Search NASA⌕ Search

SEARCH · Search NASA

Results for “Threading”

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 811 records · Page 45

Hybrid-view programming of nuclear fusion simulation code in the PGAS parallel programming language XcalableMP

Recently, the Partitioned Global Address Space (PGAS) parallel programming model has emerged as a usable distributed memory programming model. XcalableMP (XMP) is a PGAS parallel programming language that extends base languages such as C and Fortran with directives in OpenMP-like style. XMP supports a global-view model that allows programmers to define global data and to map them to a set of processors, which execute the distributed global data as a single thread. In XMP, the concept of a coarray is also employed for local-view programming. In this study, we port Gyrokinetic Toroidal Code - Princeton (GTC-P), which is a three-dimensional gyrokinetic PIC code developed at Princeton University to study the microturbulence phenomenon in magnetically confined fusion plasmas, to XMP as an example of hybrid memory model coding with the global-view and local-view programming models. In local-view programming, the coarray notation is simple and intuitive compared with Message Passing Interface (MPI) programming while the performance is comparable to that of the MPI version. Thus, because the global-view programming model is suitable for expressing the data parallelism for a field of grid space data, we implement a hybrid-view version using a global-view programming model to compute the field and a local-view programming model to compute the movement of particles. Finally, the performance is degraded by 20% compared with the original MPI version, but the hybrid-view version facilitates more natural data expression for static grid space data (in the global-view model) and dynamic particle data (in the local-view model), and it also increases the readability of the code for higher productivity.

97 MATHEMATICS AND COMPUTING↗

Performance optimization, modeling and analysis of sparse matrix-matrix products on multi-core and many-core processors

Sparse matrix-matrix multiplication (SpGEMM) is a computational primitive that is widely used in areas ranging from traditional numerical applications to recent big data analysis and machine learning. Although many SpGEMM algorithms have been proposed, hardware specific optimizations for multi- and many-core processors are lacking and a detailed analysis of their performance under various use cases and matrices is not available. In this work, we firstly identify and mitigate multiple bottlenecks with memory management and thread scheduling on Intel Xeon Phi (Knights Landing or KNL). Specifically targeting many-core processors, we develop a hash-table-based algorithm and optimize a heap-based shared-memory SpGEMM algorithm. We examine their performance together with other publicly available codes. Different from the literature, our evaluation also includes use cases that are representative of real graph algorithms, such as multi-source breadth-first search or triangle counting. Our hash-table and heap-based algorithms are showing significant speedups from libraries in the majority of the cases while different algorithms dominate the other scenarios with different matrix size, sparsity, compression factor and operation type. We wrap up in-depth evaluation results and make a recipe to give the best SpGEMM algorithm for target scenario. We build the performance model for hash-table and heap-based algorithms, which supports the recipe. A critical finding is that hash-table-based SpGEMM gets a significant performance boost if the nonzeros are not required to be sorted within each row of the output matrix. Finally, we integrate our implementations into a large-scale protein clustering code named HipMCL, accelerating its SpGEMM kernel by up to 10X and achieving an overall performance boost for the whole HipMCL application by 2.6X.

97 MATHEMATICS AND COMPUTING↗

Scalable line and plane relaxation in a parallel structured multigrid solver

The efficient solution of sparse, linear systems that arise through the discretization of partial differential equations remains a key challenge for a range of high performance scientific simulations. One approach for reducing data movement and improving performance is by exposing and exploiting structure in a problem through the use of robust structured multilevel solvers. By choosing coarsening that preserves the structure of the problem, these methods maintain efficient structured computation and communication throughout the multigrid hierarchy. However, when coarsening is not permitted to be dependent on the operator, anisotropy must be addressed by the smoother — producing error compatible for coarse-grid correction with structured coarsening. Here, the components required in a scalable parallel structured solver are described with a focus on memory and communication efficiency of robust smoothers. While the implementation of communication and memory reduction techniques in smoothers integrated in a complete 3D solver present a significant engineering challenge, a novel approach is proposed that addresses these challenges systematically through a change to the solver’s execution model. Enabled by user-level threading paired with a set of data and communication abstractions, this approach permits seamless aggregation of communication in plane smoothers — directly reusing code for a 2D distributed multilevel cycle. Results show an effective reduction in communication costs for coarse-grid problems, and result in a speedup of 8.7x in smoothing routines shown in Fig. 12 using this approach. This produces a significant improvement to strong scalability while maintaining favorable weak scaling behavior. Finally, a parallel scaling study using a series of refined meshes is included that demonstrates the effectiveness of this approach in an application of interest.

