Anjali Sandip, University of North Dakota
Irina Tezaur, Sandia National Laboratory
Recent advances in computational hardware have provided opportunities to diminish resolution constraints and increase computing speed in the development of large-scale numerical models. However, utilizing the hardware at their maximal capabilities still remains a challenge. Developing algorithms that optimally leverage the hardware capabilities with performance-portable implementations, enabling the same code to execute correctly and efficiently across a wide range of CPU and GPU architectures, has therefore become an important aspect of numerical modeling.
This mini-symposium will feature presentations on algorithms that optimally leverage the hardware capabilities with performance-portable implementations for unstructured mesh applications. Relevant topics include algorithms applied to high spatial and/or temporal resolution large-scale numerical models.