Accelerating the FlowSimulator: Performance Analysis & Optimization of DLR High-Performance Solvers using Score-P

Abstract

With rapidly growing interest in high-fidelity simulations in the field of aircraft design, high performance computing becomes indispensable. However, processors evolve towards an ever-increasing core count, raising requirements on the scalability of scientific software. Consequently, the coordinated interaction of algorithms, hardware and software is required to efficiently use the available resources. This work deals with an in-depth scalability analysis of the flow solvers CODA [12] and Musubi [11] employed at DLR. Thereafter, it is refined by profiling and tracing the applications using Score-P [15]. The analysis reveals sub-optimal communication patterns; possible solutions are proposed and their impact evaluated.