Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “parallel, scaling”

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 1,315 records · Page 73

OpenMP and MPI implementations of an elasto-viscoplastic fast Fourier transform-based micromechanical solver for fast crystal plasticity modeling

Here, we explore several parallel implementations of an elasto-viscoplastic fast Fourier transform (EVPFFT) model using Message Passing Interface (MPI), OpenMP, and a hybrid of MPI and OpenMP to efficiently predict micromechanical response of polycrystals. Performance studies using EVPFFT are performed based on domain decomposition over voxels of a periodic cell, which is a representative volume element (RVE) of polycrystalline copper. We begin by parallelizing the computationally intensive Newton–Raphson (NR) single crystal solver within EVPFFT. Next, we compare the performance of the serial and parallel FFTW (Fastest Fourier Transform in the West) using OpenMP (OpenMP-FFTW) and MPI (MPI-FFTW) with the original Numerical Recipes-based FOURN routine within EVPFFT. In the parallel environment, we find that the FFT calculations are best performed using the MPI version of FFTW. Finally, the remainder of the code, except read/write subroutines, is parallelized. Significant speedups of the original EVPFFT model are achieved using MPI on shared memory multicore workstations. Furthermore, results achieved on a distributed memory Cray supercomputer show promising strong and weak scalability and in some cases even super scalability for the single crystal NR solver in EVPFFT. MPI-FFTW also scales perfectly for microstructure RVEs larger than 64 3 FFT voxels. For example, the MPI-EVPFFT parallel version of the code accelerates the simulations for approximately two orders of magnitude using 64 cores over the old serial code for an RVE size of 128 3 . Finally, the parallel EVPFFT code developed in this work can run massive voxel-based microstructural RVEs taking the advantages of thousands of logical cores provided by more advanced clusters.

42 ENGINEERING↗

Measurement of Fast Electron Transport by Lower Hybrid Modulation Experiments in Alcator C-Mod

The Lower Hybrid Current Drive (LHCD) system on Alcator C-Mod can produce spectra with a wide range of peak parallel refractive index (n{sub parallel}). An experiment in which LH power is square-wave modulated on a time scale much faster than the current relaxation time does not significantly alter the poloidal magnetic field inside the plasma and thus allows for realistic modeling and consistent plasma conditions for different ny spectra. Boxcar binning of hard x-rays during LH power modulation allows for time resolution sufficient to resolve the build-up, steady-state, and slowing-down of fast electrons. A transport model built in Matlab has been used to determine a fast electron pinch velocity for a high-n{sub parallel} case of 1-2 m/s.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Scalable Triadic Analysis of Large-Scale Graphs: Multi-Core vs. Multi-Processor vs. Multi-Threaded Shared Memory Architectures

Triadic analysis encompasses a useful set of graph mining methods that is centered on the concept of a triad, which is a subgraph of three nodes and the configuration of directed edges across the nodes. Such methods are often applied in the social sciences as well as many other diverse fields. Triadic methods commonly operate on a triad census that counts the number of triads of every possible edge configuration in a graph. Like other graph algorithms, triadic census algorithms do not scale well when graphs reach tens of millions to billions of nodes. To enable the triadic analysis of large-scale graphs, we developed and optimized a triad census algorithm to efficiently execute on shared memory architectures. We will retrace the development and evolution of a parallel triad census algorithm. Over the course of several versions, we continually adapted the code’s data structures and program logic to expose more opportunities to exploit parallelism on shared memory that would translate into improved computational performance. We will recall the critical steps and modifications that occurred during code development and optimization. Furthermore, we will compare the performances of triad census algorithm versions on three specific systems: Cray XMT, HP Superdome, and AMD multi-core NUMA machine. These three systems have shared memory architectures but with markedly different hardware capabilities to manage parallelism.

