VSC School Seminar: Probabilistic Performance Modeling of Asynchronous Iterative Solvers

    Abstract:

    Asynchronous implementations present a mechanism to improve the parallel performance of iterative linear system solution methods on highly parallel computing platforms by removing the synchronization overhead. In this talk, after a brief overview of asynchronous computations, we present a performance modeling approach for hybrid parallel implementations of asynchronous methods. Then, we consider a class of asynchronous iterative linear system solvers that employ randomization to determine the component update orders, specifically focusing on the effects of non-uniform distributions. Results show that using such distributions may lead to a faster convergence than that when components are selected for updates uniformly.

    Speaker:

    Prof. Dr. Masha Sosonkina
    Modeling and Simulation Engineering, Old Dominion University, VA, USA

    Bio: Masha Sosonkina received her Ph.D. in Computer Science and Applications from Virginia Tech. Upon graduation, she worked as a faculty of the University of Minnesota Duluth. During 2003-2012, Dr. Sosonkina was a scientist at the U.S. Department of Energy Ames Laboratory. She is now a professor of Modeling and Simulation Engineering at Old Dominion University.
    Dr. Sosonkina's research interests include high-performance computing, large-scale simulations, parallel numerical algorithms, performance analysis, and adaptive algorithms.

    Language:

    English

    Date, Time, and Location:

    27.05.2019, 14:00 - 15:00, Seminarraum CD0520 (TU Wien, Campus Gußhaus, Gußhausstraße 27-29, Stiege 1, 5th floor)

    Registration:

    Registration for this event is not necessary.

     

    Training events of VSC:

    Opens external link in new windowvsc.ac.at/training