The demand for the ability to easily customize, reproduce and migrate applications and workflows has been steadily increasing among the HPC community as software environments and applications grow in complexity. Lightweight containers that are suitable for HPC applications at scale are considered to be a viable approach to meet this demand. Previous studies have addressed the performance aspects of most existing containers using microbenchmarks and revealed the performance overheads of the best implementations to be small. However, the feasibility of providing containerized, real-world HPC applications on HPC systems, and the impact on overall application performance at scale, has not yet been explored. Here we present a basic feasibility and performance study using the Singularity container. We evaluate what is required to enable container images to utilize the high-speed fabric present on most HPC systems and explore their performance by comparing real-world applications run both within a container and in the absence of a container (natively). The results indicate lightweight Singularity images are a promising approach to the HPC communities demands for not only customizability, reproducibility and portability, but also performance.
19 July 2016 in Bulletin of the Seismological Society of America
This paper describes a generalization of the iterative deconvolution method commonly used as a component of passive array wavefield imaging. We show that the iterative method should be thought of as a sparse output deconvolution method with the number of terms retained dependent on the convergence criteria. The generalized method we introduce uses an inverse operator to shape the assumed wavelet to a peaked function at zero lag. We show that the conventional method is equivalent to using a damped least-squares spiking filter with extremely large damping and proper scaling. In that case, the inverse operator used in the generalized method reduces to the cross-correlation operator. The theoretical insight of realizing the output is a sparse series provides a basis for the second important addition of the generalized method—an output shaping wavelet. A constant output shaping wavelet is a critical component in scattered wave imaging to avoid mixing data of variable bandwidth. We demonstrate the new approach can improve resolution by using an inverse operator tuned to maximize resolution. We also show that the signal-to-noise ratio of the result can be improved by applying a different convergence criterion than the standard method, which measures the energy left after each iteration. The efficacy of the approach was evaluated with synthetic experiment in various signal and noise conditions. We further validated the approach with real data from the USArray. We compared our results with data from the EarthScope Automated Receiver Survey and found that our results show modest improvements in consistency measured by correlation coefficients with station stacks and a reduced number of outliers.