Triadic analysis, triads, triad census, large-scal↗

Cross-Field Motion of Plasma Blobs in an Open Magnetic Field Line Configuration

The radial propagation of blobs generated from plasma instabilities is investigated in an open magnetic field line configuration. Blob cross-field velocities and sizes are obtained from internal probe measurements using pattern recognition. By varying the ion mass, the normalized vertical blob scale a-tilde is scanned from a-tilde<1 to a-tilde>1. An analytical expression for the blob velocity including cross-field ion polarization currents, parallel currents to the sheath, and ion-neutral collisions is derived and shows good quantitative agreement with the experimental data. In agreement with previous theoretical studies, this scaling shows that, for a-tilde<1, the blob velocity is limited by cross-field ion polarization currents, while for a-tilde>1 it is limited by parallel currents to the sheath.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Profiling the Usage of an Extreme-Scale Archival Storage System

Profiling the archival storage system in scientific computing environments has received much less attention compared to the parallel file system, but is equally important since it stores the final data products safely, for a long duration. In this paper, we analyze eight years worth of data transfer logs for accessing the archival file system (HPSS) in the Oak Ridge Leadership Computing Facility (OLCF), which has been hosting the world's largest supercomputers and file systems. Our analysis encompasses about 135 million data transfer activities to the 80 PB High Performance Storage System (HPSS), between 2010 and 2017. We analyze the logs from several dimensions, including studying the workload characteristics (e.g., access patterns, frequency of accesses and temporal behavior), file system characteristics (e.g., directory depth, file system scaling trends, file types), and scientific user behavior (e.g., domain-specific usage and organization). Based on the analysis, we derive insights into the future evolution of the archive in terms of provisioning, desired features and functionality from the archive software, role and right sizing of the archive tiers, quota management, and the importance of smart and efficient metadata and storage management. We believe our study will prove useful for both operating current archival storage and better provisioning future systems.

Sim, Hyogi↗

Parallelizing Single Source Shortest Path with OpenSHMEM

Single Source Shortest Path (SSSP) is one of the widely occurring graph problems where the paths are discovered from an origin vertex to all other vertices in the graph. In this paper, we discuss our experience parallelizing SSSP using OpenSHMEM. We start with the serial Dijkstra and Bellman-Ford algorithms, parallelize these algorithms, and adapt them to the Partitioned Global Address Space (PGAS) programming model. We implement the parallel algorithms using OpenSHMEM and introduce a series of optimizations to achieve higher scaling and performance characteristics. The implementation is evaluated on Titan with various graphs including synthetic Recursive Matrix (R-MAT) and small-world network graphs as well as real-world graphs from Facebook, Twitter, LiveJournal, and the road maps of California and Texas.

97 MATHEMATICS AND COMPUTING↗

Simulating the Heterogeneity in Braided Channel Belt Deposits: Part 1. A Geometric-Based Methodology and Code

A geometric-based simulation methodology was developed and incorporated into a computer code to model the hierarchical stratal architecture, and the corresponding spatial distribution of permeability, in braided channel belt deposits. The code creates digital models of these deposits as a three-dimensional cubic lattice, which can be used directly in numerical aquifer or reservoir models for fluid flow. The digital models have stratal units defined from the km scale to the cm scale. These synthetic deposits are intended to be used as high-resolution base cases in various areas of computational research on multiscale flow and transport processes, including the testing of upscaling theories. The input parameters are primarily univariate statistics. These include the mean and variance for characteristic lengths of sedimentary unit types at each hierarchical level, and the mean and variance of log-permeability for unit types defined at only the lowest level (smallest scale) of the hierarchy. The code has been written for both serial and parallel execution. The methodology is described in Part 1 of this series. In Part 2, models generated by the code are presented and evaluated.

