Support for OpenMP tasks in Nanos v4
CASCON '07 Proceedings of the 2007 conference of the center for advanced studies on Collaborative research
IEEE Transactions on Parallel and Distributed Systems
ICPP '09 Proceedings of the 2009 International Conference on Parallel Processing
Safe programmable speculative parallelism
PLDI '10 Proceedings of the 2010 ACM SIGPLAN conference on Programming language design and implementation
Common mistakes in OpenMP and how to avoid them: a collection of best practices
IWOMP'05/IWOMP'06 Proceedings of the 2005 and 2006 international conference on OpenMP shared memory parallel programming
Implementing irregular parallel algorithms with OpenMP
Euro-Par'06 Proceedings of the 12th international conference on Parallel Processing
Towards an error model for OpenMP
IWOMP'10 Proceedings of the 6th international conference on Beyond Loop Level Parallelism in OpenMP: accelerators, Tasking and more
| Hi-index | 0.00 |
Multi-core processors are at the heart of current and future trends for computer architectures. The number of cores on a single chip is rapidly increasing, so as the need for simpler and more efficient programming models. OpenMP is a powerful programming model that has been adopted by a large spectrum of research and development teams to develop parallel applications on multi-core processors. To fully exploit the available cores in multi-cores chips, the latest versions of OpenMP specification marked a transition from a thread-centric to a task-centric programming model. Hence, tasks are used to express parallelism and to execute concurrent computations. However, this programming model suffers from the lack of a useful feature where created tasks can be explicitly cancelled. Task cancellation is considered an important aspect in programs based on speculative execution and search algorithms, where computation resources should be quickly released if not needed, yielding higher computation efficiency and lower power consumption. In this paper, we present a proposal and an enhancement to the OpenMP programming model that allows users to create special type of tasks called "cancellable tasks". New easy to use extensions are added in order to support both cooperative and forced cancellation of these special tasks through specific cancellation calls. We will show that these extensions reduce execution time and response delays of parallel applications that may need cancellation and prompt resources re-allocation compared to the user cancellation approach based on flags.