@article {icl:734, title = {Kernel-assisted and topology-aware MPI collective communications on multi-core/many-core platforms}, journal = {Journal of Parallel and Distributed Computing}, volume = {73}, year = {2013}, month = {2013-07}, pages = {1000-1010}, abstract = {Multicore Clusters, which have become the most prominent form of High Performance Computing (HPC) systems, challenge the performance of MPI applications with non-uniform memory accesses and shared cache hierarchies. Recent advances in MPI collective communications have alleviated the performance issue exposed by deep memory hierarchies by carefully considering the mapping between the collective topology and the hardware topologies, as well as the use of single-copy kernel assisted mechanisms. However, on distributed environments, a single level approach cannot encompass the extreme variations not only in bandwidth and latency capabilities, but also in the capability to support duplex communications or operate multiple concurrent copies. This calls for a collaborative approach between multiple layers of collective algorithms, dedicated to extracting the maximum degree of parallelism from the collective algorithm by consolidating the intra- and inter-node communications. In this work, we present HierKNEM, a kernel-assisted topology-aware collective framework, and the mechanisms deployed by this framework to orchestrate the collaboration between multiple layers of collective algorithms. The resulting scheme maximizes the overlap of intra- and inter-node communications. We demonstrate experimentally, by considering three of the most used collective operations (Broadcast, Allgather and Reduction), that (1) this approach is immune to modifications of the underlying process-core binding; (2) it outperforms state-of-art MPI libraries (Open MPI, MPICH2 and MVAPICH2) demonstrating up to a 30x speedup for synthetic benchmarks, and up to a 3x acceleration for a parallel graph application (ASP); (3) it furthermore demonstrates a linear speedup with the increase of the number of cores per compute node, a paramount requirement for scalability on future many-core hardware. }, keywords = {Cluster, Collective communication, Hierarchical, HPC, MPI, Multicore}, doi = {10.1016/j.jpdc.2013.01.015}, url = {http://www.sciencedirect.com/science/article/pii/S0743731513000166}, author = {Teng Ma and George Bosilca and Aurelien Bouteiller and Jack Dongarra} } @article {icl:700, title = {HierKNEM: An Adaptive Framework for Kernel-Assisted and Topology-Aware Collective Communications on Many-core Clusters}, journal = {IPDPS 2012 (Best Paper)}, year = {2012}, month = {2012-05}, address = {Shanghai, China}, author = {Teng Ma and George Bosilca and Aurelien Bouteiller and Jack Dongarra} } @article {icl:646, title = {Impact of Kernel-Assisted MPI Communication over Scientific Applications: CPMD and FFTW}, journal = {18th EuroMPI}, year = {2011}, month = {2011-09}, pages = {247-254}, publisher = {Springer}, address = {Santorini, Greece}, keywords = {dague}, author = {Teng Ma and Aurelien Bouteiller and George Bosilca and Jack Dongarra}, editor = {Yiannis Cotronis and Anthony Danalis and Dimitrios S. Nikolopoulos and Jack Dongarra} } @inproceedings {icl:649, title = {Kernel Assisted Collective Intra-node MPI Communication Among Multi-core and Many-core CPUs}, journal = {Int{\textquoteright}l Conference on Parallel Processing (ICPP {\textquoteright}11)}, year = {2011}, month = {2011-09}, address = {Taipei, Taiwan}, author = {Teng Ma and George Bosilca and Aurelien Bouteiller and Brice Goglin and J. Squyres and Jack Dongarra} } @inproceedings {icl:648, title = {Process Distance-aware Adaptive MPI Collective Communications}, journal = {IEEE Int{\textquoteright}l Conference on Cluster Computing (Cluster 2011)}, year = {2011}, month = {2011-00}, address = {Austin, Texas}, author = {Teng Ma and Thomas Herault and George Bosilca and Jack Dongarra} } @techreport {icl:597, title = {Kernel Assisted Collective Intra-node Communication Among Multicore and Manycore CPUs}, journal = {University of Tennessee Computer Science Technical Report, UT-CS-10-663}, year = {2010}, month = {2010-11}, author = {Teng Ma and George Bosilca and Aurelien Bouteiller and Brice Goglin and J. Squyres and Jack Dongarra} } @inproceedings {icl:535, title = {Locality and Topology aware Intra-node Communication Among Multicore CPUs}, journal = {Proceedings of the 17th EuroMPI conference}, year = {2010}, month = {2010-09}, publisher = {LNCS}, address = {Stuttgart, Germany}, author = {Teng Ma and Aurelien Bouteiller and George Bosilca and Jack Dongarra} }