Faces of HPC: Alan Stokes

Tell us a bit about yourself

I am a post-doctoral scientist working with the SpiNNaker platform at the University of Manchester.

I have a background in information management and artificial intelligence. I currently support the development of the SpiNNaker software stack and supporting end-users in using the platform. My research aim is to utilise SpiNNaker for advancing general computing technology where suited.

What is your current job? Describe what you do in HPC. Is this your main interest, or something you fell into?

The job I’m working on now uses a unique hardware, called SpiNNaker, that's made of up to a million arm processes with a bespoke communication architecture that is very energy efficient. This allows small multicast communications to occur very quickly, enabling the simulation of neural networks, that what can not otherwise be done at the moment.

I came to this job because my previous work on sensor networks was a distributed computation process on systems that consisted of tiny sensors connected wirelessly. When I heard about SpiNNaker I immediately saw parallels to what I had already been doing and thought of it as having the potential to perform as a high performance computing platform on its own. I am here to develop the software and develop it to be useful for computational neuroscientists, but  ‘normal’ HPC problems could be applied to it. I do the HPC side of my work in my spare time, so we are still ironing out the bugs in the ‘HPC’ version of the SpiNNaker software but the HBP application that we are currently testing in beta indicates that SpiNNaker runs the system 6x faster than the supercomputer the system was originally benchmarked on.

That’s how I came into HPC – I came in hoping to make SpiNNaker an HPC machine.

As part of this project we want to celebrate the diversity of HPC, in particular to promote equality across the nine “protected” characteristics of the UK Equality Act, which are replicated in world-wide equality legislation. Do you feel an affiliation with this matter, and if so how has this interacted with or impacted your job in the HPC community?’

Before I was an undergraduate I was diagnosed with what was thought to be a terminal cancer. I spent most of my undergraduate and some of my PhD fighting the cancer . This has resulted in add-on effects on how I operate. For example, my stamina isn’t as good as everyone else in the field. This has resulted in me having a different perspective on life to many other people. I often think worrying about certain things in politics and more generally in life is a waste of time. There are bigger problems.

I was also diagnosed with dyslexia when I was a child. As a software engineer this opens up interesting viewpoints as I see things differently from other people. One of my colleagues came up with an interesting description: most people see software as loads of lines. I see software as blocks and I can move the blocks around and abstract them in my head easier than other people. This gives me a way of refactoring code that is different from other people and is a positive aspect to my dyslexia.

Those two things together mean I approach projects with a unique viewpoint. I take more abstract views and consider different ways of using things. Because of my cancer background I have a more motivated viewpoint in a way: my time is scarce because I don’t know what will happen in the future. I am currently having outpatient treatment for Chronic kidney disease (CKD) and IGA nephropathy which are other side effects from the cancer treatment and together augment the reduction in stamina.  There are plans for a variety of treatments and complications in the future due to these health issues.

Were there any challenges when you entered the field of HPC?

The group I work with don’t consider the SpiNNaker platform to be an HPC machine. This means that when I try and adapt the work to sell as an HPC platform or supercomputer it can result in a backlash. Although the group now see that SpiNNaker can be a general purpose high performance computer, there is still resistance to giving the research time to pushing this rather than using the time for the research projects that use the platform, as was the original purpose of the SpiNNaker work. I usually have to use some of my own time, but it is now getting to a stage where the proof of application will show valid research novelty, which means hopefully we can get funding for this side of the work. Hopefully this means that we can sell the platform as a way of doing HPC with many low-powered processors.

What’s the best thing about working in HPC?

I have spent a long time fighting that I am in HPC. I don’t use many other systems, but my platform has amazing capabilities and the low power aspect means we could have huge impact on reducing the energy impact of HPC applications, including for usage such as weather prediction. I see this as revolutionising a lot of areas that have massive energy demand and reduces the geographical limitations of HPC based on feasibility of receiving power to the location of the supercomputer.

What’s next for you in HPC?

I don’t know the fundamental answer. If everything fell into place I see myself running a department working on SpiNNaker solving HPC problems – that would be my ideal world. Given that I’m a young postdoc and therefore I haven’t been in the academic world for very long, and HPC is a completely different world from where I am currently, I suspect I will be doing  my current job or similar in the next 5 years.

Alan was interviewed by Toni Collis in May 2017.