Skip to Main Content U.S. Department of Energy
PNNL: High Performance Computing

ComEx: Communications Runtime for Extreme Scale

The Communication runtime for Extreme Scale (ComEx) is a successor of the Aggregate Remote Memory Copy Interface (ARMCI). ComEx uses native interfaces for facilitating one-sided communication primitives in Global Arrays. As an example, ComEx has been designed to use Openfabrics Verbs (OFA) for InfiniBand and RoCE Interconnects, Distributed Memory Applications (DMAPP) for Cray Gemini Interconnect, and PAMI for x86, PERCS, and Blue Gene/Q Interconnects. The specification is being extended to support multi-threading, group aware communication, non-cache-coherent architectures and generic active messages. ComEx provides abstractions for RMA operations such as get, put and atomic memory operations and provides location consistency.

The latest version of ComEx is bundled with the Global Arrays software as of GA 5.4b. Future versions of ComEx will be available stand-alone. Please see the GA download page to download ComEx as part of GA 5.4b.

Acknolwedging ComEx

We request you to include the following reference in your paper, if you are using Global Arrays 5.3 or above.

Jeff Daily, Abhinav Vishnu, Bruce Palmer, Hubertus van Dam, and Darren Kerbyson. On the use of MPI as a PGAS Runtime. International Conference on High Performance Computing (HiPC) 2014. (pdf), (bib).

Acknowledgements

The research and development on ComEx is supported by Extreme Scale Computing Initiative, Center for Programming Models for Scalable Parallel Computing, Environmental and Molecular Science Laboratory and grants from Intel.

Site Links

Our Research

Past Research

ComEx Links

News and Events

Feedback

Contacts