Case Study: NASA’s Global Modeling and Assimilation Office achieves faster, higher-resolution weather and climate modeling with Cornelis Networks Omni-Path

Discover cluster with 6.7 petaFLOPS peak capacity will enable high resolution weather predictions and climate simulation and modeling

NASA’s Global Modeling and Assimilation Office (GMAO) needed immense computing power to run high-resolution global weather predictions and climate simulations to support various NASA missions. Their challenge was processing over five million observations every six hours to create forecasts and models, a task that required a significant upgrade to their existing supercomputing infrastructure.

Cornelis Networks, through its Intel® Omni-Path Architecture fabric, was part of the solution implemented in the Discover supercomputer’s latest upgrade. This included 640 new nodes with Intel® Xeon® Gold processors, boosting the cluster's peak capacity to 6.7 petaFLOPS. The results enabled GMAO to achieve a groundbreaking 1.5 km² global simulation and transition their research into a new operational 3 km² forecasting product, significantly enhancing the detail and accuracy of their predictions for NASA and the global scientific community.

NASA’s Global Modeling and Assimilation Office

Bill Putman

Researcher