How Sailfish Optimized and Accelerated Numerical Ocean Modeling

To solve critical regional problems, such as water quality issues near the coast, in the shelf and estuaries, researchers must simulate ocean circulation. As the observational ocean data is sparse, numerical ocean models are essential for building a complete picture of all the physical processes and making predictions. 

The global numerical ocean models present their own set of challenges. Non-linear in their nature, they require hundreds of computationally costly simulations to produce accurate forecasts. Also, the sheer size and complexity of the existing models mean they do not perform to their best potential speed, nor do they scale easily to larger problems or added physical parameters.
 

Sailfish was created as a fast and efficient counterpart to traditional numerical ocean models. This performant and energy-efficient GPU-based model is being developed by the team at Herbert Wertheim College of Engineering University of Florida led by Jose Maria Gonzalez Ondina, Ph.D., of the Center for Coastal Solutions.

Watch the presentation to learn how Sailfish performs 1500 times faster than Regional Ocean Modeling System (ROMS),
a traditional ocean numerical model, on certain modelling cases. Watch now.