Analysis of Data Reuse in Task-Parallel Runtimes Miquel Pericàs?, Abdelhalim Amer†

Abstract—This paper proposes a methodology to study the data reuse quality of task-parallel runtimes. We introduce an extension to the reuse distance method called the Kernel Reuse Distance (KRD). The metric is a low-overhead alternative designed to ana-lyze data reuse at the socket level while minimizing perturbation to the parallel schedule. Using the KRD metric we show that reuse depends considerably on the system configuration (sockets, cores) and on the runtime scheduler. Furthermore, we correlate KRD with hardware metrics such as cache misses and work time inflation. Overall we found that KRD can be used effectively to assess data reuse in parallel applications. The study also revealed that several current runtimes suffer from severe bottlenecks at scale which often dominate performance. I. INTRODUCTION AND BACKGROUND Tasking has become an established technique to progra