The UCL Centre for Advanced Research Computing (ARC) - supported by funding from DiRAC, which we gratefully acknowledge - ran a student HPC careers day in February 2022 and a follow-up hands-on student cluster challenge in May 2022, with the aim of developing a framework for the HPC community to grow undergraduate and graduate students’ interest in pursuing a career in HPC. We used the strapline "Kickstart your HPC journey", hashtag "KickStart HPC", and associated logo on Twitter, LinkedIn and the ARC website to promote the aims of the project, and the events we hosted. We hope that others find the resources in this repository useful if planning to run a similar challenge.
The careers day was run as a Zoom webinar and attracted attendees from across UCL and several external institutions, with a variety of speakers from both academia and industry talking about their own careers. The webinar was open to all, with no requirement for students status or intent to participate in the cluster challenge, although obviously this was encouraged.
In the event, the cluster challenge was limited to UCL undergraduate and postgraduate students, partly because of the logistics of providing access to the UCL cluster; we did invite all of those who attended the careers day to consider entering the challenge and were prepared if necessary to make up mixed teams with both external participants and UCL students working on the challenge together, with a UCL student submitting the jobs to the cluster. In order to make the hands-on challenge as accessible as possible, we ran this as a hybrid event with those unable to attend in-person joining online via Teams. We chose Python as the programming language to encourage those not familiar with the more traditionally-used HPC languages. Although Python does not provide as much flexibility and potential for performance tuning as, for example, C++, there are enough parallel features of Python to introduce core parallel programming concepts and to hook participants on "making code run real fast".
We strongly recommended that in order to get the most out of the challenge, the students should have a minimum level of knowledge/experience including familiarity with Unix/Linux command line, some experience with Python, and ideally, having attended an HPC Carpentry workshop (or similar) for background knowledge of HPC clusters and batch job submission.
We provided a list of suitable self-paced resources for students to work through before the two days of the hands-on challenge, that includes the DiRAC Foundation HPC-skills Course which offers a complete suite of training designed to give you an introduction to the principles of HPC and the tools needed to work on an HPC system.
The student cluster challenge days both had a taught element in the morning, followed by a hands-on challenge in the afternoon. We used Timo Betcke's Techniques of High-Performance Computing as a basis for the teaching portion of the event. More details can be found in the challenge folder of this repository.
Feedback from the careers day was collected via sli.do, and we used a feedback form for the hands-on cluster challenge as well as positive/negative post-it notes on the day. This feedback can be seen in the examples folder of this repository, as well as a summary of the lessons we learned in the prerequisites folder.