A Step Ahead: Novel HPC Approaches to provide Early Warnings for Severe Weather
Scientific Software Development
TimeTuesday, June 26th2:07pm - 2:29pm
DescriptionSevere weather warnings based on numerical weather prediction models are a public service offered by the government and form part of basic meteorological provision. Accurate and timely warnings play an important role in protecting the public and informing the economic sector. But, the weather and climate community is struggling to keep up with the fast-paced developments as HPC system designs strive towards the Exascale. Only few production-ready community codes are able to leverage emerging hardware architectures. This has become known as the software productivity gap. While programming models - such as OpenACC - are available to target these hardware architectures, they further increase the complexity of the code and thus exacerbate the problem. This talk summarizes a co-design effort for a regional weather and climate model that presents an alternative pathway. It will showcase how MeteoSwiss is developing and deploying novel HPC approaches to improve its real-time weather forecasting system. Significant parts of the code have been rewritten based on a domain-specific language (DSL). The DSL allows a high-level hardware agnostic formulation of models of physical processes which can be compiled to different hardware targets. Finally, we will demonstrate the capability of the refactored code on the leadership-class heterogeneous HPC system Piz Daint in Switzerland -- currently the #3 supercomputer worldwide -- and discuss the implications for global km-scale weather and climate simulations together with the question of how far do we still have to go to make the next leap forward in weather and climate modeling.