Portrait Andy Pimentel


Always curious about how things work

Ever since his high school days, Andy Pimentel has been fascinated by computer systems. ‘I asked my mathematics teacher to explain to me what was inside my Commodore 64.’ Even today, as a full professor heading the Parallel Computing Systems group at the University of Amsterdam, he is happiest when he understands how a system works and manages to create something new to make it better.

How did you end up in computer science?

‘My fascination for the subject started as a kid, playing computer games on my Commodore 64. Very early on, I was not only programming new stuff, but also got interested in the hardware. What parts was such a computer made of, and how did all of that together enable all of the amazing things it could do? During my studies, one of my first classes was about computer organization, architecture and components. I was immediately hooked.
What I like about computer science is the fact that you are creating something new that didn’t exist before. Especially when programming, there is an instant satisfaction when you can show others what you’ve made. And working in research is the best job for someone like me, who is endlessly curious.’

What is it that you like most about working as an academic?

‘First of all, I like the atmosphere of a university. There are loads of young people around, eager to learn new things and looking at problems from different perspectives. Though I originate from and still work at a general university, in my work I take an engineering approach. Looking at the interaction between hardware and software, I am interested in the question how to make systems faster, more reliable and use less energy.’

What types of topics have you been working on over the years?

‘The common thread in my work is formed by computer systems, and more specifically, embedded systems, from the smallest to the largest of scales. So, I could be working on smartphones, but just as well on ASML machines. In my group, we look at what we call extra-functional system behavior – the question is not so much what the system is doing, but rather how it does that. How much memory is used, and where and when? How much energy is consumed? How fast are processes executed? We build analytical and simulation models to understand and optimize this behavior.
We also work on design space exploration: how many and which processors to use, which network, which memory, where to execute what function, et cetera. The aim of this research is to help designers make the right choices. That is extremely hard, because all in all the possibilities are almost endless. So, we need to devise smart ways to explore this space and evaluate the quality of the choices that can be made. To this end, we develop smart algorithms to efficiently search these vast design spaces, for example based on genetic algorithms, that we enrich with domain specific expertise.’

What types of topics does the group currently work on?

‘We develop general methods to optimize embedded systems in terms of energy usage, sustainability, CO2 footprint, efficiency and efficacy, based on a variety of use cases.
For example, we study edge AI, where we are researching how to bring large neural networks as close as possible to small devices with limited computing power, energy capacity and memory. And we think about how to execute AI algorithms on systems that are not reliable, for example because their battery can deplete or they can lose connection to the network.
A second example is research we conduct together with ASML on what the next generation of hard- and software for their giant lithography machines should look like. The big challenge there is how to automatically generate digital twins of those machines, since they are way too complex to model by hand.
Another aspect that is of importance in computer systems is heat. The smaller chips get, and the more layers are stacked on top of each other, the harder it becomes to dissipate the heat. This determines the lifetime of a chip. By developing smart designs and algorithms, we want to make sure not too much heat is generated at a specific spot, and that parts of the chip get the time to cool down in between operations.’

Why did you decide to join the board of IPN?

‘I have been a director of education at the University of Amsterdam for a long time. By the end of 2019, the time had come for me to step down. Since I learnt a lot about academic administration and I like to help bring the computer science community further, joining the board of IPN felt like a logical next step.’

What is your ambition for IPN?

‘Computer science as a field has grown tremendously, both in size and in importance. The next step is to make IPN reflect that growth, for example by empowering the Special Interest Groups and giving them a more prominent role.
Computer science is in desperate need for more funding for fundamental research. Personally, I am rather worried about the development that, for example, the sector plan funds have resulted in many new staff members who now need to build a career of their own and all start competing for the scarce funds that are available. Many calls are earmarked on specific themes, which diminishes chances for other subdomains. And many calls require substantial contributions from industry, in cash. That is a major problem in our field.
IPN is addressing this issue wherever and whenever it can. We are the right platform to do this, since we represent the entire field. In the coming years, I hope we can make this message resonate within the different governmental and administrative bodies.
In addition, I’d like to improve the contacts between computer science and industry on a more structural level. Ideally, every Special Interest Group would have a dedicated liaison who could act as a single point of contact for a specific field. That is something we are currently working on in the board.
And finally, the time has come to better connect the next generation of computer scientists to this type of managerial and lobby activities that are for the greater good of the entire field. I like to delve into this topic and find solutions to achieve a continuous and gradual transition from one generation to fresh and diverse groups of enthusiasts managing the field.’