Скачать или смотреть Benchmarks of WRF v4.2 on Google Cloud and Cheyenne

In this livestream, we'll share with you benchmarks of WRF v4 on Google Cloud Platform and the Cheyenne supercomputer at NCAR. Specifically, we'll look at the CONUS 2.5km and CONUS 12km benchmarks across machine types available on Google Cloud. The benchmarks are presented with both the time-to-solution and cost for running short forecasts on GCP, and we examine the impacts of these metrics on process affinity, specifically with relation to the use of hyptherthreading. I'll show you a strategy to optimize file IO performance through the use of parallel file IO in WRF and a cloud-native Lustre file-system on Google Cloud.

We welcome your comments, thoughts, questions, and feedback at this livestream! At the end of the presentation, there will be time for Q&A.

Additional Resources
Codelab : https://codelabs.developers.google.co...

HPC Apps GCP Repository : https://github.com/FluidNumerics/hpc-...

WRF on GCP : https://www.fluidnumerics.com/product...