Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Computing methodologies”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 703 records · Page 39

AREVA NP's industrial CFD single-phase methodology and applications for nuclear fuel

During the course of the last decade, AREVA NP has developed Computational Fluid Dynamics (CFD) methods capable of accurately predicting flow fields and thermal mixing within fuel bundles, as well as pressure losses and hydrodynamic forces on fuel assembly components such as spacer grids, lower and upper nozzles. The present work summarizes the evolution of AREVA NP's single-phase CFD methodology and the extensive validation efforts, with focus on improving confidence level in CFD results. Currently, it is possible to produce models for large sections of a fuel assembly and produce CFD simulation results with an uncertainty proven to be similar to the corresponding measurement uncertainty. Extensive validation has been conducted against in-house data obtained for a wide range of geometries representative of actual hardware, in multiple test facilities, with various measurement techniques. In addition to in-house validation data, AREVA NP has also exercised its best-practices against international benchmarks; blind test results were in very good agreement with the data. The current high level of confidence in single-phase CFD predictions allows investigating a wide variety of physics in Pressurized Water Reactor (PWR) as well as Boiling Water Reactor (BWR) fuel assemblies. CFD is used as a design and optimization tool as well as a supplement to regular testing to improve the quality of tests and optimize hardware design features for specific targets. CFD played a major role in the design of AREVA NP's latest fuel assembly designs (ATRIUM{sup TM1} 11 and GAIA). In this paper we first illustrate AREVA NP's comprehensive validation database for single-phase CFD (pressure drop, flow field), then provide several examples of applications: virtual testing for parts optimization and support of core operation. (authors)

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

A frequentist approach to computer model calibration

The paper considers the computer model calibration problem and provides a general frequentist solution. Under the framework proposed, the data model is semiparametric with a non-parametric discrepancy function which accounts for any discrepancy between physical reality and the computer model. In an attempt to solve a fundamentally important (but often ignored) identifiability issue between the computer model parameters and the discrepancy function, the paper proposes a new and identifiable parameterization of the calibration problem. It also develops a two-step procedure for estimating all the relevant quantities under the new parameterization. This estimation procedure is shown to enjoy excellent rates of convergence and can be straightforwardly implemented with existing software. For uncertainty quantification, bootstrapping is adopted to construct confidence regions for the quantities of interest. As a result, the practical performance of the methodology is illustrated through simulation examples and an application to a computational fluid dynamics model.

97 MATHEMATICS AND COMPUTING↗

Systemwide Planning with a Branch-and-Price Algorithm for Pavement-Marking Assessment Data Collection via the Mobile Retroreflectivity Unit Routing Model

The visibility of pavement markings is one of the most critical factors for traffic safety, and a periodical assessment plan is crucial for maintaining this function. Traditional assessment methods, such as visual windshield surveys or manual testing using handheld devices, are unsafe, time-consuming, and labor-intensive. In recent years, transportation agencies have begun to adopt the use of mobile retroreflectivity units (MRUs) for condition assessment of pavement markings. MRUs, different from other manual methods, can be utilized to collect large-scale retroreflectivity data in an efficient manner. However, no relevant research has yet proposed a mathematical optimization model for arranging the evaluation schedule and paths of MRUs. This study aims to propose a MRU routing model, and an efficient solution methodology. A branch-and-price algorithm, including column generation and branch-and-bound, was implemented. Computational experiments have been conducted based on actual tasks from the Florida MRU program for validation. In conclusion, results show that the proposed solution methodology with a set partitioning model in this study not only finds the optimal solution for problems with tasks less than 60, but also effectively narrows the solution gap to be within 1.0% for problems with tasks less than 131.

