The documents distributed by this server have been provided by the contributing authors as a means to ensure timely dissemination of scholarly and technical work on a noncommercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, notwithstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.
Publications of SPCL
|MPI Beyond 3.0 and Towards Larger-Scale Computing
(Presentation - presented in Denver, CO, USA, Nov. 2013)
Keynote at ExaMPI 2013 Workshop (in conjunction with SC13)
AbstractThe MPI-3.0 standardization effort finished and many state-of-the art parallel programming techniques have been added to MPI. In this talk, we will briefly discuss the new MPI features, their optimized implementations, and their prospects to change the state of the art in HPC. It can be argued that MPI was design to evolve from the library interface into a parallel language. We will discuss how the current specification enables to handle MPI as a de-facto language and facilitate automatic transformations. We continue by discussing several more radical ideas to extend MPI to support emerging applications such as graph analytics or more dynamic computing settings. We conclude by noting that MPI is well suited to address needs for large-scale computing of the next decades.