EPCC @ The Museum

Author: Alistair Grant
Posted: 22 Apr 2014 | 14:20

A visit to the museum is often filled with wonders about the past, other cultures, science and art. For five days in April, a visit to the National Museum of Scotland included an introduction to supercomputers. As part of the University of Edinburgh's programme of family events, EPCC staff and university students were there with a drop-in exhibit: "What Makes Supercomputers Super?". The exhibit had activities which the public, young and old, could interact with and gain insight into how supercomputers support the science being researched at the University of Edinburgh, building on the history displayed by the museum.

We ran three hands-on activities: a ball-sorting algorithm, supercomputing hardware and dinosaur simulator. In addition, we displayed posters about the dinosaur simulation, an illustration of how Moore's Law affects supercomputers, and a brief history of supercomputers at EPCC. Our activities were staffed by members of EPCC and students from our MSc in HPC and the School of Physics & Astronomy. You can read some comments from them below.

Across the event, we had a total of seven people helping to run our activities and answer questions. From EPCC there were Iain Bethune, Eilidh Troup, Nick Brown and myself, and the students were Justs Zarins, Emma Ryan and Kostas Ladopoulos (pictured left).

Ball-sorting

In this activity participants work out a parallel sorting algorithm to sort a bag of coloured balls into boxes. To demonstrate that it is hard to do operations to the same entity at the same time without undoing the work of someone else or losing some data, some restrictions were placed on the operations that could be done. The initial step is for a single person to sort as many balls as they can in a minute. The average person will sort about a third to half of the bag of coloured balls.

For the next step, several people work out how to sort the balls as a team, coming up with an 'algorithm' (even if we didn't call it that to begin with). They quickly realise that they get in each other's way or things get blocked. Most of the participants came up with a good way to solve the problem, and the questions which arose allowed them to link the sorting of coloured balls to other more difficult problems or other real-life situations, for example post code sorting. The success of the parallel sort varied but mostly was more effective in a minute, and across the 5 days a lot of teams managed to sort all of the balls.
 
 
 
 

I did the coloured ball sort and got a good amount of interest. The children got really competitive trying to get the most balls in the time allowed. Some of the older children thought about different algorithms - like taking a colour each to sort and problems like contention as they got in each other's way. Eilidh, EPCC

This activity attracted a real spread of age ranges. Our youngest participant was only two and our oldest was a lot lot older than that (we're too polite to say how old).

The ball sorting activity was probably our busiest, with a lot of groups trying to figure out the best way of doing it. Some would attempt it alone and others were in groups, like in the pictures below.

 

 

 

 

 

 

The high points of the event were when people, especially younger kids, were really interested in listening to me talk about home computers, supercomputers, scientific simulation and how all these hang together. In addition to the people who were happy to just get to know about EPCC, these excited individuals made the event seem really worthwhile. While operating the stands and answering questions, I also enjoyed thinking about what other props would make for good demonstrations of serial and parallel computing issues. Surprisingly, sorting coloured balls illustrates many computational concepts, both hardware and software! Justs, student volunteer

Hardware

We used pieces of decommissioned machines to introduce what supercomputers are made of. To contrast and highlight similarities between supercomputers and a regular computer we took a standard desktop system as well. Our four exhibits ranged from three years old to over twenty, indeed several of the pieces were older than those viewing them. The hardware sparked a lot of interest in how the size of technology has changed, which linked nicely to our poster about how big ARCHER would be without the effects of Moores' Law. The questions raised from this set of exhibits were perhaps the most wide ranging of all, from how are they put together and how much do they cost, to what are they used for and are they good for playing games? 

The hardware also allowed people to go onto tangents of interest about parallel and high-performance computing in a manner that the other exhibits didn't spur quite as much. Perhaps this is because it is more of a visual and information-gathering activity rather than a hands-on, doing something activity. Some visitors reminisced about using systems in our display. Indeed one person asked if we could sell parts from one of the older supercomputers as his employer still used one. For younger visitors, it was a chance to see the origins of familiar technology. It was quite a surprise for them to see how big computers used to be in comparison to today's tablets and laptops.

Tuesday started quieter than previous days, but as it progressed more and more people came to see us. The younger kids were kept occupied with the ball sorting and dinosaur racing, which all went down very well. The ball sorting illustrates some key concepts of parallelism and how, by parallelising a problem, one can solve it much faster than when done sequentially. The dinosaur racing demo allows participants to configure their own dinosaur, simulate it racing against other people’s configurations and see who can create the fastest creature. This illustrates how simulation can play a vital role in the sciences where experimentation is not possible.

Many of the older children and adults showed interest in our “hardware through the ages” exhibit, which illustrated four generations of HPC machine and the technological innovations that allowed the leap from one to the next. This was all put into perspective by the “Moores’ Law” poster which showed that if we used 1980s technology to gain the level of processing power we have in ARCHER, we would require a machine hall 15m high and covering most of southern Scotland. Nick, EPCC, one of the developers of the dinosaur simulation software

Dinosaur simulation

The dinosaur simulation has become a staple of our outreach work. As well as showing how computer simulation can help our understanding of dinosaur physiology, it also gives an easy introduction to the technology, firing the imagination of the youngest and showing older users how the underlying techniques can be applied to other areas of research.

One of the things that caught a lot of people by surprise was that the dinosaur models could fall over. This allowed us to illustrate that simulations are not predetermined successes or failures but are attempting to model reality - we still need to test things in the real world, but simulation can help reduce the initial costs before we do physical experiments.

We also explained that simulations can be used to determine if our models are feasible in situations where experimentation is not possible. For example, I don't think people would be happy if we tried to blow up the Moon. This was helped by getting the participants, especially the younger ones, to come up with things that we could put into a simulation that were either really expensive or really dangerous. The ideas that came forth were good, from modelling dinosaur extinction to new aircraft to zombie outbreaks.

The EISF was a totally new experience for me, as it was the first time I had to interact with other people as the core of my post. This was engaging, but also more difficult than I imagined. Apparently, it's really easy to lose the visitors' attention! From the 3 exhibits EPCC had to offer, I think the most successful was the dinosaur race simulation. This reminded me how much people of all ages like dinosaurs. Some of them, however, didn't like the fact that it was an actual simulation and their weirdly-proportioned dinosaurs fell over. Sorry guys! Kostas, student volunteer

The posters and pull-up display often led to questions that started to stretch to the more science-fictional aspects of computers and machine intelligence, but the level of interest shown was impressive as was the willingness to step into something which for a lot of people is an unfamiliar and remote area of science.

Summing up

My own impressions are that for a relative newcomer to this style of event the EPCC exhibits did well, and we have many things to think about improving or adding to the range of activities we provide. We had a busy five days at the Museum. At the end of each day we felt physically tired, but at the same time our minds were not ready to rest and wanted to keep going.

The efforts of Eilidh, Emma, Iain, Justs, Kostas and Nick were excellent and their approachability helped to draw people into discussions. Having experts on hand makes a huge difference to the outcome of events like this, they don't require a script or canned responses to facilitate discussion, which helps put people at ease and retain information. The assistance provided by Mario Antonioletti, Toni Collis, Amy Krause and Nick Brown who set up the exhibits for the event is also greatly appreciated.

I think this event has been very positive for EPCC and hopefully we will continue our involvement in the future. There will always be new questions and new activities to try out.

Author

Alistair Grant, EPCC