99 GENERAL AND MISCELLANEOUS↗

Self-interaction corrected density functional calculations of molecular Rydberg states

A method is presented for calculating the wave function and energy of Rydberg excited states of molecules. A good estimate of the Rydberg state orbital is obtained using ground state density functional theory including Perdew-Zunger self-interaction correction and an optimized effective potential. The total energy of the excited molecule is obtained using the Delta Self-Consistent Field method where an electron is removed from the highest occupied orbital and placed in the Rydberg orbital. Results are presented for the first few Rydberg states of NH{sub 3}, H{sub 2}O, H{sub 2}CO, C{sub 2}H{sub 4}, and N(CH{sub 3}){sub 3}. The mean absolute error in the energy of the 33 molecular Rydberg states presented here is 0.18 eV. The orbitals are represented on a real space grid, avoiding the dependence on diffuse atomic basis sets. As in standard density functional theory calculations, the computational effort scales as NM{sup 2} where N is the number of orbitals and M is the number of grid points included in the calculation. Due to the slow scaling of the computational effort with system size and the high level of parallelism in the real space grid approach, the method presented here makes it possible to estimate Rydberg electron binding energy in large molecules.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Interacting quantum and classical waves: Resonant and non-resonant energy transfer to electrons immersed in an intense electromagnetic wave

Dynamics of electrons subjected to a constant amplitude classical electromagnetic (EM) wave is investigated as a fundamental, representative problem in the physics of interacting quantum and classical waves. In the nonrelativistic regime (electrons as Schrödinger waves), the electron energy acquires a constant and a time dependent part. Driven by EM waves, both parts scale strongly with the amplitude, but we expect no resonant enhancement since the parallel electron “speed” of nonrelativistic electrons could never match the wave phase velocity. In the relativistic regime (electron as a Klein–Gordon wave), however, a class of electron waves (with parallel speed matching the EM phase speed) are resonantly excited to extremely high energies. Such a direct resonant energy transfer from intense electromagnetic waves constitutes a mechanism that could, in principle, power the most energetic of cosmic rays (this mechanism will work on protons just as well). Some predictions of the theory will, hopefully, be tested in laboratory laser experiments. In conclusion, the nonrelativistic calculations will also be examined in the context of recent experiments using photon-induced near-field electron microscopy in detail.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Conformal consistency relations for single-field inflation

We generalize the single-field consistency relations to capture not only the leading term in the squeezed limit — going as 1/q{sup 3}, where q is the small wavevector — but also the subleading one, going as 1/q{sup 2}. This term, for an (n+1)-point function, is fixed in terms of the variation of the n-point function under a special conformal transformation; this parallels the fact that the 1/q{sup 3} term is related with the scale dependence of the n-point function. For the squeezed limit of the 3-point function, this conformal consistency relation implies that there are no terms going as 1/q{sup 2}. We verify that the squeezed limit of the 4-point function is related to the conformal variation of the 3-point function both in the case of canonical slow-roll inflation and in models with reduced speed of sound. In the second case the conformal consistency conditions capture, at the level of observables, the relation among operators induced by the non-linear realization of Lorentz invariance in the Lagrangian. These results mean that, in any single-field model, primordial correlation functions of ζ are endowed with an SO(4,1) symmetry, with dilations and special conformal transformations non-linearly realized by ζ. We also verify the conformal consistency relations for any n-point function in models with a modulation of the inflaton potential, where the scale dependence is not negligible. Finally, we generalize (some of) the consistency relations involving tensors and soft internal momenta.

79 ASTRONOMY AND ASTROPHYSICS↗

Momentum flux parasitic to free-energy transfer

An often-neglected portion of the radial $\boldsymbol{E}\times \boldsymbol{B}$ drift is shown to drive an outward flux of co-current momentum when free energy is transferred from the electrostatic potential to ion parallel flows. This symmetry breaking is fully nonlinear, not quasilinear, necessitated simply by free-energy balance in parameter regimes for which significant energy is dissipated via ion parallel flows. The resulting rotation peaking is counter-current and has a scaling and order of magnitude that are comparable with experimental observations. Finally, the residual stress becomes inactive when frequencies are much higher than the ion transit frequency, which may explain the observed relation of density peaking and counter-current rotation peaking in the core.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Frontiers of Performance Analysis on Leadership-Class Systems

The number of cores in high-end systems for scientific computing are employing is increasing rapidly. As a result, there is an pressing need for tools that can measure, model, and diagnose performance problems in highly-parallel runs. We describe two tools that employ complementary approaches for analysis at scale and we illustrate their use on DOE leadership-class systems.