42 ENGINEERING↗

The MCNP6 Analytic Criticality Benchmark Suite

Analytical benchmarks provide an invaluable tool for verifying computer codes used to simulate neutron transport. Several collections of analytical benchmark problems [1-4] are used routinely in the verification of production Monte Carlo codes such as MCNP® [5,6]. Verification of a computer code is a necessary prerequisite to the more complex validation process. The verification process confirms that a code performs its intended functions correctly. The validation process involves determining the absolute accuracy of code results vs. nature. In typical validations, results are computed for a set of benchmark experiments using a particular methodology (code, cross-section data with uncertainties, and modeling) and compared to the measured results from the set of benchmark experiments. The validation process determines bias, bias uncertainty, and possibly additional margins. Verification is generally performed by the code developers, while validation is generally performed by code users for a particular application space. The VERIFICATION_KEFF suite of criticality problems [1,2] was originally a set of 75 criticality problems found in the literature for which exact analytical solutions are available. Even though the spatial and energy detail is necessarily limited in analytical benchmarks, typically to a few regions or energy groups, the exact solutions obtained can be used to verify that the basic algorithms, mathematics, and methods used in complex production codes perform correctly. The present work has focused on revisiting this benchmark suite. A thorough review of the problems resulted in discarding some of them as not suitable for MCNP benchmarking. For the remaining problems, many of them were reformulated to permit execution in either multigroup mode or in the normal continuous-energy mode for MCNP. Execution of the benchmarks in continuous-energy mode provides a significant advance to MCNP verification methods.

97 MATHEMATICS AND COMPUTING↗

GPU Acceleration of Kernel Density Estimators in Monte Carlo Neutron Transport Simulations

Kernel Density Estimators (KDEs) have recently been explored for use in Monte Carlo radiation transport simulations as an alternative to histogram tallies for capturing spatially-resolved quantities such as neutron flux and reaction rates. Histogram tallies suffer from large uncertainties when detailed spatial resolution of the quantity of interest is required; increasing the resolution of the underlying mesh increases the uncertainty in each histogram bin. KDEs obtain estimates of underlying densities on user-defined points, with the uncertainty at those points being independent of the spatial resolution desired since collisions and particle-track segments are allowed to contribute to the score at multiple tally points. Thus, KDEs show potential for obtaining estimates of a smoother spatially-resolved with reduced variance when compared to a histogram. However, this ability that enables KDEs to obtain lower uncertainties per particle history is more computationally expensive than using a traditional histogram bins on a structured mesh. While KDEs are capable of producing lower tally uncertainties compared to histogram binning for a given number of histories, their increased computational cost may make them less efficient. Even so, it is possible to reduce this additional cost by exporting the KDE tally routines onto Graphics Processing Units (GPUs). GPUs are commonly found on high-performance computing machines as a means of increasing performance without adding additional compute nodes. While it is difficult for existing Monte Carlo radiation transport codes to take advantage of GPUs without rewriting large portions of the code, it is possible to leverage GPUs through heterogeneous computing. This work uses a modified version of OpenMC and exports the KDE tally routines onto a GPU to obtain problem-dependent speedups ranging between 1.6 and 5.0 for the problem studied in this paper. While this methodology is applied to KDEs, the same ideas can be applied to other computationally expensive vectorizable portions of Monte Carlo algorithms. (authors)

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

New Version of the MCNP Analytic Criticality Benchmark Suite

Analytical benchmarks provide an invaluable tool for verifying computer codes used to simulate neutron transport. Several collections of analytical benchmark problems are used routinely in the verification of production Monte Carlo codes such as MCNP{sup R}. Verification of a computer code is a necessary prerequisite to the more complex validation process. The verification process confirms that a code performs its intended functions correctly. The validation process involves determining the absolute accuracy of code results vs. nature. In typical validations, results are computed for a set of benchmark experiments using a particular methodology (code, cross-section data with uncertainties, and modeling) and compared to the measured results from the set of benchmark experiments. The validation process determines bias, bias uncertainty, and possibly additional margins. Verification is generally performed by the code developers, while validation is generally performed by code users for a particular application space. The VERIFICATION-KEFF suite of criticality problems was originally a set of 75 criticality problems found in the literature for which exact analytical solutions are available. Even though the spatial and energy detail is necessarily limited in analytical benchmarks, typically to a few regions or energy groups, the exact solutions obtained can be used to verify that the basic algorithms, mathematics, and methods used in complex production codes perform correctly. The present work has focused on revisiting this benchmark suite. A thorough review of the problems resulted in discarding some of them as not suitable for MCNP benchmarking. For the remaining problems, many of them were reformulated to permit execution in either multigroup mode or in the normal continuous-energy mode for MCNP. Execution of the benchmarks in continuous-energy mode provides a significant advance to MCNP verification methods. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

