Written by Murphy McDonald-Smith
This is the fifth in a series of blog posts from our Year 10 Information Technology in a Global Society (ITGS) students on the impact of High Performance Computing (HPC) in a global society. Each year the Research Computing Centre (RCC) at UQ, provide funding for four QASMT students to attend a Super Computing Conference. In Year 10 we introduce our ITGS students to HPC and learning to blog as a 21st century skill.
Scientists at Berkeley Lab & ORNL and engineers from NVIDIA use deep learning techniques to simulate climate events.
Section 1: Presentation of the HPC application
Co-winning the Gordon Bell Prize at SC18 for breaking the exaop* barrier with climate simulations, the collaborative efforts of scientists at Berkeley Lab and Oak Ridge National Laboratory and engineers from NVIDIA used the computational ability of the Summit supercomputer to process terabytes of climate data to train neutral networks and thus produce accurate climate predictions (Kincade, 2018).
These simulations and models generated can now be used by the US government, or even governments worldwide, to predict and prepare for extreme weather events that arise due to the changing climate. Especially because it is quite accurate, making it not only beneficial for the population but now financially viable in terms of preparing for these weather events (Kurth, et al., 2018).
This team of scientists and engineers on the project needed to train the neural networks of DeepLabv3+ (the software) to recognise patterns in the seemingly erratic data and teach it to apply those pattern-finding algorithms to simulations of future weather. Therefore, it was necessary to employ the computation power of the Summit supercomputer to process the enormous number of calculations required (Lawrence Berkeley National Laboratory, 2018).
Read more
