Abstract
The Global Address Space Programming Interface (GASPI) is a Partitioned Global Address Space (PGAS) API specification. The GASPI API specification is focused on three key objectives: scalability, flexibility and fault tolerance. It offers a small, yet powerful API composed of synchronization primitives, synchronous and asynchronous collectives, fine-grained control over one-sided read and write communication primitives, global atomics, passive receives, communication groups and communication queues. GASPI has been designed for one-sided RDMA-driven communication in a PGAS environment. As such, GASPI aims to initiate a paradigm shift from bulk-synchronous two-sided communication patterns towards an asynchronous communication and execution model. In order to achieve its much improved scaling behaviour GASPI leverages request based asynchronous dataflow with remote completion. In GASPI request based remote completion indicates that the operation has completed at the target window. The target hence can (on a per request basis) establish whether a one sided operation is complete at the target. A correspondingly implemented fine-grain asynchronous dataflow model can achieve a largely improved scaling behaviour relative to MPI.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
MPI Forum: MPI: A Message Passing Interface Standard, http://www.mpi-forum.org/docs/docs.html (2014)
Chamberlain, B.L., Callahan, D., Zima, H.P.: Parallel programmability and the chapel language. Int. J. High Perform. Comput. Appl. 21(3), 291–312 (2007)
Consortium, U.: UPC Specifications v1.2. Lawrence Berkeley National Lab Tech Report LBNL-59208 (2005)
Hilfinger, P., Bonachea,D., Datta, K., Gay, D., Graham, S., Liblit, B., Pike, G., Su, J., Yelick, K.: Titanium Language Reference Manual. U.C. Berkeley Tech Report UCB/EECS-2005-15 (2005)
Numrich, R.W., Reid, J.: Co-array Fortran for parallel programming. ACM Fortran Forum 17, 1–31 (1998)
Machado, R., Lojewski, C.: The Fraunhofer virtual machine: a communication library and runtime system based on the RDMA model. Comput. Sci. Res. Dev. 23, 125–132 (2009)
Fraunhofer ITWM: GPI - Global Address space programming Interface. http://www.gpi-site.com (2013)
Simmendinger, C., Jägersküpper, J., Machado, R., Lojewski, C.: A PGAS-based implementation for the unstructured CFD solver TAU. In: 5th Conference on Partitioned Global Address Space Programming Models, Tremont House, Galveston Island (2011)
Machado, R., Lojewski, C., Abreu, S., Pfreundt, F.-J.: Unbalanced tree search on a manycore system using the GPI programming model. Comput. Sci. Res. Dev. 26(3–4), 229–236 (2011)
Poole, S.W., Hernandez, O., Kuehn, J.A., Shipman, G.M., Curtis, A., Feind, K.: OpenSHMEM - toward a unified RMA model. In: Encyclopedia of Parallel Computing, pp. 1379–1391 (2011)
Acknowledgements
The authors would like to thank the German Ministry of Education and Science for funding the GASPI project (funding code 01IH11007A) within the program “ICT 2020 - research for innovation”. Further more, the authors are grateful to all of project partners for having fruitful and constructive discussions.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2015 Springer International Publishing Switzerland
About this paper
Cite this paper
Simmendinger, C., Rahn, M., Gruenewald, D. (2015). The GASPI API: A Failure Tolerant PGAS API for Asynchronous Dataflow on Heterogeneous Architectures. In: Resch, M., Bez, W., Focht, E., Kobayashi, H., Patel, N. (eds) Sustained Simulation Performance 2014. Springer, Cham. https://doi.org/10.1007/978-3-319-10626-7_2
Download citation
DOI: https://doi.org/10.1007/978-3-319-10626-7_2
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-10625-0
Online ISBN: 978-3-319-10626-7
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)