On‐site surrogates for large‐scale calibration

Abstract Motivated by a computer model calibration problem from the oil and gas industry, involving the design of a honeycomb seal, we develop a new Bayesian methodology to cope with limitations in the canonical apparatus stemming from several factors. We propose a new strategy of on‐site design and surrogate modeling for a computer simulator acting on a high‐dimensional input space that, although relatively speedy, is prone to numerical instabilities, missing data, and nonstationary dynamics. Our aim is to strike a balance between data‐faithful modeling and computational tractability in a calibration framework—tailoring the computer model to a limited field experiment. Situating our on‐site surrogates within the canonical calibration apparatus requires updates to that framework. We describe a novel yet intuitive Bayesian setup that carefully decomposes otherwise prohibitively large matrices by exploiting the sparse blockwise structure. Empirical illustrations demonstrate that this approach performs well on toy data and our motivating honeycomb example.

Huang, Jiangeng↗

Computer usage and national energy consumption: Results from a field-metering study

The electricity consumption of miscellaneous electronic loads (MELs) in the home has grown in recent years, and is expected to continue rising. Consumer electronics, in particular, are characterized by swift technological innovation, with varying impacts on energy use. Desktop and laptop computers make up a significant share of MELs electricity consumption, but their national energy use is difficult to estimate, given uncertainties around shifting user behavior. This report analyzes usage data from 64 computers (45 desktop, 11 laptop, and 8 unknown) collected in 2012 as part of a larger field monitoring effort of 880 households in the San Francisco Bay Area, and compares our results to recent values from the literature. We find that desktop computers are used for an average of 7.3 hours per day (median = 4.2 h/d), while laptops are used for a mean 4.8 hours per day (median = 2.1 h/d). The results for laptops are likely underestimated since they can be charged in other, unmetered outlets. Average unit annual energy consumption (AEC) for desktops is estimated to be 194 kWh/yr (median = 125 kWh/yr), and for laptops 75 kWh/yr (median = 31 kWh/yr). We estimate national annual energy consumption for desktop computers to be 20 TWh. National annual energy use for laptops is estimated to be 11 TWh, markedly higher than previous estimates, likely reflective of laptops drawing more power in On mode in addition to greater market penetration. This result for laptops, however, carries relatively higher uncertainty compared to desktops. Different study methodologies and definitions, changing usage patterns, and uncertainty about how consumers use computers must be considered when interpreting our results with respect to existing analyses. Finally, as energy consumption in On mode is predominant, we outline several energy savings opportunities: improved power management (defaulting to low-power modes after periods of inactivity as well as power scaling), matching the rated power of power supplies to computing needs, and improving the efficiency of individual components.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Current Capabilities at SNL for the Integration of Small Modular Reactors onto Smart Microgrids Using Sandia's Smart Microgrid Technology High Performance Computing and Advanced Manufacturing.

Smart grids are a crucial component for enabling the nation’s future energy needs, as part of a modernization effort led by the Department of Energy. Smart grids and smart microgrids are being considered in niche applications, and as part of a comprehensive energy strategy to help manage the nation’s growing energy demands, for critical infrastructures, military installations, small rural communities, and large populations with limited water supplies. As part of a far-reaching strategic initiative, Sandia National Laboratories (SNL) presents herein a unique, three-pronged approach to integrate small modular reactors (SMRs) into microgrids, with the goal of providing economically-competitive, reliable, and secure energy to meet the nation’s needs. SNL’s triad methodology involves an innovative blend of smart microgrid technology, high performance computing (HPC), and advanced manufacturing (AM). In this report, Sandia’s current capabilities in those areas are summarized, as well as paths forward that will enable DOE to achieve its energy goals. In the area of smart grid/microgrid technology, Sandia’s current computational capabilities can model the entire grid, including temporal aspects and cyber security issues. Our tools include system development, integration, testing and evaluation, monitoring, and sustainment.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Design Load Analysis for Wave Energy Converters: Preprint

This study demonstrates a systematic methodology for establishing the design loads of a wave energy converter (WEC). The proposed design load methodology incorporates existing design guidelines, where they exist, and follows a typical design progression; namely, advancing from many, quick, order-of-magnitude accurate, conceptual stage design computations to a few, computationally intensive, high-fidelity, design validation simulations. The goal of the study is to streamline and document this process based on quantitative evaluations of the design loads' accuracy at each design step and consideration for the computational efficiency of the entire design process. For the WEC, loads, and site conditions considered, this study demonstrates an efficient and accurate methodology of evaluating the design loads.

