SC20 Proceedings

The International Conference for High Performance Computing, Networking, Storage, and Analysis

TaskWorks: A Task Engine for Empowering Asynchronous Operations in HPC Applications

Authors: KAI-YUAN HOU (Northwestern University, Lawrence Berkeley National Laboratory) and Quincey Koziol and Suren Byna (Lawrence Berkeley National Laboratory)

Abstract: TaskWorks is a portable, high-level task engine designed for HPC workloads. Applications can create tasks and define dependencies between them with the task engine. Once the task is defined and submitted to TaskWorks, the TaskWorks engine will execute it according to the specified dependencies, without additional input from the application. TaskWorks has an integrated event manager that monitors files, sockets, timers and MPI operations, and these events can be used as task dependencies. TaskWorks is compatible with MPI and is designed to work efficiently with HPC applications that perform inter-process (node) communication.

Best Poster Finalist (BP): no

Poster: PDF
Poster summary: PDF

Back to Poster Archive Listing