Using surrogate-based modeling to predict optimal I/O parameters of applications at the extreme scale
Published in 20th IEEE International Conference on Parallel and Distributed Systems (ICPADS), 2014
Recommended citation: M Matheny, S Herbein, N Podhorszki, S Klasky, M Taufer. Performance impact of I/O on QMCPack simulations at the petascale and beyond. In 20th IEEE International Conference on Parallel and Distributed Systems. 2014 https://ieeexplore.ieee.org/abstract/document/7097855
On petascale systems, the selection of optimal values for I/O parameters without taking into account the I/O size and pattern can cause the I/O time to dominate the simulation time, compromising the application’s scalability. In this paper, we adopt and adapt an engineering method called surrogate-based modeling to efficiently search for the optimal I/O parameter values and accurately predict the associated I/O times at the extreme scale. Our approach allows us to address both the search and prediction in a short time, even when the application’s I/O is large and exhibits irregular patterns.