Abstract of LCHD Project

Predicting the performance of fusion plasmas in terms of amplification factor, namely the ratio of the fusion power over the injected power, is among the key challenges in fusion plasma physics. In this perspective, turbulence and heat transport need being modeled within the most accurate theoretical framework, using first-principle non-linear simulation tools. Using the non-linear global full-f gyrokinetic 5D code GYSELA, a simulation very close to ITER-size plasmas with kinetic ions and adiabatic electrons has been performed on 8192 cores during 31 days. This type of simulation is at the front-edge of current research in fusion plasma modeling.

In each part of the code GYSELA, parallel algorithms have been designed in order to scale up to thousands of cores using a hybrid MPI/OpenMP approach. Some bottlenecks concerning memory scalability and collective communication costs have been recently removed. With such improvements, the relative efficiency reaches 78% on 65k cores.

The aim of the new step proposed here is the improvement of the outputs of the code in order to compress 3D diagnostics and huge restart files. For very large simulations on several thousands of cores (Curie – France, Helios – IFERC) that we expect in the next years, we will have 100 TB of exported data at each run. We absolutely need a refined strategy to reduce this amount of data in order to: speedup simulation time because file system bandwidth is a bottelneck, and retrieve/download results without difficulties.