AMITH RAJITH MAMIDALA RNO 1092, 101 Curl Drive, email: mamidala@cse.ohio-state.edu Columbus, Ohio 43210. phone: (614) 209-5018 web: http://www.cse.ohio-state.edu/~mamidala OBJECTIVE To obtain a challenging full-time research position in systems software. My interests lie in improving Performance and Resource scalability of Communication Protocols and Algorithms over modern Interconnects and Multi-core processors. EDUCATION The Ohio State University, Columbus, Ohio PhD, Computer Science and Engineering Winter 2008 (Expected) Indian Institute of Technology, Chennai, India BTech, Computer Science and Engineering May 2002 EXPERIENCE Argonne National Laboratory, Argonne, IL June 2006 - September 2006 Summer Intern, Mathematics and Computer Science Division The Ohio State University, Columbus, OH June 2003 - Present Graduate Research Assistant The Ohio State University, Columbus, OH September 2002 - May 2003 Graduate Teaching Assistant Instructor for C++ PUBLICATIONS * Jiuxing Liu, Amith R Mamidala and Dhabaleswar K Panda. Performance Evaluation of InfiniBand with PCI-Express, IEEE Micro Journal, 2005 * Rahul Kumar, Amith R Mamidala and Dhabaleswar K Panda. Scaling Alltoall Collective on Multicore Systems, Workshop on Communication Architecture for Clusters, IPDPS 2008 * Amith R Mamidala, Rahul Kumar, Debraj De and Dhabaleswar K Panda. MPI Collectives on Modern Multicore Clusters: Performance Optimizations and Communication Characteristics, Int’l Symposium on the Cluster Computing and the Grid (CCGrid), May 2008 * Amith R Mamidala, Sundeep Narravula, Abhinav Vishnu, Gopal K Santhanaraman and Dhabaleswar K Panda. Using Connection-Oriented and Connection-Less transport on Performance and Scalability of Collective and One-sided operations: Trade-offs and Impact, Accepted at ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming, PPoPP 07 * Gopal K Santhanaraman, Sundeep Narravula, Amith R Mamidala, and Dhabaleswar K Panda. MPI-2 One Sided Usage and Implementation for Read Modify Write operations: A case study with HPCC, In Proceedings of EuroPVM/MPI '07, Paris, France * Sundeep Narravula, Amith R Mamidala, Abhinav Vishnu, Karthik Vaidyanathan and Dhabaleswar K Panda. High Performance Distributed Lock Management Services using Network-based Remote Atomic Operations, Int'l Symposium on Cluster Computing and the Grid (CCGrid), Rio de Janeiro - Brazil, May 2007 * Sundeep Narravula, Amith R Mamidala, Abhinav Vishnu and Dhabaleswar K Panda. High Performance MPI over iWARP: Early Experiences, International Conference for Parallel Processing, ICPP 2007 * Abhinav Vishnu, Matt Koop, Adam Moody, Amith R Mamidala, Sundeep Narravula and Dhabaleswar K Panda. Hot-Spot Avoidance With Multi-Pathing Over InfiniBand: An MPI Perspective, Int'l Symposium on Cluster Computing and the Grid (CCGrid), Rio de Janeiro - Brazil, May 2007 * Abhinav Vishnu, Amith R Mamidala, Sundeep Narravula and Dhabaleswar K Panda. Automatic Path Migration over InfiniBand: Early Experiences, Third International Workshop on System Management Techniques, Processes, and Services, to be held in conjunction with IPDPS '07, March 2007 * Abhinav Vishnu, Prachi Gupta, Amith R Mamidala, and Dhabaleswar K Panda. A Software Based Approach for Providing Network Fault Tolerance in Clusters with uDAPL interface: MPI Level Design and Performance Evaluation, Supercomputing, SC 06 * Amith R Mamidala, Abhinav Vishnu and Dhabaleswar K Panda. Efficient Shared Memory and RDMA based design for MPI_Allgather over InfiniBand, EuroPVM/MPI Conference, 2006 * Amith R Mamidala, Lei Chai, Hyun-Wook Jin and Dhabaleswar K Panda. Efficient SMP-Aware MPI-Level Broadcast over InfiniBand’s Hardware Multicast, Workshop on Communication Architecture for Clusters, IPDPS 2006 * Jiuxing Liu, Amith R Mamidala and Dhabaleswar K Panda. Performance Evaluation of InfiniBand with PCI-Express, IEEE Micro, 2005 * Amith R Mamidala, Hyun-Wook Jin and Dhabaleswar K Panda. Efficient Hardware Multicast Group Management for Multiple MPI Communicators over InfiniBand, EuroPVM/MPI