16 TIDAL AND WAVE POWER↗

Adapting Code Verification Methodology to Model Form

Code verification is an essential part of credibility analysis for computational models. It assesses whether the mathematical model is implemented correctly into the code and whether the numerical methods behave consistently, and is done before solution verification and validation. Robust guidance for code verification exists in the literature. However, there is no known, concise guide for selecting the approach based on the model form that also presents an overview of the common elements. This document was written to address this gap as an accessible reference for beginning a code-verification effort.

97 MATHEMATICS AND COMPUTING↗

Petascale Computing Enabling Technologies Project Final Report

The Petascale Computing Enabling Technologies (PCET) project addressed challenges arising from current trends in computer architecture that will lead to large-scale systems with many more nodes, each of which uses multicore chips. These factors will soon lead to systems that have over one million processors. Also, the use of multicore chips will lead to less memory and less memory bandwidth per core. We need fundamentally new algorithmic approaches to cope with these memory constraints and the huge number of processors. Further, correct, efficient code development is difficult even with the number of processors in current systems; more processors will only make it harder. The goal of PCET was to overcome these challenges by developing the computer science and mathematical underpinnings needed to realize the full potential of our future large-scale systems. Our research results will significantly increase the scientific output obtained from LLNL large-scale computing resources by improving application scientist productivity and system utilization. Our successes include scalable mathematical algorithms that adapt to these emerging architecture trends and through code correctness and performance methodologies that automate critical aspects of application development as well as the foundations for application-level fault tolerance techniques. PCET's scope encompassed several research thrusts in computer science and mathematics: code correctness and performance methodologies, scalable mathematics algorithms appropriate for multicore systems, and application-level fault tolerance techniques. Due to funding limitations, we focused primarily on the first three thrusts although our work also lays the foundation for the needed advances in fault tolerance. In the area of scalable mathematics algorithms, our preliminary work established that OpenMP performance of the AMG linear solver benchmark and important individual kernels on Atlas did not match the predictions of our simple initial model. Our investigations demonstrated that a poor default memory allocation mechanism degraded performance. We developed a prototype NUMA library to provide generic mechanisms to overcome these issues, resulting in significantly improved OpenMP performance. After additional testing, we will make this library available to all users, providing a simple means to improve threading on LLNL's production Linux platforms. We also made progress on developing new scalable algorithms that target multicore nodes. We designed and implemented a new AMG interpolation operator with improved convergence properties for very low complexity coarsening schemes. This implementation will also soon be available to LLNL's application teams as part of the hypre library. We presented results for both topics in an invited plenary talk entitled 'Efficient Sparse Linear Solvers for Multi-Core Architectures' at the 2009 HPCMP Institutes Annual Meeting/CREATE Annual All-Hands Meeting. The interpolation work was summarized in a talk entitled 'Improving Interpolation for Aggressive Coarsening' at the 14th Copper Mountain Conference on Multigrid Methods and in a research paper that will appear in Numerical Linear Algebra with Applications. In the area of code correctness, we significantly extended our behavior equivalence class identification mechanism. Specifically, we not only demonstrated it works well at very large scales but we also added the ability to classify MPI tasks not only by function call traces, but also by specific call sites (source code line numbers) being executed by tasks. More importantly, we developed a new technique to determine relative logical execution progress of tasks in the equivalence classes by combining static analysis with our original dynamic approach. We applied this technique to a correctness issue that arose at 4096 tasks during the development of the new AMG interpolation operator discussed above. This scale isat the limit of effectiveness of production tools, but our technique quickly located the erroneous source code, demonstrating the power of understanding relationships between behavioral equivalence classes. This work is the subject of a paper recently accepted to SC09, as well as a presentation entitled 'Providing Order to Extreme Scale Debugging Chaos' given at the ParaDyn Week annual conference in College Park, MD. In addition to this theoretical extension, we have made significant progress in developing a front end for this tool set, and the front-end is now available on several of LLNL's largescale computing resources. In addition, we explored mechanisms to identify exact locations of erroneous MPI usage in application source code. In this work, we developed a new model that led to a highly efficient algorithm for detecting deadlock during dynamic software testing. This work was the subject of a well-received paper at ICS 2009 [4].

