Proxy Application Driven Codesign for Power Systems Analytics on Extreme-Scale Architectures

Modern scientific applications are complicated and require coordination of several components. Proxy application driven software-hardware codesign plays a vital role in driving innovation among the developments of applications, software infrastructure and hardware architecture. Proxy applications are self-contained and simplified codes that are intended to model the performance-critical computations within applications. Power systems are susceptible to congestion, renewable curtailment, and loss of load. Numerical simulations and machine learning models to study the impact of potentially adverse events on the power grid under variable and uncertain generation are actively researched. Modeling dynamical systems such as the power grid on HPC systems often exhibit irregular computation and memory accesses, reasons behind performance bottlenecks at scale. Such use cases are rarely explored for performance characterization. In this panel, we discuss the prospect of proxy applications in analyzing the performance of power systems analytics on extreme-scale hardware architectures: spanning GPUs to massively parallel system-on-chip architectures.