Scalable Solvers on Emerging Architectures

Luke Olson, University of Illinois at Urbana-Champaign

Usage Details

Sparse solvers are a critical piece of many scientific simulations. Yet, the irregular data patterns and communication demands of the underlying sparse kernels challenge scalability and efficiency on current parallel machines such as Blue Waters. The focus of this work is on developing new algorithms to address the needs in recent advances in multigrid methods. Moreover, the main objective of this work is to position sparse solvers and the supporting software stack for upcoming and emerging parallel machines. In this sense, Blue Waters is a critical resource for the study as both traditional CPU compute nodes are available as well as nodes with accelerators. This work seeks to extend the compute capabilities of multigrid methods in both a structured and an unstructured setting.