Conference 2005 * Sayantan Sur, Uday Bondhugula, Amith R Mamidala, Hyun-Wook Jin and Dhabaleswar K Panda. High Performance RDMA Based All-to-all Broadcast for InfiniBand Clusters, International Conference on High Performance Computing, HiPC 2005 * Abhinav Vishnu, Amith R Mamidala, Hyun-Wook Jin and Dhabaleswar K Panda. Performance Modeling of Subnet Management on Fat Tree InfiniBand Networks using OpenSM, Workshop on System Management Tools for Large-Scale Parallel Systems, IPDPS 2005 * Amith R Mamidala, Jiuxing Liu and Dhabaleswar K Panda. Efficient Barrier and Allreduce on InfiniBand clusters using Hardware Multicast and Adaptive algorithms, International Conference on Cluster Computing, Cluster 2004 * Jiuxing Liu, Amith R Mamidala and Dhabaleswar K Panda. Fast and Scalable MPI-Level Broadcast using InfiniBand’s Hardware Multicast Support in proceedings of International Parallel and Distributed Processing Symposium, IPDPS 2004 * Jiuxing Liu, Amith R Mamidala and Dhabaleswar K Panda. Performance Evaluation of InfiniBand with PCI-Express, Hot Interconnects: Symposium on high performance interconnects, HOTI 2004 * Jiesheng Wu, Amith R Mamidala and Dhabaleswar K Panda. Can NIC memory in InfiniBand Benefit Communication Performance? – A Study with Mellanox Adapter, Technical report, OSU-CISRC-4/04-TR20 PROJECTS MVAPICH Software: * Most widely used Message Passing Communication Software for InfiniBand interconnection network. Used by 3rd ranked supercomputer in TOP 500 list (November 07 ranking) * Designed and Integrated different software modules for Efficient and Scalable means of group communication Communication Protocols: * Understanding message communication within a Cluster of commodity processors interconnected by InfiniBand Network. The salient issues being the trade-offs in performance and memory usage for two modes of transport: Reliable Connection and Unreliable Datagram * Design of Communication substrate for messaging over IWARP (latest RDMA-enabled Ethernet standard) * Design of Scalable and Reliable means of Group Communication (e.g. Multicast) between thousands of processors connected over InfiniBand network Communication Algorithms: * Design of Efficient algorithms for inter-process communication for an imperfectly load balanced application * Exploiting the semantic advantages of RDMA for Group Communication compared to traditional Send/Recv approaches * Using Analytical models to study and capture the behavior of different communication algorithms for Multi-core clusters Data-center Project: * Providing for Distributed Shared Locking support for Data-centers over InfiniBand Shared Memory for Multi-cores: * Design of efficient inter-process communication system within Multi-core processors (e.g. Intel Clovertown) and optimizations for group communication * Using integrated RDMA and Shared Memory mechanism for All-to-all Broadcast of data over InfiniBand * Ongoing work: Cache-conscious message communication for Multi-cores Multi-Pathing: * Exploring automatic path migration mechanism in InfiniBand for providing network-level Fault Tolerance * Designing Congestion Avoidance techniques over InfiniBand Performance Benchmarking: * Construction of a set of micro-benchmarks such as Latency and Bandwidth to evaluate the performance of InfiniBand with PCI-Express I/O interfaces * Comprehensive evaluation of advantages of using memory in the network card for inter-process communication SKILLS Programming Languages: C, C++, Visual C++, Lisp, Shell scripts and MPI-level programming Operating Systems: Linux, Windows (98/NT), DOS and Solaris Other: User Level Protocols for High-Speed networks (InfiniBand, IWARP), CORBA, Java RMI and 80x86 Assembly AWARDS and HONORS * Merit certificate for top 50 students entering IIT-Madras * Placed in top 50 in IIT-JEE Engineering Entrance Examination taken by over 1,00,000 students PROFESSIONAL ACTIVITIES * External reviewer for IEEE TPDS, TKDE journals * External reviewer for IPDPS, SC, ICPP, ICS conferences REFERENCES * Available on Request