97 MATHEMATICS AND COMPUTING↗

Leaf isoprene and monoterpene emission distribution across hyperdominant tree genera in the Amazon basin

Tropical forests are acknowledged to be the largest global source of isoprene (C 5 H 8 ) and monoterpenes (C 10 H 16 ) emissions, with current synthesis studies suggesting few tropical species emit isoprenoids (20-38%) and do so with highly variable emission capacities, including within the same genera. This apparent lack of a clear phylogenetic thread has created difficulties both in linking isoprenoid function with evolution and for the development of accurate biosphere-atmosphere models. In this work, we present a systematic emission study of "hyperdominant" tree species in the Amazon Basin. Across 162 individuals, distributed among 25 botanical families and 113 species, isoprenoid emissions were widespread among both early and late successional species (isoprene: 61.9% of the species; monoterpenes: 15.0%; both isoprene and monoterpenes: 9.7%). The hyperdominant species (69) across the top five most abundant genera, which make up about 50% of all individuals in the Basin, had a similar abundance of isoprenoid emitters (isoprene: 63.8%; monoterpenes: 17.4%; both 11.6%). Among the abundant genera, only Pouteria had a low frequency of isoprene emitting species (15.8% of 19 species). In contrast, Protium, Licania, Inga, and Eschweilera were rich in isoprene emitting species (83.3% of 12 species, 61.1% of 18 species, 100% of 8 species, and 100% of 12 species, respectively). Light response curves of individuals in each of the five genera showed light-dependent, photosynthesis-linked emission rates of isoprene and monoterpenes. Importantly, in every genus, we observed species with light-dependent isoprene emissions together with monoterpenes including β-ocimene. These observations support the emerging view of the evolution of isoprene synthases from β-ocimene synthases. Our results have important implications for understanding isoprenoid function-evolution relationships and the development of more accurate Earth System Models.

59 BASIC BIOLOGICAL SCIENCES↗

A morphology independent methodology for quantifying planview river change and characteristics from remotely sensed imagery

Remotely sensed imagery of rivers has long served as a means for characterizing channel properties and detection of planview change. In the last decade the dramatic increase in the availability of satellite imagery and processing tools has created the potential to greatly expand the spatial and temporal scale of our understanding of river morphology and dynamics. To date, the majority of GIS and automated analyses of planview changes in rivers from remotely sensed data has been developed for single-threaded meandering river systems. These methods have limited applicability to many of the earth's rivers with complex multi-channel planforms. Here we present the methodologies of a set of analysis algorithms collectively called Spatially Continuous Riverbank Erosion and Accretion Measurements (SCREAM). SCREAM analyzes planview river metrics regardless of river morphology. These algorithms quantify both the erosion and accretion rates of riverbanks from binary masks of channels generated from imagery acquired at two time periods. Additionally, the program quantifies the area of change between river channels and the surrounding floodplain and area of islands lost or formed between these two time periods. To examine variations in erosion rates in relation to local channel attributes and make rate comparisons between river systems of varying sizes, the program determines channel widths and bank curvature at every bank pixel. SCREAM was developed and tested on rivers with diverse and complex planform morphologies in imagery acquired from a range of observational platforms with varying spatial resolutions. Here, validation and verification of SCREAM-generated metrics against manual measurements show no significant measurement errors in determination of channel width, erosion, and bank aspects. SCREAM has the potential to provide data for both the quantitative examination of the controls on erosion rates and for the comparison of these rates across river systems ranging broadly in size and planform morphology.

