Learn about climate, environment, and weather on AWS HPC
Weather simulation is a reliably difficult workload for almost any HPC architecture and is often used as a litmus test by many customers before they look at a novel technologies or different systems. Customers have asked us frequently about our performance, and that’s been even more the case since we launched the Elastic Fabric Adapter and several new HPC-oriented instance families.
The Hpc6a is built on AMD’s Milan, and the AWS Graviton is our Arm64-based processor. Both perform excellently with a variety of weather and climate codes, as you can see in the blog posts, videos and discussions linked on this page.
Numerical weather prediction (NWP) codes
-
An overview of Numerical Weather Prediction (NWP) workloads on AWS HPC-optimized instances. We test three popular NWP codes: WRF, MPAS, and FV3GFS on the Hpc6a (Blog).
-
NAVGEM on the Cloud: Computational Evaluation of Cloud HPC with a Global Atmospheric Model - A talk from the HPC User Forum about a project comparing the performance of the NAVGEM on an in-house supercomputer system against several cloud offerings including AWS. (Video, Fall 2019).
-
Running the Harmonie weather model on AWS - the Danish Meteorological Institute (DMI) ported and ran their full cycling dataflow with the Harmonie model on AWS (Blog).
-
Training forecasters to warn severe hazardous weather on AWS (Blog).
The impact of AWS Graviton in modeling
- WRF Performance on AWS Graviton2 - (Blog || Video) - an exploration of how WRF performs on AWS Graviton2, delivering suprising performance results.
- Arm a world-leading forecast model with AWS Graviton and Lambda - the UK MetOffice describe their experience of gathering their SurfaceNet data gathering of over 1 billion observations a year using AWS Lamba and their switch to using AWS Graviton (Blog).
Getting hands on (with help) in a workshop
These workshops lead you through getting started with AWS ParallelCluster including installation and use of WRF, MPAS, and Unified Forecase System FV3GFS. (Workshop)
You can work through them in your own AWS account, or use them in a class-room scenario for a group.