AN intelligent road traffic management system using NVIDIA GPU
The original goal of this project was to investigate and compare the experimental performance and ease of programming of algorithms for geometric shortest path finding using shared memory and message passing programming styles on a shared memory machine. However, due to the extended unavailability of a suitable shared memory machine, this goal was only partially met, though a system suitable for testing the hypothesis was implemented. The results gained indicated that the programming style did not have a major impact on run time, though the shared memory style appeared to have a lower overhead. It was found that the message passing style was both easier to program and required less code. Additional experiments were performed on queue type and partitioning method to determine their impact on the performance. It was found that use of a sorted queue had a serious negative impact on the parallelisability of the shortest path algorithms tested, compared with use of an unsorted queue. The use of a multilevel over-partitioning scheme (multidimensional fixed partitioning) gave improved performance with an asynchronous parallel algorithm (by Lanthier et al.), but worsened the performance of a synchronous parallel algorithm (simple parallelisation of Dijkstra’s algorithm).