47 OTHER INSTRUMENTATION↗

ED-cPSD: Fast Phase-Size Distribution via Sequential Erosion-Dilation

The Erosion-Dilation continuous Phase-Size Distribution, ED-cPSD, is an application for calculating continuous pore and particle-size distribution from digital reconstructions and/or image-based structural data. It is based on the erosion-dilation continuous phase-size distribution method. A continuous size distribution is a measure of the probability density of finding a particle or pore of a certain size. These distributions are of interest in any field of study involving porous media, including but not limited to electrochemistry, petroleum engineering, geology, and food science. The algorithm behind the software provides a computationally efficient way to calculate phase-size distributions for large domains. For a 3D battery electrode reconstruction with 1.3 x 10 8 voxels, the particle size distribution is derived in under 2 min on a desktop, while also retaining flexibility and computational efficiency for HPC-scale multi-threading. The software can handle structures with over 10 9 voxels. The algorithm is roughly 280 times faster than a previous version on the same task.

Characterization↗

Reynolds number dependence of Lagrangian dispersion in direct numerical simulations of anisotropic magnetohydrodynamic turbulence

Large-scale magnetic fields thread through the electrically conducting matter of the interplanetary and interstellar medium, stellar interiors and other astrophysical plasmas, producing anisotropic flows with regions of high-Reynolds-number turbulence. It is common to encounter turbulent flows structured by a magnetic field with a strength approximately equal to the root-mean-square magnetic fluctuations. In this work, direct numerical simulations of anisotropic magnetohydrodynamic (MHD) turbulence influenced by such a magnetic field are conducted for a series of cases that have identical resolution, and increasing grid sizes up to $2048^3$ . The result is a series of closely comparable simulations at Reynolds numbers ranging from 1400 up to 21 000. We investigate the influence of the Reynolds number from the Lagrangian viewpoint by tracking fluid particles and calculating single-particle and two-particle statistics. The influence of Alfvénic fluctuations and the fundamental anisotropy on the MHD turbulence in these statistics is discussed. Single-particle diffusion curves exhibit mildly superdiffusive behaviours that differ in the direction aligned with the magnetic field and the direction perpendicular to it. Competing alignment processes affect the dispersion of particle pairs, in particular at the beginning of the inertial subrange of time scales. Scalings for relative dispersion, which become clearer in the inertial subrange for a larger Reynolds number, can be observed that are steeper than indicated by the Richardson prediction.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

A tutorial introduction to the statistical theory of turbulent plasmas, a half-century after Kadomtsev’s Plasma Turbulence and the resonance-broadening theory of Dupree and Weinstock

In honour of the 50th anniversary of the influential review/monograph on plasma turbulence by B. B. Kadomtsev as well as the seminal works of T. H. Dupree and J. Weinstock on resonance-broadening theory, an introductory tutorial is given about some highlights of the statistical–dynamical description of turbulent plasmas and fluids, including the ideas of nonlinear incoherent noise, coherent damping, and self-consistent dielectric response. The statistical closure problem is introduced. Incoherent noise and coherent damping are illustrated with a solvable model of passive advection. Self-consistency introduces turbulent polarization effects that are described by the dielectric function ${\mathcal{D}}$ . Dupree’s method of using ${\mathcal{D}}$ to estimate the saturation level of turbulence is described; then it is explained why a more complete theory that includes nonlinear noise is required. The general theory is best formulated in terms of Dyson equations for the covariance $C$ and an infinitesimal response function $R$ , which subsumes ${\mathcal{D}}$ . An important example is the direct-interaction approximation (DIA). It is shown how to use Novikov’s theorem to develop an $\boldsymbol{x}$ -space approach to the DIA that is complementary to the original $\boldsymbol{k}$ -space approach of Kraichnan. A dielectric function is defined for arbitrary quadratically nonlinear systems, including the Navier–Stokes equation, and an algorithm for determining the form of ${\mathcal{D}}$ in the DIA is sketched. The independent insights of Kadomtsev and Kraichnan about the problem of the DIA with random Galilean invariance are described. The mixing-length formula for drift-wave saturation is discussed in the context of closures that include nonlinear noise (shielded by ${\mathcal{D}}$ ). The role of $R$ in the calculation of the symmetry-breaking (zonostrophic) instability of homogeneous turbulence to the generation of inhomogeneous mean flows is addressed. The second-order cumulant expansion and the stochastic structural stability theory are also discussed in that context. In conclusion, various historical research threads are mentioned and representative entry points to the literature are given. In addition, some outstanding conceptual issues are enumerated.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Approximation algorithms in combinatorial scientific computing