99 GENERAL AND MISCELLANEOUS↗

Dipolar Antiferromagnetism and Quantum Criticality in LiErF4

Magnetism has been predicted to occur in systems in which dipolar interactions dominate exchange. We present neutron scattering, specific heat, and magnetic susceptibility data for LiErF{sub 4}, establishing it as a model dipolar-coupled antiferromagnet with planar spin-anisotropy and a quantum phase transition in applied field H{sub c{parallel}} = 4.0 {+-} 0.1 kilo-oersteds. We discovered non-mean-field critical scaling for the classical phase transition at the antiferromagnetic transition temperature that is consistent with the two-dimensional XY/h{sub 4} universality class; in accord with this, the quantum phase transition at H{sub c} exhibits three-dimensional classical behavior. The effective dimensional reduction may be a consequence of the intrinsic frustrated nature of the dipolar interaction, which strengthens the role of fluctuations.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Performance tuning of CEED software and 1st and 2nd wave apps

The goal of this milestone was the performance tuning of the CEED software, as well as the use and tuning of CEED to accelerate the first and second wave of targeted ECP applications. In this milestone, the CEED team developed optimization techniques and tuned for performance the CEED software to accelerate the first and second wave target ECP applications. Specifically, the focus was on the following: (1) Efficient use of the memory sub-system for managing data motion and interactions among different physics packages and libraries; this included integration of the new developments in a libCEED 0.5 software release; (2) Enhancing the CEED libraries with GPU and AMD GPU support in close interaction with vendors; (3) Optimal data locality and motion, and enhanced scalability and parallelism; this included vendors interactions for improvements in data motion and making strong scaling easier and more efficient; and (4) Continue boosting performance for the first-wave and second-wave of ECP target applications, including ExaSMR, ExaWind, Urban, MARBL, E3SM, and ExaAM. A main part of this milestone was the performance tuning of the CEED first-wave and second-wave ECP applications. This included the ExaSMR application – Coupled Monte Carlo Neutronics and Fluid Flow Simulation of Small Modular Reactors (ORNL), the MARBL application – Next-Gen Multi-physics Simulation Code (LLNL), and the ExaAM ExaConstit – a miniapp for the Transforming Additive Manufacturing through Exascale Simulation applications project (ExaAM), as well as ExaWind, Urban, and E3SM. The artifacts delivered include performance improvements in CEED’s 1st and 2nd wave of applications, and tuned CEED software for various architectures through a number of backends, freely available in the CEED’s repository on GitHub. See the CEED website, http://ceed.exascaleproject.org and the CEED GitHub organization, http://github.com/ceed for more details. In addition to details and results from the above R&D efforts, in this document we are also reporting on other project-wide activities performed in Q4 of FY19 including: CEED’s third annual meeting, MFEM GPU kernels scaling improvements, libCEED 0.5 and NekRS releases, an overview paper covering latest results on a set of bake-off high-order finite element problems (BPs), and other outreach efforts.

97 MATHEMATICS AND COMPUTING↗

Development of a Traveling-Wave (TW) SRF cavity

Traveling-wave technology can push the accelerator field gradient of niobium SRF cavity to 70MV/m or higher beyond the fundamental limit of 50~60MV/m in Standing-Wave regime. The 1st demonstration of TW resonance excitation in a proof-of-principle 3-cell SRF cavity in 2K liquid helium was successfully carried out at Fermilab in collaboration with Euclid Techlabs. In parallel with that, to advancing TW technologies necessary more for future accelerator-scale one, Fermilab has proposed a preliminary RF design of a half-meter scale TW SRF cavity by considering the physical dimensions of existing SRF facilities and the lessons learned from the 3-cell,. It consists of a 7-cell structure and a power feedback waveguide (WG) loop with new RF configurations to control TW resonance. Development efforts toward a TW 7-cell has been in progress. Here we report the recent activities and progresses of those.

