Real-time Scheduling upon Heterogeneous Multiprocessors

Abstract

As the computational demands made by ever more complex embedded real-time applications continue to increase, there is a need for enhanced performance capabilities from these platforms. Initially, the approach adopted for obtaining such enhanced performance was to increase core counts in multiprocessor CPUs. Soon, however, chip makers began to distinguish themselves not only by offering more general-purpose cores but also by providing specialized hardware components that accelerate particular computations. Examples include multicore CPUs with specialized graphics processing cores, specialized signal-processing cores, specialized floating-point units, customizable FPGAs, etc. Computing platforms such as these with specialized components are called unrelated or heterogeneous [143] multiprocessor platforms.