We survey recent work on approximation algorithms for computing degree-constrained subgraphs in graphs and their applications in combinatorial scientific computing. The problems we consider include maximization versions of cardinality matching, edge-weighted matching, vertex-weighted matching and edge-weighted $b$ -matching, and minimization versions of weighted edge cover and $b$ -edge cover. Exact algorithms for these problems are impractical for massive graphs with several millions of edges. For each problem we discuss theoretical foundations, the design of several linear or near-linear time approximation algorithms, their implementations on serial and parallel computers, and applications. Our focus is on practical algorithms that yield good performance on modern computer architectures with multiple threads and interconnected processors. We also include information about the software available for these problems.

Mathematics↗

Large Area and Depth-Profiling Dislocation Imaging and Strain Analysis in Si/SiGe/Si Heterostructures

We demonstrate the combined use of large area depth-profiling dislocation imaging and quantitative composition and strain measurement for a strained Si/SiGe/Si sample based on nondestructive techniques of electron beam-induced current (EBIC) and X-ray diffraction reciprocal space mapping (XRD RSM). Depth and improved spatial resolution is achieved for dislocation imaging in EBIC by using different electron beam energies at a low temperature of ~7 K. Images recorded clearly show dislocations distributed in three regions of the sample: deep dislocation networks concentrated in the “strained” SiGe region, shallow misfit dislocations at the top Si/SiGe interface, and threading dislocations connecting the two regions. Dislocation densities at the top of the sample can be measured directly from the EBIC results. XRD RSM reveals separated peaks, allowing a quantitative measurement of composition and strain corresponding to different layers of different composition ratios. High-resolution scanning transmission electron microscopy cross-section analysis clearly shows the individual composition layers and the dislocation lines in the layers, which supports the EBIC and XRD RSM results.

36 MATERIALS SCIENCE↗

Conjugated Macrocycles in Organic Electronics

This Account describes a body of research on the design, synthesis, and application of a new class of electronic materials made from conjugated macrocycles. Our macrocyclic design takes into consideration the useful attributes of fullerenes and what properties make fullerenes efficient n-type materials. We identified four electronic and structural elements: (1) a three-dimensional shape; (2) a conjugated and delocalized π-space; (3) the presence of an interior and exterior to the π-surface; and (4) low-energy unoccupied molecular orbitals allowing them to accept electrons. The macrocyclic design incorporates some of these properties, including a three-dimensional shape, an interior/exterior to the π-surface, and low-lying LUMOs maintaining the n-type semiconducting behavior, yet we also install synthetic flexibility in our approach in order to tune the properties further. Each of the macrocycles comprises perylenediimide cores wound together with linkers. The perylenediimide building block endows each macrocycle with the ability to accept electrons, while the synthetic flexibility to install different linkers allows us to create macrocycles with different electronic properties and sizes. We have created three macrocycles that all absorb well into the visible range of the solar spectrum and possess different shapes and sizes. We then use these materials in an array of applications that take advantage of their ability to function as n-type semiconductors, absorb in the visible range of the solar spectrum, and possess intramolecular cavities. This Account will discuss our progress in incorporating these new macrocycles in organic solar cells, organic photodetectors, organic field effect transistors, and sensors. The macrocycles outperform acyclic controls in organic solar cells. We find the more rigid macrocyclic structure results in less intrinsic charges and lower dark current in organic photodetectors. Our macrocyclic-based photodetector has the highest detectivity of non-fullerene acceptors. The macrocycles also function as sensors and are able to recognize nuanced differences in analytes. Perylenediimide-based fused oligomers are efficient materials in both organic solar cells and field effect transistors. Here, we will use the oligomers to construct macrocycles for use in solar energy conversion. In addition, we will incorporate different electron-rich linkers in our cycles in an attempt to engineer the HOMO/LUMO gap further. Looking further into the future, we envision opportunities in applying these conjugated macrocycles as electronic host/guest materials, as concatenated electronic materials by threading the macrocycles with electroactive oligomers, and as a locus for catalysis that is driven by light and electric fields.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Reduced Dislocation Introduction in III–V/Si Heterostructures with Glide-Enhancing Compressively Strained Superlattices