99 GENERAL AND MISCELLANEOUS↗

TAOI B- Computational Microstructural Optimization Design Tool for High Temperature Structural Materials

The objectives of this research were two-fold: (a) develop a methodology for microstructural optimization of alloys - genetic algorithm approach for alloy microstructural optimization using theoretical models based on fundamental micro-mechanisms, and (b) develop a new computationally designed Ni-Cr alloy for coal-fired power plant applications. The broader outcome of these objectives is expected to be creation of an integrated approach for ‘structural materials by microstructural design’. Three alloy systems were considered for computational optimization and validation, (i) Ni-20Cr (wt.%) base alloy using only solid solution strengthening, (ii) nano-Y 2 O 3 containing Ni-20Cr-1.2Y 2 O 3 (wt.%) alloy for dispersion strengthening and (iii) a sub-micron Al 2 O 3 for composite strengthening, Ni-20Cr-1.2Y 2 O 3 -5.0Al 2 O 3 (wt.%). The specimens were synthesized by mechanical alloying and consolidated using spark plasma sintering. Detailed microstructural characterization was done along with initial mechanical properties to validate the computational prediction. A key target property is to have creep rate of 1x10 -9 s -1 at 100 MPa and 800°C. The initial results were quite promising and require additional quantification of strengthening contributions from dislocation-particle attractive interaction and load transfer. The observed creep rate was in order of 10 -9 s -1 for longer time creep test of Ni-20Cr -1.2Y 2 O 3 -5Al 2 O 3 , lending support to the overall approach pursued in this project.

36 MATERIALS SCIENCE↗

Nonlinear simulation of under-resolved flows with shocks

Here, we consider the numerical simulation of advection-dominated flows whose wide range of physical length scales exceed the memory capacity of finite computers. Simulating flows with shocks and turbulence presented challenges for the earliest computers that were quickly overcome by the development of new numerical methodology. Principal among those new ideas were artificial viscosity and finite volume methods, concepts that remain in common use today. We begin by describing the history of those methods, the innovators and their motivations. We then describe the development of finite scale theory, a reformulation of Navier–Stokes theory that exposes the physical principles on which artificial viscosity is based. We discuss the essential properties of the finite scale equations, the observer, unresolved kinetic energy and inviscid energy dissipation. We briefly consider the implementation of the finite scale equations on the computer from the point of view of Gisin’s conjectures about finite information.

97 MATHEMATICS AND COMPUTING↗

Design and execution of a verification, validation, and uncertainty quantification plan for a numerical model of left ventricular flow after LVAD implantation

Left ventricular assist devices (LVADs) are implantable pumps that act as a life support therapy for patients with severe heart failure. Despite improving the survival rate, LVAD therapy can carry major complications. Particularly, the flow distortion introduced by the LVAD in the left ventricle (LV) may induce thrombus formation. While previous works have used numerical models to study the impact of multiple variables in the intra-LV stagnation regions, a comprehensive validation analysis has never been executed. The main goal of this work is to present a model of the LV-LVAD system and to design and follow a verification, validation and uncertainty quantification (VVUQ) plan based on the ASME V&V40 and V&V20 standards to ensure credible predictions. The experiment used to validate the simulation is the SDSU cardiac simulator, a bench mock-up of the cardiovascular system that allows mimicking multiple operation conditions for the heart-LVAD system. The numerical model is based on Alya, the BSC’s in-house platform for numerical modelling. Alya solves the Navier-Stokes equation with an Arbitrary Lagrangian-Eulerian (ALE) formulation in a deformable ventricle and includes pressure-driven valves, a 0D Windkessel model for the arterial output and a LVAD boundary condition modeled through a dynamic pressure-flow performance curve. The designed VVUQ plan involves: (a) a risk analysis and the associated credibility goals; (b) a verification stage to ensure correctness in the numerical solution procedure; (c) a sensitivity analysis to quantify the impact of the inputs on the four quantities of interest (QoIs) (average aortic root flow $Q^{avg}_{Ao}$, maximum aortic root flow $Q^{max}_{Ao}$, average LVAD flow $Q^{avg}_{VAD}$, and maximum LVAD flow $Q^{max}_{VAD}$; (d) an uncertainty quantification using six validation experiments that include extreme operating conditions. Numerical code verification tests ensured correctness of the solution procedure and numerical calculation verification showed a grid convergence index (GCI)95% <3.3%. The total Sobol indices obtained during the sensitivity analysis demonstrated that the ejection fraction, the heart rate, and the pump performance curve coefficients are the most impactful inputs for the analysed QoIs. The Minkowski norm is used as validation metric for the uncertainty quantification. It shows that the midpoint cases have more accurate results when compared to the extreme cases. The total computational cost of the simulations was above 100 [core-years] executed in around three weeks time span in Marenostrum IV supercomputer. This work details a novel numerical model for the LV-LVAD system, that is supported by the design and execution of a VVUQ plan created following recognised international standards. We present a methodology demonstrating that stringent VVUQ according to ASME standards is feasible but computationally expensive.

