New frontiers of high performance computing for climate change studies

Posted on

The complexity and resolution of current climate models increase computational and memory requirements.
The development of high-resolution models is prohibitive in sequential mode, but it can be supported by high-performance computing systems if an efficient parallel implementation is available. In other words, we need to re-design the sequential problem as a list of sub-problems to be executed (as much as possible) independently and efficiently on the modern HPC systems.

Silvia Mocavero, CMCC researcher at ASC – Advanced Scientific Computing Division explores this topic in the seminar “Design and implementation of the SHYFEM parallel model”. She illustrates the parallelization steps of the Shallow Water Hydrodynamic Finite Element Model (SHYFEM), a three-dimensional hydrodynamic model used by CMCC researchers  to implement the sub-regional costal SANIFS (Southern Adriatic Northern Ionian coastal Forecasting System) configuration.

The SANIFS configuration, developed by CMCC at OPA – Ocean Predictions and Applications Division, is considered as test case to demonstrate the speedup of the implemented parallel solution.

Watch the video:

Start typing and press Enter to search

Shopping Cart