Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallel processing”

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 1,063 records · Page 59

Block-Parallel Data Analysis with DIY2

DIY2 is a programming model and runtime for block-parallel analytics on distributed-memory machines. Its main abstraction is block-structured data parallelism: data are decomposed into blocks; blocks are assigned to processing elements (processes or threads); computation is described as iterations over these blocks, and communication between blocks is defined by reusable patterns. By expressing computation in this general form, the DIY2 runtime is free to optimize the movement of blocks between slow and fast memories (disk and flash vs. DRAM) and to concurrently execute blocks residing in memory with multiple threads. This enables the same program to execute in-core, out-of-core, serial, parallel, single-threaded, multithreaded, or combinations thereof. This paper describes the implementation of the main features of the DIY2 programming model and optimizations to improve performance. DIY2 is evaluated on benchmark test cases to establish baseline performance for several common patterns and on larger complete analysis codes running on large-scale HPC machines.

97 MATHEMATICS AND COMPUTING↗

Roll-to-Roll SOFC Manufacturing for Low Cost Energy Generation

Oak Ridge National Laboratory (ORNL) and Redox Power Systems, LLC. collaborated on this project to develop an integrated, continuous roll-to-toll (R2R) manufacturing processes to produce solid oxide fuel cells (SOFCs). SOFCs are ceramic-based electrochemical devices that efficiently produce electricity from hydrogen-containing fuels, including natural gas. Standard processes for manufacturing SOFCs include tape casting of the anode and electrolyte, followed by stacking of the layers, and lamination using a uniaxial press or isostatic press. After high temperature firing, the cathode and electrical contact are screen printed and fired once again to complete the cell. This effort aimed to further develop R2R tape fabrication processes for green cell fabrication with the DOE Battery Manufacturing Facility (BMF) at ORNL as next generation improvements to current methods, focusing on R2R calendering and slot die coating for lower cost, high volume approaches for multilayer green cell fabrication. Two paths were investigated in parallel in this project. The first path was to optimize R2R processing using traditional tape cast materials from the Redox production chain on the BMF’s heated calender for compressing and laminating multilayer materials. The second path was to use BMF’s pilot-scale slot-die coater for R2R manufacturing as an alternative to tape casting of some or all of the layers in the green cell. Processes have been developed and and process conditions have been identified. Additionally, basic process economics were estimated for comparison to Redox current manufacturing approaches for the applicable process modules. The SBV program results will enable Redox to advance lower cost R2R manufacturing processes to produce millions of green cells per year on a single production line. As a result of this work, one invention disclosure has been submitted. In addition, one paper is under preparation for submission to a peer-reviewed journal.

30 DIRECT ENERGY CONVERSION↗

Scalable Deep-Learning-Accelerated Topology Optimization for Additively Manufactured Materials

Topology optimization (TO) is a popular and powerful computational approach for designing novel structures, materials, and devices. Two computational challenges have limited the applicability of TO to a variety of industrial applications. First, a TO problem often involves a large number of design variables to guarantee sufficient expressive power. Second, many TO problems require a large number of expensive physical model simulations, and those simulations cannot be parallelized. To address these issues, we propose a general scalable deep-learning (DL) based TO framework, referred to as SDL-TO, which utilizes parallel schemes in high performance computing (HPC) to accelerate the TO process for designing additively manufactured (AM) materials. Unlike the existing studies of DL for TO, our framework accelerates TO by learning the iterative history data and simultaneously training on the mapping between the given design and its gradient. The surrogate gradient is learned by utilizing parallel computing on multiple CPUs incorporated with a distributed DL training on multiple GPUs. The learned TO gradient enables a fast online update scheme instead of an expensive update based on the physical simulator or solver. Using a local sampling strategy, we achieve to reduce the intrinsic high dimensionality of the design space and improve the training accuracy and the scalability of the SDL-TO framework. The method is demonstrated by benchmark examples and AM materials design for heat conduction. The proposed SDL-TO framework shows competitive performance compared to the baseline methods but significantly reduces the computational cost by a speed up of around 8.6x over the standard TO implementation.

Bi, Sirui↗

Load Balancing of Dynamical Nucleation Theory Monte Carlo Simulations through Resource Sharing Barriers

The dynamical nucleation theory Monte Carlo (DNTMC) application from the NWChem computational chemistry suite utilizes a Markov chain Monte Carlo, two-level parallel structure, with periodic synchronization points that assemble the results of independent finer-grained calculations. Like many such applications, the existing code employs a static partitioning of processes into groups and assigns each group a piece of the finer-grained parallel calculation. A significant cause of performance degradation is load imbalance among groups since the time requirements of the inner-parallel calculation varies widely with the input problem and as a result of the Monte Carlo simulation. We present a novel approach to load balancing such calculations with minimal changes to the application. We introduce the concept of a resource sharing barrier (RSB) – a barrier that allows process groups waiting on other processes’ work to actively contribute to their completion. The RSB load balancing technique is applied to the production DNTMC application code, resulting in a small code change of 200 lines and a reduction in execution time of up to 37%.

