person at that stage, often taking a very quiet and considered approach, but consistent and convincing, that there was a need for Pawsey, although it didn't have a name then, but also that we had the expertise in the state to do it.  

Because I was so involved with the Square Kilometre Array radio telescope discussions, it was clear that Pawsey had to be in WA. I thought from the outset it was positive that it wouldn't be entirely devoted to radio astronomy. 

In fact, I think at the beginning it was a quarter for radio astronomy, with agriculture, the environment, medical and industry all having an involvement. I thought this was a good and balanced approach to take, as it was, playing to our strengths and playing to the need that was very real.  

As Chief Scientist, one of my roles was to make sure everyone was on the same page. And I remember we held events, particularly in the ARC Building, where we brought everyone together to let them have a voice.  

And I think that was an important part of the process, because everyone had to buy in to see this not only as something that was needed by the Square Kilometre Array, but was also an opportunity for everyone. 

It brings enormous value — we see it in the medical research area, where there have been such advances in molecular techniques and genomics, where you need to crunch lots of data. 

But you see that in all the other disciplines too. You see the opportunity for remote sensing, which is influential in understanding climate change, looking at our environment, and driving the patterns of agriculture that take place in the state.  

This goes way beyond radio astronomy, and to have industry working with academe and to use this facility positively has been important.  

A lot of basic and important discoveries that we've made in Western Australia, for instance in geoscience, will have been influenced by data that's been crunched at the Pawsey centre.  

And I think that's also the fact that it has had an open-door policy, exemplified by CEO Mark Stickells and owned by the community. 

That has been an important aspect too, because I always reckon it's four legs on the table, government, industry, academe and community, and I include everything from primary school right through to university at the Third Age.   

I think those four legs on the table have been appreciated by Pawsey, and that has been an important part of its evolution."

Andrew Rohl

A leading expert in the application of supercomputing technologies in materials chemistry and a former Executive Director of iVEC, Andrew Rohl was instrumental in our early years. Here he talks about the partnerships that helped progress the Pawsey vision. 

"I became the iVEC director because I saw the job advertised one day, and I spent a lot of time, you know, as a supercomputer user, so I thought it might be an interesting job.  

When I came over, I was the first director that was full time — it was my kind of full-time role. Around the same time Mal Bryce became the Chair, so it was, you know, having a new board chair and the new director ... everything was pretty much up for grabs.  

Mal was very, very dynamic, and had some quite strong views on the role of ICT in society. He was prosecuting that argument because he'd been doing it all of his life, and then I was working with him trying to, you know, build up a centre. In not having a full-time director, it hadn't really thrived at all.  

When I came along, we were able to attract more funding from the state government that allowed us to have a full-time staff, improve our computing, and then everything dramatically changed. I got a phone call from the Department, which would have been the Department of Industry and Resources then, saying, “oh, Andrew, you need to come to the Budget lock-up meeting tomorrow.”  

I went, "don't be ridiculous. It's a long way from Perth to Canberra to go for the two-hour meeting," went to bed, and next morning, Mal phoned and said, “you know, Andrew, you've got to get on that plane to Canberra.”  

So indeed, I did get on the plane, and at that announcement, was announced the Super Science Initiative. NCI, where I'm at now, was one of the points of investment. I'm in the building that was built through that funding, which is the same funding that built the Pawsey centre.  

There's one centre, and it's based in Canberra. And then there are two centres of equal size on either side of the country. The role of preparing for the Square Kilometre Array cannot be played down, in my opinion, because I think that's why it ended up in Western Australia, rather than somewhere else.  

What took a long time was building Pawsey Centre itself and then putting the computer into the Pawsey Centre. So first of all, we had to do it quickly, because super sites was one of the government's responses to the GFC: go buy a boat that's going to go to Antarctica. Go build two supercomputing centres.  

There was a huge drive at that point to get the buildings built right, because that was how the money could be seen to be stimulating the economy. We worked pretty closely with the Federal Government on the name. They knew they wanted to name it in the radio astronomy space, because I think that, again, helped their bid for siting the SKA in Australia.  

For me, probably the most moving moment I had was when we actually opened the Pawsey Centre. And Joseph Pawsey’s son came to the opening, and his father actually had died, you know, just after he came back to Australia from the war, and before he actually took up the position for doing radio astronomy in Australia. His son was in his 60s at the time. To me that was really moving.  

I think it, to me, it helped emphasise that the purpose was the whole community. And you know, the government were really clear on that as well. And I just thought that that tied that all together quite beautifully when we opened the centre."