MPI Detach - Asynchronous Local Completion
Authors:
Open Access
- 21 September 2020
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 71-80
Abstract
When aiming for large scale parallel computing, waiting time due to network latency, synchronization, and load imbalance are the primary opponents of high parallel efficiency. A common approach to hide latency with computation is the use of non-blocking communication. In the presence of a consistent load imbalance, synchronization cost is just the visible symptom of the load imbalance. Tasking approaches as in OpenMP, TBB, OmpSs, or C++20 coroutines promise to expose a higher degree of concurrency, which can be distributed on available execution units and significantly increase load balance. Available MPI non-blocking functionality does not integrate seamlessly into such tasking parallelization. In this work, we present a slim extension of the MPI interface to allow seamless integration of non-blocking communication with available concepts of asynchronous execution in OpenMP and C++.Keywords
Funding Information
- European Union's Horizon 2020 (824080)
This publication has 7 references indexed in Scilit:
- CHAMELEON: Reactive Load Balancing for Hybrid MPI+OpenMP Task-Parallel ApplicationsJournal of Parallel and Distributed Computing, 2020
- Improving the Interoperability between MPI and Task-Based Programming Models2018
- OpenSHMEM Specification 1.42017
- Trace-Based Detection of Lock Contention in MPI One-Sided Communication2017
- Enabling communication concurrency through flexible MPI endpointsThe International Journal of High Performance Computing Applications, 2014
- The deep computing messaging framework2008
- Dynamic load balancing for distributed memory multiprocessorsJournal of Parallel and Distributed Computing, 1989