Arafat, Md H.↗

Development of PFLOTRAN Transport Capability for Use in the Waste Isolation Pilot Plant Performance Assessment - 20545

Waste Isolation Pilot Plant (WIPP) performance assessment (PA) calculations estimate the probability of radionuclide release from the repository to the land surface and across the land withdrawal boundary for a regulatory period of 10,000 years after facility closure. Simulations of flow and transport in the repository and the surrounding Salado Formation are foundational to the PA. Because proposed additional waste emplacement panels would result in an asymmetric repository layout, the US Department of Energy (DOE) is preparing to transition to use of a three-dimensional (3-D) model domain for simulation of flow and transport instead of the two-dimensional (2-D) flared grid domain currently used. DOE has charged Sandia National Laboratories with developing the capability necessary to simulate processes affecting flow and transport in the WIPP in PFLOTRAN, an open-source massively parallel multi-phase flow and reactive transport code. The new flow and transport capabilities developed in PFLOTRAN incorporate WIPP-specific process models and will replace the 2-D simulators (BRAGFLO and NUTS) that are currently utilized for Salado flow and transport calculations in WIPP PA. The focus of this paper is on the development of a new Nuclear Waste Transport (NWT) mode in PFLOTRAN that has all of the capabilities necessary for Salado transport simulations, including the ability to handle complete dry-out (100% gas saturation) of arbitrary cells in the model domain, radionuclide mass conservation at step changes in porosity associated with borehole intrusion, and the ability to calculate fluxes on a flared grid. The new PFLOTRAN transport capability and a suite of verification tests were designed around a list of functional requirements for WIPP PA calculations. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Ionisation of a quantum dot by electric fields

We have derived analytical formulas for differential and total ionisation probabilities of a two-dimensional quantum dot by a constant electric field. In the adiabatic approximation, we have calculated the probability of this process in the field of a plane electromagnetic wave and in a superposition of constant and alternating electric fields. The imaginary-time method is used to obtain the momentum distribution of the ionisation probability of a bound system by an intense field generated by a superposition of parallel constant and alternating electric fields. The total probability of the process per unit time is calculated with exponential accuracy. The dependence of the results obtained on the characteristic parameters of the problem is investigated. (laser applications and other topics in quantum electronics)

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Adjusting process count on demand for petascale global optimization

There are many challenges that need to be met before efficient and reliable computation at the petascale is possible. Many scientific and engineering codes running at the petascale are likely to be memory intensive, which makes thrashing a serious problem for many petascale applications. One way to overcome this challenge is to use a dynamic number of processes, so that the total amount of memory available for the computation can be increased on demand. This paper describes modifications made to the massively parallel global optimization code pVTdirect in order to allow for a dynamic number of processes. In particular, the modified version of the code monitors memory use and spawns new processes if the amount of available memory is determined to be insufficient. The primary design challenges are discussed, and performance results are presented and analyzed.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Optimizing Excited-State Electronic-Structure Codes for Intel Knights Landing: A Case Study on the BerkeleyGW Software

We profile and optimize calculations performed with the BerkeleyGW code on the Xeon-Phi architecture. BerkeleyGW depends both on hand-tuned critical kernels as well as on BLAS and FFT libraries. We describe the optimization process and performance improvements achieved. We discuss a layered parallelization strategy to take advantage of vector, thread and node-level parallelism. We discuss locality changes (including the consequence of the lack of L3 cache) and effective use of the on-package high-bandwidth memory. We show preliminary results on Knights-Landing including a roofline study of code performance before and after a number of optimizations. We find that the GW method is particularly well-suited for many-core architectures due to the ability to exploit a large amount of parallelism over plane-wave components, band-pairs, and frequencies.

97 MATHEMATICS AND COMPUTING↗

What Hinders Electron Transfer Dissociation (ETD) of DNA Cations?