Furuta, F.↗

Final Technical Report

Most prokaryotes of interest to DOE are poorly understood. Even when full genomic sequences are available, the function of only a small number of gene products are clear. The critical question is how to best infer the most probable network architectures in cells that are poorly characterized. The project goal is to create a computational hypothesis testing (CHT) framework that combines large-scale dynamical simulation, a database of bioinformatics-derived probable interactions, and numerical parallel architecture data-fitting routines to explore many “what if ?” hypotheses about the functions of genes and proteins within pathways and their downstream effects on molecular concentration profiles and corresponding phenotypes. From this framework we expect to infer signal transduction pathways and gene expression networks in prokaryotes. Detailed mechanistic models of E. Coli have been developed that directly incorporate DNA sequence information. The CHT framework is implemented in the NIEngine network inference software. NIEngine has been applied to recover gene regulatory networks in E. coli to assess performance. Application to Shewanel la oneidensi and other organism of interest DOE will be conducted in partnership with Jim Collin's Lab at Boston University and other academic partners. The CHT framework has also found broad application in the automated learning of biology for purposes of improving human health.

59 BASIC BIOLOGICAL SCIENCES↗

Module-Fluidics: Building Blocks for Spatio-Temporal Microenvironment Control

Generating the desired solute concentration signal in micro-environments is vital to many applications ranging from micromixing to analyzing cellular response to a dynamic microenvironment. We propose a new modular design to generate targeted temporally varying concentration signals in microfluidic systems while minimizing perturbations to the flow field. The modularized design, here referred to as module-fluidics, similar in principle to interlocking toy bricks, is constructed from a combination of two building blocks and allows one to achieve versatility and flexibility in dynamically controlling input concentration. The building blocks are an oscillator and an integrator, and their combination enables the creation of controlled and complex concentration signals, with different user-defined time-scales. We show two basic connection patterns, in-series and in-parallel, to test the generation, integration, sampling and superposition of temporally-varying signals. All such signals can be fully characterized by analytic functions, in analogy with electric circuits, and allow one to perform design and optimization before fabrication. Such modularization offers a versatile and promising platform that allows one to create highly customizable time-dependent concentration inputs which can be targeted to the specific application of interest.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

4 kW Test of Solid Oxide Electrolysis Stacks with Advanced Electrode-Supported Cells

A new test stand has been developed at the Idaho National Laboratory for multi-kW testing of solid oxide electrolysis stacks. This test stand will initially be operated at the 4 KW scale. The 4 kW tests will include two 60-cell stacks operating in parallel in a single hot zone. The stacks are internally manifolded with an inverted-U flow pattern and an active area of 100 cm2 per cell. Process gases to and from the two stacks are distributed from common inlet/outlet tubing using a custom base manifold unit that also serves as the bottom current collector plate. The solid oxide cells incorporate a negative-electrode-supported multi-layer design with nickel-zirconia cermet negative electrodes, thin-film yttria-stabilized zirconia electrolytes, and multi-layer lanthanum ferrite-based positive electrodes. Treated metallic interconnects with integral flow channels separate the cells and electrode gases. Sealing is accomplished with compliant mica-glass seals. A spring-loaded test fixture is used for mechanical stack compression. Due to the power level and the large number of cells in the hot zone, process gas flow rates are high and heat recuperation is required to preheat the cold inlet gases upstream of the furnace. Heat recuperation is achieved by means of two inconel tube-in-tube counter-flow heat exchangers. A current density of 0.3 A/cm2 will be used for these tests, resulting in a hydrogen production rate of 25 NL/min. Inlet steam flow rates will be set to achieve a steam utilization value of 50%. The 4 kW test will be performed for a minimum duration of 1000 hours in order to document the long-term durability of the stacks. Details of the test apparatus and initial results will be provided.

08 HYDROGEN↗