Guest Editors' note: Large-Scale Parallel Processing

As parallel processors scale to thousands of nodes, often with custom interconnects and file systems, managing their operation and tuning performance becomes more challenging. This special issue focuses on research that is considering the impact of architectural scaling across orders of magnitude, from various perspectives. Three of the papers in this issue are concerned with analyzing and improving performance through modeling or direct measurement of system performance factors. The other two papers are focused on paradigm shifts that are needed as systems scale up.