Radical activation methods, such as electron transfer dissociation (ETD), produce structural information complementary to collision-induced dissociation. Herein, electron transfer dissociation of 3-fold protonated DNA hexamers was studied to gain insight into the fragmentation mechanism. The fragmentation patterns of a large set of DNA hexamers confirm cytosine as the primary target of electron transfer. The reported data reveal backbone cleavage by internal electron transfer from the nucleobase to the phosphate linker leading either to a•/w or d/z• ion pairs. This reaction pathway contrasts with previous findings on the dissociation processes after electron capture by DNA cations, suggesting multiple, parallel dissociation channels. However, all these channels merely result in partial fragmentation of the precursor ion because the charge-reduced DNA radical cations are quite stable. Two hypotheses are put forward to explain the low dissociation yield of DNA radical cations: it is either attributed to non-covalent interactions between complementary fragments or to the stabilization of the unpaired electron in stacked nucleobases. MS{sup 3} experiments suggest that the charge-reduced species is the intact oligonucleotide. Moreover, introducing abasic sites significantly increases the dissociation yield of DNA cations. Consequently, the stabilization of the unpaired electron by π–π-stacking provides an appropriate rationale for the high intensity of DNA radical cations after electron transfer. .

74 ATOMIC AND MOLECULAR PHYSICS↗

Maximization of performance in multi-tube latent heat storage – Optimization of fins topology, effect of materials selection and flow arrangements

This paper addresses the need of multi-tube latent heat thermal storage (LHTES) systems with enhanced heat transfer performance. Uniquely, this work draws from topology optimization method for thermal energy storage to search for the optimal configuration of fins in multi-tube LHTES systems with different phase change materials (PCMs), flow arrangements and design constraints. The design freedom of topology optimization allows the discovery of innovative LHTES designs and elucidate the link between design and physical processes occurring during charging/discharging. Three key results of this study are: i) the optimized fin design is tightly connected to the type of storage duty cycle, which demonstrates the necessity to account for realistic operating conditions in the optimization process. ii) The fin material should be chosen in parallel with the layout of the fins and not sequentially as commonly done; this indicates that the optimization of LHTES systems is a co-design challenge. iii) Topology optimized multi-tube LHTES units surpass in performance fins optimized for a single-tube configuration in a multi-tube unit. Finally, this work demonstrates for the first time the manufacturability of topology-optimized LHTES units by using 3D printing.

25 ENERGY STORAGE↗

High-fidelity wind farm simulation methodology with experimental validation

The complexity and associated uncertainties involved with atmospheric-turbine-wake interactions produce challenges for accurate wind farm predictions of generator power and other important quantities of interest (QoIs), even with state-of-the-art high-fidelity atmospheric and turbine models. A comprehensive computational study was undertaken with consideration of simulation methodology, parameter selection, and mesh refinement on atmospheric, turbine, and wake QoIs to identify capability gaps in the validation process. For neutral atmospheric boundary layer conditions, the massively parallel large eddy simulation (LES) code Nalu-Wind was used to produce high-fidelity computations for experimental validation using high-quality meteorological, turbine, and wake measurement data collected at the Department of Energy/Sandia National Laboratories Scaled Wind Farm Technology (SWiFT) facility located at Texas Tech University’s National Wind Institute. The wake analysis showed the simulated lidar model implemented in Nalu-Wind was successful at capturing wake profile trends observed in the experimental lidar data.

17 WIND ENERGY↗

Structural basis of GDP release and gating in G protein coupled Fe[superscript 2+] transport

G proteins are key molecular switches in the regulation of membrane protein function and signal transduction. The prokaryotic membrane protein FeoB is involved in G protein coupled Fe{sup 2+} transport, and is unique in that the G protein is directly tethered to the membrane domain. Here, we report the structure of the soluble domain of FeoB, including the G protein domain, and its assembly into an unexpected trimer. Comparisons between nucleotide free and liganded structures reveal the closed and open state of a central cytoplasmic pore, respectively. In addition, these data provide the first observation of a conformational switch in the nucleotide-binding G5 motif, defining the structural basis for GDP release. From these results, structural parallels are drawn to eukaryotic G protein coupled membrane processes.

59 BASIC BIOLOGICAL SCIENCES↗

Analysis of laser-induced modification of collagen structure using nonlinear optical microscopy

Nonlinear optical microscopy is used to monitor changes in the collagen structure of patellar ligament after IR laser heating up to a temperature of 60 °C. Pre-denaturation modification of the tissue matrix exhibits specific properties and consists in splitting of initially parallel fibre bundles into array of divergent fibres. Image processing employing Fast Fourier transform analysis is used for characterising images of the fibre structure and determining the collagen orientation index. An increase in manifestation of laser-induced disordering with an increase in IR laser radiation power from 0.3 to 2 W is accompanied by a decrease in the collagen orientation index. The specific disorganisation of the collagen tissue structure during laser heating is supposed to be due to fast speed and local action of the laser heating, which leads to mechanical distortions of the structural elements of tissue matrix. (paper)

