Performance comparison of multiprocessor programming libraries for multicore environments
Pien, Hu Yun · ProQuest Demo Repository
Sequentially coding is usually the quickest way to complete a given task when building a software application. The CPU executes program code one line after another during the process of the application. However, most modern computing devices even smart phones now have multiple cores available. Due to this technology, sequential execution would waste the potential performance when only a single core is working. To write code in parallel requires the understanding of parallel patterns in order to preserve the proper implementation of an algorithm. At this point, libraries such as Intel Threading Building Blocks (TBB) and Microsoft Task Parallel Library (TPL) are the easy-to-use tools for multicore environment that help programmers build their parallel solutions. This paper contains recommended steps to exploit multi-core parallelism. A real case application is also implemented to demonstrate the discovered potential parallel computing power and to compare the performance of these TBB and TPL.