Adventures in Antarctic Computing, or How I Learned to Stop Worrying and Love the Neutrino

IceCube--a neutrino telescope that encompasses a cubic kilometer of Antarctic ice at the South Pole, collecting and processing data from 5,160 optical sensors buried a mile deep in the icecap--presents considerable challenges, from overcoming power and bandwidth limitations to simulating the complexities of Antarctic ice, which continue to stretch computing technology. IceCube has faced many computational challenges during its evolution: successfully deploying 5,160 robust, ultralow-power networked computers under 1.5 kilometers of ice; sifting through a terabyte of data daily to cull out just 100 Gbytes of events to send from the SouthPole to the north; processing this data, using sophisticated, CPU-hungry algorithms; and developing satisfactory tools, building appropriate GPU clusters, and managing data to facilitate simulation production. IceCube's success in meeting these challenges is demonstrated by its successful physics results.