On May 31st, 2010, Ms. Ping Lai, Dr. Sayantan Sur, and Dr. D. K. Panda will be presented with the ISC Award for the Best Paper submitted for presentation at the 2010 ISC- the HPC Event in Hamburg, Germany. In this work, the team addresses the problems with multi-core processors using the popular MPI (Message Passing Interface) which do not provide support for pure one-sided intra-node RMA (Remote Memory Access) communication. The recent development of MPI-2 RMA allows for efficient one-sided communication for MPI applications. DK, Sayantan, and Ping have worked on a new "Design of Truly One-sided Intra-node Data Transfer using two Kernel-based Direct Copy Alternatives: Basic Kernel-assisted Approach and I/OAT-assisted Approach." This design eliminates the need for using two-sided operations and involvement from the remote side. The paper also proposes a "series of benchmarks" to evaluate various performance aspects of the new design over multi-core architectures (Intel Clovertown, Intel Nehalem and AMD Barcelona). The results show that the new design obtains up to 39% lower latency for small and medium messages and demonstrates 29% improvement in large message bandwidth. Moreover, it provides superior performance in terms of better scalability, reduced cache misses, higher resilience to process skew and increased computation and communication overlap. Finally, up to 10% performance benefit is demonstrated for a real scientific application AWM-Olsen (recently renamed to AWM-ODC)."
The ISC conference typically accepts a maximum of 24 papers for its technical program from the total number of submissions. This paper has been selected to receive the Best Paper Award from this competitively selected set of papers. The ISC Award is sponsored by Gauss Center for Supercomputing (GCS).
Sayantan Sur is a Research Scientist at the Department of Computer Science at The Ohio State University. His research interests include high speed interconnection networks, high performance computing, fault tolerance, and parallel computer architecture. He has published more than 18 papers in major conferences and journals related to these research areas. He is a member of the Network-Based Computing Laboratory lead by Dr. D. K. Panda. He is currently collaborating with National Laboratories and leading InfiniBand and iWARP companies on designing various subsystems of next generation high performance computing platforms. He has contributed significantly to the MVAPICH/MVAPICH2 (High Performance MPI over InfiniBand and 10GigE/iWARP) open-source software packages. The software developed as a part of this effort is currently used by over 1135 organizations in 58 countries. In the past, he has held the position of Post-doctoral researcher at IBM T. J. Watson Research Center, Hawthorne, and Member Technical Staff at Sun Microsystems. Dr. Sur received his Ph.D. degree from The Ohio State University in 2007.
Ping Lai is a Ph.D. candidate at Computer Science & Engineering Department. Her primary research interests include high performance computing, communication protocols, and high performance data-centers. She has published (including co-authored) about 10 papers in journals and conferences related to these research areas. She is a member of the Network-Based Computing Laboratory lead by Professor D. K. Panda.
Dr. Dhabaleswar K. (DK) Panda is a Professor of Computer Science at the Ohio State University. He obtained his Ph.D. in computer engineering from the University of Southern California. DK's research interests include parallel computer architecture, high performance computing, communication protocols, files systems, network-based computing, and Quality of Service. He has published over 270 papers in major journals and international conferences related to these research areas. Dr. Panda is a Fellow of IEEE and a member of ACM and has served on many conference committees and boards in various capacities.
Dr. Panda leads the Network-Based Computing Research Group. Students and staff members of this group are involved in multiple state-of-the-art research projects. The team has been doing extensive research on modern networking technologies including InfiniBand, 10GE/iWARP and RDMA over Enhanced Ethernet (RoCE). Currently the group is collaborating with National Laboratories and leading InfiniBand and 10GE/iWARP and RoCE companies on designing various subsystems of next generation high-end systems. The MVAPICH/MVAPICH2 (High Performance MPI over InfiniBand and iWARP) open-source software project, developed by his research group, is being used by more than 1,135 organizations worldwide (in 58 countries). This software has enabled several InfiniBand clusters (including the 5th and 9th ranked ones) to get into the latest TOP500 ranking of the powerful computer systems and with few exceptions InfiniBand clusters have been included in each biannual list over the past five years. These software packages are also available with the Open Fabrics stack for network vendors (InfiniBand, iWARP and RoCE), server vendors and Linux distributors. Dr. Panda's research is supported by funding from US National Science Foundation, US Department of Energy, Ohio Board of Regents and several industry including Intel, Cisco, SUN, Mellanox, QLogic and NetApp. More details on this project are available at http://mvapich.cse.ohio-state.edu.