One of the fundamental roadblocks to achieving acceptable levels of performance of multi-physics simulations like white dwarf (WD) mergers is the poor utilization of available computational resources. The STAR (Scalable toolkit for Transformative Astrophysics Research) research project contributes significantly to the deeper understanding of such scientific applications, and to the understanding and capability of efficient scalable computing on any scale. In this project we plan to explore new techniques in the management of parallel execution and resource management in order to overcome the massive workload and communication requirements imposed by modern astrophysics simulations.
Through the collaboration of an interdisciplinary team of astrophysicists and computer scientists, the STAR team is working on an application which will exploit the performance benefits offered by cutting edge techniques implemented in HPX and apply them to WD mergers. This application will allow astrophysicists to produce more accurate results in a shorter period of time. The application’s development will be driven by novel research codes developed by members of the astrophysics team. These new techniques, supported by more efficient parallization paradigm, will enable larger, more refined simulations necessary for the advancement of astrophysics research.
To track the progress of our work please visit our GitHub repository.
This work is supported by NSF Award Number: 1240655