Task Parallelism for TwigStack Algorithm on a Multi-core System

Abstract The advancement of multi-core processor technology has changed the course of computing and enabled us to maximize the computing performance. In this study, we present an approach of task parallelism for the TwigStack algorithm on a multi-core system to find all occurrences of an XML query twig pattern in a large XML database. The TwigStack agorithm comprises two tasks. The first task is to perform query pattern matching against XML data and to generate partial solutions. Meanwhile, the second task is to merge the partial solutions generated by the first task for final solutions. To achieve query performance improvement, the task parallelism employs the pipeline parallelism technique for the two tasks. The experimental results show that the query performance is significantly improved especially for queries having relatively more complex structures and/or higher selectivities.