36 MATERIALS SCIENCE↗

Quantum fusion of domain walls with fluxes

We study how fluxes on the domain wall world volume modify quantum fusion of two distant parallel domain walls into a composite wall. The elementary wall fluxes can be separated into parallel and antiparallel components. The parallel component affects neither the binding energy nor the process of quantum merger. The antiparallel fluxes, instead, increase the binding energy and, against naive expectations, suppress quantum fusion. In the small-flux limit we explicitly find the bounce solution and the fusion rate as a function of the flux. We argue that at large (antiparallel) fluxes there exists a critical value of the flux (versus the difference in the wall tensions), which switches off quantum fusion altogether. This phenomenon of flux-related wall stabilization is rather peculiar: it is unrelated to any conserved quantity. Our consideration of the flux-related wall stabilization is based on substantiated arguments that fall short of complete proof.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Adaptive Spatially Aware I/O for Multiresolution Particle Data Layouts

Large-scale simulations on nonuniform particle distributions that evolve over time are widely used in cosmology, molecular dynamics, and engineering. Such data are often saved in an unstructured format that neither preserves spatial locality nor provides metadata for accelerating spatial or attribute subset queries, leading to poor performance of visualization tasks. Furthermore, the parallel I/O strategy used typically writes a file per process or a single shared file, neither of which is portable or scalable across different HPC systems. We present a portable technique for scalable, spatially aware adaptive aggregation that preserves spatial locality in the output. We evaluate our approach on two supercomputers, Stampede2 and Summit, and demonstrate that it outperforms prior approaches at scale, achieving up to 2.5× faster writes and reads for nonuniform distributions. Furthermore, the layout written by our method is directly suitable for visual analytics, supporting low-latency reads and attribute-based filtering with little overhead.

Usher, Will↗

Rapid Characterization and Statistical Analysis of High-Volume Field-Harvested Photovoltaic Connectors

Photovoltaic (PV) installations heavily depend on connectors for efficient module and string interconnections without requiring skilled labor. Yet this seemingly innocuous component of PV systems is a leading cause of module failures, multiple high-profile fires, and lawsuits in the PV industry. This work aims to answer critical questions regarding why connectors fail and the contributing factors to their failure. The study involves collecting and analyzing more than 17,000 field-harvested connectors from various solar installations across the United States. The vast dataset, which includes connector metadata, visual inspections, and resistance measurements, provides unprecedented insight into the state of health of PV connectors across the US, including the geographic locations, connector types, and installation practices most prone to failures. The work presented here describes a novel rapid characterization method for processing large numbers of connectors and is supported by parallel forensic analysis to discern the root causes of failures as well as a levelized cost of lifetime model to determine the economic ramifications of connector failure. Ultimately, the findings may inform PV developers about the best practices to extend connector longevity and lead to more resilient and reliable PV systems.

connectors↗

Communication Characterization and Optimization of Applications Using Topology-Aware Task Mapping on Large Supercomputers

On large supercomputers, the job scheduling systems may assign a non-contiguous node allocation for user applications depending on available resources. With parallel applications using MPI (Message Passing Interface), the default process ordering does not take into account the actual physical node layout available to the application. This contributes to non-locality in terms of physical network topology and impacts communication performance of the application. In order to mitigate such performance penalties, this work describes techniques to identify suitable task mapping that takes the layout of the allocated nodes as well as the application's communication behavior into account. During the first phase of this research, we instrumented and collected performance data to characterize communication behavior of critical US DOE (United States - Department of Energy) applications using an augmented version of the mpiP tool. Subsequently, we developed several reordering methods (spectral bisection, neighbor join tree etc.) to combine node layout and application communication data for optimized task placement. We developed a tool called mpiAproxy to facilitate detailed evaluation of the various reordering algorithms without requiring full application executions. This work presents a comprehensive performance evaluation (14,000 experiments) of the various task mapping techniques in lowering communication costs on Titan, the leadership class supercomputer at Oak Ridge National Laboratory.

97 MATHEMATICS AND COMPUTING↗

A two-level GPU-accelerated incomplete LU preconditioner for general sparse linear systems

This paper presents a parallel preconditioning approach based on incomplete LU (ILU) factorizations in the framework of Domain Decomposition (DD) for general sparse linear systems. We focus on distributed memory parallel architectures, specifically, those that are equipped with graphic processing units (GPUs). In addition to block-Jacobi, we present general purpose two-level ILU Schur complement-based approaches, where different strategies are presented to solve the coarse-level reduced system. These strategies are combined with modified ILU methods in the construction of the coarse-level operator, in order to effectively remove smooth errors by targeting an algebraically smooth vector. We leverage available GPU-based sparse matrix kernels to accelerate the setup and the solve phases of the proposed ILU preconditioner. We evaluate the efficiency of the proposed methods as a smoother for algebraic multigrid (AMG) and as a preconditioner for Krylov subspace methods on challenging anisotropic diffusion problems and a collection of general sparse matrices.

97 MATHEMATICS AND COMPUTING↗