Artemis Phase 3: Another boost for compute-intensive research
Artemis, the University of Sydney’s supercomputer, is currently undergoing its most significant expansion since its original launch in 2015. With a continuing growth of users, Artemis has been heavily utilised over the last year, necessitating further capacity to support more users and boost research capability. Therefore, we have purchased forty-nine compute nodes and seven GPU-accelerated nodes, adding 2,352 compute cores and twenty-eight GPUs, and this capability will be available to researchers in the next several weeks.
All Artemis users will be able to use twenty-eight of NVIDIA’s latest and most powerful general-purpose GPU, which are well-suited for deep learning applications. Two V100 GPUs using the V100 “tensor cores” have more computing power than all of Artemis before this upgrade!
We have also partnered with NVIDIA to deliver two GPGPU-related programming training sessions on 26 and 27 February - please visit the Sydney Informatics Hub website to learn more. This initiative helps ensure that our researchers can leverage the incredible compute potential of these new GPUs. So far, eighty people have registered for this training.
Individual schools can purchase compute nodes or GPGPU accelerated nodes dedicated to School-specific research – for example, Dacheng Tao from the School of Information Technology bought eighty V100 GPGPUs for his research group. Dacheng is a pioneer in the machine learning and computer vision fields, and these new GPUs will ensure his research group has the compute power to keep up with the latest deep learning and artificial intelligence research.