The novel use of a GaAs y P 1-y /GaP compressively-strained superlattice (CSS) to provide enhanced control over misfit dislocation (MD) evolution and threading dislocation density (TDD) during GaP/Si metamorphic heteroepitaxy is demonstrated. Here, the insertion of the CSS just after critical thickness, and thus prior to substantial dislocation introduction, is found to yield significantly reduced TDD in relaxed, 500 nm thick, n-type GaP/Si versus comparable control samples. The impact of CSS period count on average TDD and the overall dislocation network morphology was examined, supported by quantitative microstructural characterization, revealing a nearly 20× relative TDD reduction (to 2.4 ± 0.4×10 6 cm -2 ) with a 3-period CSS structure. A similarly low TDD (3.0 ± 0.6×10 6 cm -2 ) is maintained when the resultant n-GaP/Si virtual substrate is used for the growth of a subsequent n-type GaAs 0.75 P 0.25 -terminal GaAs y P 1-y step-graded metamorphic buffer. Although the physical mechanism for TDD reduction provided by these structures is not yet entirely understood, this initial work suggests that enhanced glide dynamics of MDs at or within the CSS placed early in the growth leads to a reduction in the total number of dislocations introduced overall, as opposed to annihilation-based reduction that occurs in conventional strained-layer superlattice dislocation filter approaches.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Development of High-Efficiency GaAs Solar Cells Grown on Nanopatterned GaAs Substrates

One approach to reducing the cost of high-efficiency III–V devices involves adding patterned layers to heteroepitaxial or homoepitaxial substrates to facilitate substrate removal and reuse. However, few studies have focused explicitly on high-quality devices grown over patterned substrates, which is required for any cost saving to be beneficial. In this work, we demonstrate the growth of high-efficiency GaAs solar cells on GaAs substrates patterned with an array of nanoscale SiOX mask stripes. We show that reducing the pattern dimensions to submicron length scales with nanoimprint lithography enables defect-free coalescence. By varying the growth conditions, faceting of the epilayer material during overgrowth of the patterned mask was also controlled. A V/III ratio of 200 during MOVPE overgrowth produced smooth coalesced epilayers, which is desirable for the growth of subsequent device layers. Inverted GaAs front homojunction devices grown on patterned GaAs(001) substrates achieved threading dislocation densities below 5 × 10 5 cm –2 and maintained >23% solar cell efficiencies at one sun illumination, equivalent to control devices grown on unpatterned epi-ready substrates.

14 SOLAR ENERGY↗

Heteroepitaxial Integration of ZnGeN 2 on GaN Buffers Using Molecular Beam Epitaxy

Recently theorized hybrid II-IV-N 2 /III-N heterostructures, based on current commercialized (In,Ga)N devices, are predicted to significantly advance the design space of highly efficient optoelectronics in the visible spectrum, yet there are few epitaxial studies of II-IV-N 2 materials. In this work, we present heteroepitaxial ZnGeN 2 grown on GaN buffers and AlN templates. We demonstrate that a GaN nucleating surface is crucial for increasing the ZnGeN 2 crystallization rate to combat Zn desorption, extending the stoichiometric growth window from 215 degrees C on AlN to 500 degrees C on GaN buffers. Structural characterization reveals well-crystallized films with threading dislocations extending from the GaN buffer. These films have a critical thickness for relaxation of 20-25 nm as determined by reflection high energy electron diffraction (RHEED) and cross-sectional scanning electron microscopy (SEM). The films exhibit a cation-disordered wurtzite structure, with lattice constants a = 3.216 +/- 0.004 A and c = 5.215 +/- 0.005 A determined by RHEED and X-ray diffraction (XRD). This work demonstrates a significant step toward the development of hybrid ZnGeN 2 -GaN integrated devices.