59 BASIC BIOLOGICAL SCIENCES↗

Position Papers for the ASCR Workshop on the Science of Scientific-Software Development and Use

Software is an increasingly important component in the pursuit of scientific discovery. Both its development and use are essential activities for many scientific teams. At the same time, very little scientific study has been conducted to understand, characterize, and improve the development and use of software for science. Computational science teams have diversified over time to include contributions from domain scientists who provide expertise in scientific and engineering disciplines, applied mathematicians and computer scientists who provide optimal algorithms and data structures, and software and data engineers who provide methodologies and tools adapted and adopted from other software domains. These diverse contributions have enabled tremendous advances in the pursuit of scientific discovery, even as models, computer architectures, and software environments have become more complicated. With this increasing diversity, we believe the next opportunity for qualitative improvement comes from applying the scientific method to understanding, characterizing, and improving how scientific software is developed and used. We believe that this pursuit requires expertise from computational scientists themselves, and from the cognitive and social sciences as well as the software engineering research community. As we look to increase the productivity and sustainability of the scientific-software-development-and-use cycle, a more systematic application of the scientific method to understand processes for software development and use will be a valuable tool to guide future work and result in more usable and sustainable software. This workshop will bring together computer scientists, software engineering researchers, computational scientists, applied mathematicians, social scientists, cognitive scientists, and others, to explore how we can conduct such systematic investigations, what can be learned, and how doing so will benefit the scientific enterprise. The workshop will be structured around a set of breakout sessions, with every attendee expected to participate actively in the discussions. Afterward, workshop attendees — from DOE, industry, and academia — will produce a report for ASCR that summarizes the findings of the workshop.

42 ENGINEERING↗

Investigating the influence of particle distribution on force and torque statistics using hierarchical machine learning

An accurate representation of hydrodynamic force and torque experienced by every particle in a distribution can be obtained from particle resolved (PR) simulations. These unique quantities are influenced by the deterministic position of surrounding particles. However, systems simulated with this methodology are typically limited to particles due to the involved computational cost. This resource requirement is a major bottleneck in analyzing the effect of variations in particle distribution. Here, this article attempts to address this bottleneck by availing relatively inexpensive deep learning models. The surrogate models that we employ in this article use a physics‐based hierarchical framework and symmetry‐preserving neural networks to achieve robustness with limited training data. This article first performs additional generalizability tests on PR data of distinct distributions that are not involved in the training process. The models are then deployed on several different particle distributions. Impact of clustering and structure on the observed statistics are investigated.

42 ENGINEERING↗

Performance Tuning of Fock Matrix and Two-Electron Integral Calculations for NWChem on Leading HPC Platforms

Attaining performance in the evaluation of two-electron repulsion integrals and constructing the Fock matrix is of considerable importance to the computational chemistry community. Due to its numerical complexity improving the performance behavior across a variety of leading supercomputing platforms is an increasing challenge due to the significant diversity in high-performance computing architectures. In this paper, we present our successful tuning methodology for these important numerical methods on the Cray XE6, the Cray XC30, the IBM BG/Q, as well as the Intel Xeon Phi. Our optimization schemes leverage key architectural features including vectorization and simultaneous multithreading, and results in speedups of up to 2.5x compared with the original implementation.

Shan, Hongzhan↗