Parallel rendering of human-computer interaction industrial applications on multi-/many-core platforms

Abstract

Industrial Human Computer Interaction (Industrial HCI) devices are beginning the transition from single-core to multi-/many-core technology. In practice, improving the real-time response time of graphical user interface (GUI) applications in multi-/many-core is difficult. This paper presents a novel parallel rendering approach targeted to improve the performance of Industrial HCI applications in multi-/many-core technology. This is accomplished through the identification of coarse-grain parallelism during the application design, and the exploitation of fine-grain parallelism during runtime using a dynamic scheduling algorithm and true parallel execution of GUI workloads. Using a real benchmark application, we show that response time can be reduce by up to 217% in a quad-core processor. © 2013 Springer-Verlag.