Proceedings of the 50th Hawaii International Conference on System Sciences | 2017 A Comparison of Task Parallel Frameworks based on Implicit Dependencies in Multi-core Environments Basilio B. Fraguela Universidade da Coruna,˜ A Coruna,˜ Spain Email:
[email protected] Abstract—The larger flexibility that task parallelism offers minimum effort from the developers, while they allow to with respect to data parallelism comes at the cost of a higher build extremely complex task graphs and provide maximum complexity due to the variety of tasks and the arbitrary pat- parallelism thanks to the graph scheduling algorithms they terns of dependences that they can exhibit. These dependencies should be expressed not only correctly, but optimally, i.e. avoid- implement, makes them particularly interesting. ing over-constraints, in order to obtain the maximum perfor- Despite the relevance of this approach, we have not found mance from the underlying hardware. There have been many comparative studies of the existing practical alternatives proposals to facilitate this non-trivial task, particularly within of this kind beyond some performance comparisons [36]. the scope of nowadays ubiquitous multi-core architectures. A very interesting family of solutions because of their large For this reason this paper tackles this issue describing scope of application, ease of use and potential performance and comparing some of the high-level approaches available are those in which the user declares the dependences of each nowadays with a particular focus on their semantics and ease task, and lets the parallel programming framework figure out of use, our main aim being to help identify the best tool which are the concrete dependences that appear at runtime for each problem at hand.