Combining loop unrolling strategies and code predication to reduce the worst-case execution time of real-time software

Worst-case execution time (WCET) is a parameter necessary to guarantee timing constraints on real-time systems. The higher the worst-case execution time of tasks, the higher will be the resource demand for the associated system. The goal of this paper is to propose a different way to perform loop unrolling on data-dependent loops using code predication targeting WCET reduction, because existing techniques only consider loops with fixed execution counts. We also combine our technique with existing unrolling approaches. Results showed that this combination can produce aggressive WCET reductions when compared with the original code.