36 MATERIALS SCIENCE↗

Implementation of the Molecular Electrostatic Potential over Graphics Processing Units

The molecular electrostatic potential (MEP) generated by quantum chemistry methods and Gaussian functions is evaluated over graphics processing units (GPUs). This implementation is based on full-range Rys polynomials with nodes and weights obtained in each thread of a GPU. For high angular moments, the corresponding integral is solved using a one dimension vertical recurrence relation. Thus, we computed the MEP with minimal approximations. We show that this implementation is stable and very efficient since the time consumed over GPUs is quite small compared with similar implementations over CPUs. The implementation was done by using CUDA-C programming techniques within the Graphics Processing Units for Atoms and Molecules (GPUAM) project, which has been designed to analyze quantum chemistry fields over heterogeneous computational resources. With this new scalar field GPUAM is a useful application for the quantum chemistry community, in particular for people interested in chemical reactivity analysis.

Cruz-Monterrosas, Julio-Cesar↗

Acceleration of Semiempirical QM/MM Methods through Message Passage Interface (MPI), Hybrid MPI/Open Multiprocessing, and Self-Consistent Field Accelerator Implementations

The strategy and implementation of scalable and efficient semiempirical (SE) QM/MM methods in CHARMM are described. The serial version of the code was first profiled to identify routines that required parallelization. Afterward, the code was parallelized and accelerated with three approaches. The first approach was the parallelization of the entire QM/MM routines, including the Fock matrix diagonalization routines, using the CHARMM message passage interface (MPI) machinery. In the second approach, two different self-consistent field (SCF) energy convergence accelerators were implemented using density and Fock matrices as targets for their extrapolations in the SCF procedure. In the third approach, the entire QM/MM and MM energy routines were accelerated by implementing the hybrid MPI/open multiprocessing (OpenMP) model in which both the task- and loop-level parallelization strategies were adopted to balance loads between different OpenMP threads. The present implementation was tested on two solvated enzyme systems (including <100 QM atoms) and an S N 2 symmetric reaction in water. The MPI version exceeded existing SE QM methods in CHARMM, which include the SCC-DFTB and SQUANTUM methods, by at least 4-fold. The use of SCF convergence accelerators further accelerated the code by ~12–35% depending on the size of the QM region and the number of CPU cores used. Although the MPI version displayed good scalability, the performance was diminished for large numbers of MPI processes due to the overhead associated with MPI communications between nodes. This issue was partially overcome by the hybrid MPI/OpenMP approach which displayed a better scalability for a larger number of CPU cores (up to 64 CPUs in the tested systems).

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Cylindrical Confinement of Nanocolloidal Cholesteric Liquid Crystal

The organization of nanocolloidal liquid crystals in constrained geometries has fundamental and practical importance, since under confinement, liquid crystals contain stable topological defects that can serve as templates for nanoparticle organization. Three-dimensional confinement of cholesteric (Ch) liquid crystals formed by cellulose nanocrystals (CNCs) have been extensively studied; however, their two-dimensional confinement remains underinvestigated. In this work, we report the results of systematic experimental studies of two-dimensional confinement of Ch–CNC liquid crystal in cylindrical capillaries with varying inner diameters. Confinement resulted in phase separation of the Ch–CNC liquid crystal into a Ch shell formed by concentric CNC pseudolayers with the helicoidal axis perpendicular to the inner surface of the capillary walls, and a micrometer-diameter isotropic core thread running parallel to the long axis of the capillary. The morphology of the confined Ch–CNC liquid crystal varied when progressively increasing the degree confinement. Finally, we show that phase separation of the Ch–CNC liquid crystal into a Ch shell and an isotropic core is preserved in flexible capillary tubing, suggesting the applicability of this system for the fabrication of flexible optical waveguides.

59 BASIC BIOLOGICAL SCIENCES↗