This talk will focus on the set of features available in MVAPICH2 and MVAPICH2-GDR MPI libraries to exploit overlap of computation and communication on modern clusters. Sample features will include: job start-up, point-to- point operations, RMA operations, kernel-based collectives, and non-blocking collectives (with and without core- direct support). For MVAPICH2-GDR, we will additionally focus on the use of GPU Direct RDMA, kernel-based reduction and datatype operations. Performance benefits of these features will be presented.
DK Panda is a Professor and University Distinguished Scholar of Computer Science and Engineering at the Ohio State University. He has published over 400 papers in the area of high-end computing and networking. The MVAPICH2 (High Performance MPI and PGAS over InfiniBand, Omni-Path, iWARP and RoCE) libraries, designed and developed by his research group (http://mvapich.cse.ohio-state.edu), are currently being used by more than 2,875 organizations worldwide (in 86 countries). More than 451,000 downloads of this software have taken place from the project's site. This software is empowering several InfiniBand clusters (including the 1 st , 9 th , 12 th , 17 th , and 48 th ranked ones) in the TOP500 list. The RDMA packages for Apache Spark, Apache Hadoop and Memcached together with OSU HiBD benchmarks from his group (http://hibd.cse.ohio-state.edu) are also publicly available. These libraries are currently being used by more than 275 organizations in 34 countries. More than 25,300 downloads of these libraries have taken place. A high-performance and scalable version of the Caffe framework is available from https://hidl.cse.ohio-state.edu. Prof. Panda is an IEEE Fellow. More details about Prof. Panda are available at http://www.cse.ohio-state.edu/~panda.
| Attachment | Size |
|---|---|
 dk-overlap_seaconf18.pdf | 2.29 MB |
