The 2050-carbon-neutral goal has sparked increasing installation of renewables, leading to a growing number of power electronic devices in electric power systems. As a result, the requirements on the simulation speed and scalability has grown, posing high challenge to the traditional simulation techniques.

In this talk, we’ll start with an introduction on the computational effort as well as the modeling approach to improve the simulation performance using GPU, including the exploitation of data and task parallelism, etc., as well as the extension of compatibility for a wide range of GPU devices. Afterwards, we’ll introduce the optimization techniques to further improve the code performance for simulation on GPU.