Constraint Propagation on GPU: A Case Study for the Cumulative Constraint

The Cumulative constraint is one of the most important global constraints, as it naturally arises in a variety of problems related to scheduling with limited resources. Devising fast propagation algorithms that run at every node of the search tree is critical to enable the resolution of a wide range of applications. Since its introduction, numerous propagation algorithms have been proposed, providing different tradeoffs between computational complexity and filtering power. Motivated by the impressive computational power that modern GPUs provide, this paper explores the use of GPUs to speed up the propagation of the Cumulative constraint. The paper describes the development of a GPU-driven propagation algorithm, motivates the design choices, and provides solutions for several design challenges. The implementation is evaluated in comparison with state-of-the-art constraint solvers on different benchmarks from the literature. The results suggest that GPU-accelerated constraint propagators can be competitive by providing strong filtering in a reasonable amount of time.