The ISC Student Cluster Competition 2020 brought together students representing universities all over the world to take part in a friendly competition to showcase their expertise in high performance computing. Due to Covid-19, this year’s competition was fully virtual, with student teams using the supercomputing resources at the National Supercomputing Centre Singapore to complete a variety of challenges - four HPC applications, the Coding Challenge, and the AI challenge. This year’s coding challenge was based on Charm++. The coding challenge required students to write code for particle movement in a 2-dimensional grid (divided into cells) in order to send particles to their correct cells after being perturbed. Each team also had to use Charm++ asynchronous custom reductions to determine the cells with maximum and minimum number of particles. Students were also encouraged to use the visualization tool in the Charm++ ecosystem, liveViz, to visualize the movement of particles and post their visualization videos on Twitter. Another Charm++ based application in the domain of cosmology, called ChaNGa, was listed as one of the benchmarking applications. Students used ChaNGa to run a cosmological simulation of a galaxy cluster with an unprecedented resolution (using the h1.768.vis.param input file). ChaNGa simulation also created excellent visualizations of galaxy clusters, which were posted on Twitter by the participating teams. Thanks to Nitin Bhat from Charmworks, Inc. and Tom Quinn from University of Washington for being a part of this year’s ISC SCC organizing committee and leading the efforts in organizing the coding challenge and ChaNGa respectively. The competition ended on July 17th and the award ceremony took place on July 24th. We would like to thank all the participants for using Charm++ and ChaNGa, and asking interesting questions about it. We would also like to take the opportunity to congratulate the winners of this year's competition, the University of Science and Technology of China (USTC), China.
To learn more about Charm++, click here.