Over the three days of 24-26 June, we welcomed 90 undergraduate and postgraduate taught students to #BEARChallenge2025. In an attempt to meet demand for places, this year we expanded to allow 15 teams but demand also rose at a pace, with 31 teams registering their interest! It is great to see the level of interest in taking part in the Challenge, gaining skills in High Performance Computing (HPC) and team work, as well as hearing about the varied opportunities for careers in HPC.
We welcomed teams from across the University; Computer Science, Physics & Astronomy, Mechanical Engineering, Accounting, International Development, and Strategy & International Business. Students ranged from year one undergraduates, through to postgraduates studying for Masters courses. All the participants were granted access to Baskerville, our national Tier 2 High Performance Computing (HPC) system with 228 NVIDIA GPUs.
Rather than describe the event ourselves, we thought it would be best to let you hear about it from the students who experienced it themselves.
“Ever tried telling people you spent three days teaching a computer to paint, break codes & sort galaxies & genuinely had the time of your life doing it?
Joyal Pasricha, MSc student in Data Science and AI – team ‘Neural Net Ninjas’
I had the chance to dive into the world of High Performance Computing (HPC) at the #BEARChallenge2025 hosted by the University of Birmingham & Advanced Research Computing (ARC) in collaboration with NVIDIA & Lenovo.We worked on Baskerville, the University’s flagship HPC cluster, powered by NVIDIA A100 GPUs, submitting SLURM batch jobs.The challenges which pushed our limits were:
Challenge 1 – Neural Style Transfer with VGG19
Blended the power of art & AI, transferring visual styles using deep neural networks to create hybrid pieces.
Engineered a lightning-fast decryption pipeline, demonstrating the power of parallel computation for cybersecurity tasks.
Used Zoobot (ConvNeXt-based deep learning model) to classify galaxy morphologies, blending astronomy with cutting-edge vision AI.
Tackled large-scale data sorting problems by designing memory-efficient algorithms that operate even when GPU memory is maxed out.
Built a cost-effective, high-performance compute cluster, optimized for diverse scientific workloads, hands-on systems thinking in action!”
The students took part in five challenges, as Joyal describes above. Students created some great team logos in Challenge 1, with the bear-related logos being particularly appreciated within the ARC team:
As well as the challenges, throughout the three days the students also heard from speakers describing the varied career options within HPC, as Anshuman describes below:
“The event gave us access to Baskerville’s state-of-the-art system with 8 A100 GPUs and I couldn’t believe that we were submitting SLURM batch scripts to the HPC system from day one—just like a researcher using the system would! It was an amazing experience.
Anshuman Singh, MSc student in Data Science and AI – team leader of ‘AlgoRythms’
To give our adrenaline some time to stabilise after the challenges, we had talks from a wide variety of people during the event: from BEAR champions to researchers who use HPC in their domains. These talks emphasised the awesome ways in which HPC systems are used in research. We also had talks from NVIDIA, Lenovo, and ARC, giving us a glimpse of prospective careers in the field of HPC and sharing their career journeys.
The event was well organised and executed by the ARC team at the University of Birmingham. My teammates and I had great fun during those three days, eating more pizza and snacks than we would like to admit… and the fact that we won second prize in the end was the icing on the cake. Overall, I would highly recommend the BEAR Challenge to any student curious about High Performance Computing.”
At the end of the three days of challenges, the results were announced! A team of year 3 undergraduate Physics & Astronomy students came first, second was ‘AlgoRythms’ (MSc students from Computer Science), and third place went to ‘Beariables’ (a mixed team of first year undergraduates from the School of Physics & Astronomy, Computer Science, and Metallurgy & Materials). It was great to see a team from Physics & Astronomy win the competition for the first time, particularly as they are undergraduates competing against students studying for Masters degrees, and to see a mixed discipline-team placing third. This is the first time that a team outside of Computer Science has won the challenge, and shows the increasing knowledge and skills in programming present in other disciplines.
The top 3 placed teams being awarded their prizes by Professor Stephen Jarvis (Provost and Vice-Principal), and Simon Thompson (Director, HPC Solutions, Lenovo). Left: 1st place was awarded to ‘Under the Hood’, middle: 2nd place went to ‘AlgoRythms’, and right: 3rd placed was ‘Beariables’.
‘Beariables’ team leader, Harry Vicaradge, describes how his team found the event below:
The BEAR challenge is definitely a challenge. It is fast paced and competitive yet the skills you will learn are invaluable. As first years, we found some of the challenges harder than the other teams, but through determination we placed 3rd! A surreal result and experience.”
Harry Vicaradge, BSc Theoretical Physics and Applied Maths – team leader of ‘Beariables’
As ever, we owe many thanks to Lenovo and NVIDIA for continuing to support the event through providing speakers, challenges, and sponsorship of prizes and catering. Thanks to all the speakers who provided such excellent and varied talks on their careers and research, and to the students who participated with such enthusiasm. Lastly, thanks to Professor Stephen Jarvis (Provost and Vice-Principle) for sharing his passion for HPC and presenting the prizes to the winning teams. We look forward to hearing more about how teams will be participating in the newly-designed national CIUK Cluster Challenge and hope that we will be able to support team(s) from the BEAR Challenge to take part once again, so that they can continue their journey into exploring HPC in the short-term, and hopefully as a career in the future.
I leave this challenge with deeper knowledge, new friendships & greater excitement for what lies ahead in AI & HPC.”
Joyal Pasricha, MSc in Data Science and AI – team ‘Neural Net Ninjas’