Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallelization efficiency”

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 973 records · Page 54

Performance Evaluation and Modeling Techniques for Parallel Processors

In practice, the performance evaluation of supercomputers is still substantially driven by singlepoint estimates of metrics (e.g., MFLOPS) obtained by running characteristic benchmarks or workloads. With the rapid increase in the use of time-shared multiprogramming in these systems, such measurements are clearly inadequate. This is because multiprogramming and system overhead, as well as other degradations in performance due to time varying characteristics of workloads, are not taken into account. In multiprogrammed environments, multiple jobs and users can dramatically increase the amount of system overhead and degrade the performance of the machine. Performance techniques, such as benchmarking, which characterize performance on a dedicated machine ignore this major component of true computer performance. Due to the complexity of analysis, there has been little work done in analyzing, modeling, and predicting the performance of applications in multiprogrammed environments. This is especially true for parallel processors, where the costs and benefits of multi-user workloads are exacerbated. While some may claim that the issue of multiprogramming is not a viable one in the supercomputer market, experience shows otherwise. Even in recent massively parallel machines, multiprogramming is a key component. It has even been claimed that a partial cause of the demise of the CM2 was the fact that it did not efficiently support time-sharing. In the same paper, Gordon Bell postulates that, multicomputers will evolve to multiprocessors in order to support efficient multiprogramming. Therefore, it is clear that parallel processors of the future will be required to offer the user a time-shared environment with reasonable response times for the applications. In this type of environment, the most important performance metric is the completion of response time of a given application. However, there are a few evaluation efforts addressing this issue.

Dimpsey, Robert Tod↗

Time-domain analysis of planar microstrip devices using a generalized Yee-algorithm based on unstructured grids

The generalized Yee-algorithm is presented for the temporal full-wave analysis of planar microstrip devices. This algorithm has the significant advantage over the traditional Yee-algorithm in that it is based on unstructured and irregular grids. The robustness of the generalized Yee-algorithm is that structures that contain curved conductors or complex three-dimensional geometries can be more accurately, and much more conveniently modeled using standard automatic grid generation techniques. This generalized Yee-algorithm is based on the the time-marching solution of the discrete form of Maxwell's equations in their integral form. To this end, the electric and magnetic fields are discretized over a dual, irregular, and unstructured grid. The primary grid is assumed to be composed of general fitted polyhedra distributed throughout the volume. The secondary grid (or dual grid) is built up of the closed polyhedra whose edges connect the centroid's of adjacent primary cells, penetrating shared faces. Faraday's law and Ampere's law are used to update the fields normal to the primary and secondary grid faces, respectively. Subsequently, a correction scheme is introduced to project the normal fields onto the grid edges. It is shown that this scheme is stable, maintains second-order accuracy, and preserves the divergenceless nature of the flux densities. Finally, for computational efficiency the algorithm is structured as a series of sparse matrix-vector multiplications. Based on this scheme, the generalized Yee-algorithm has been implemented on vector and parallel high performance computers in a highly efficient manner.

Gedney, Stephen D.↗

Techniques for Enabling Highly Efficient Message Passing on Many-Core Architectures

Many-core architecture provides a massively parallel environment with dozens of cores and hundreds of hardware threads. Scientific application programmers are increasingly looking at ways to utilize such large numbers of lightweight cores for various programming models. Efficiently executing these models on massively parallel many-core environments is not easy, however and performance may be degraded in various ways. The first author's doctoral research focuses on exploiting the capabilities of many-core architectures on widely used MPI implementations. While application programmers have studied several approaches to achieve better parallelism and resource sharing, many of those approaches still face communication problems that degrade performance. In the thesis, we investigate the characteristics of MPI on such massively threaded architectures and propose two efficient strategies -- a multi-threaded MPI approach and a process-based asynchronous model -- to optimize MPI communication for modern scientific applications.

Si, Min↗

Performance of the Wavelet Decomposition on Massively Parallel Architectures

Traditionally, Fourier Transforms have been utilized for performing signal analysis and representation. But although it is straightforward to reconstruct a signal from its Fourier transform, no local description of the signal is included in its Fourier representation. To alleviate this problem, Windowed Fourier transforms and then wavelet transforms have been introduced, and it has been proven that wavelets give a better localization than traditional Fourier transforms, as well as a better division of the time- or space-frequency plane than Windowed Fourier transforms. Because of these properties and after the development of several fast algorithms for computing the wavelet representation of any signal, in particular the Multi-Resolution Analysis (MRA) developed by Mallat, wavelet transforms have increasingly been applied to signal analysis problems, especially real-life problems, in which speed is critical. In this paper we present and compare efficient wavelet decomposition algorithms on different parallel architectures. We report and analyze experimental measurements, using NASA remotely sensed images. Results show that our algorithms achieve significant performance gains on current high performance parallel systems, and meet scientific applications and multimedia requirements. The extensive performance measurements collected over a number of high-performance computer systems have revealed important architectural characteristics of these systems, in relation to the processing demands of the wavelet decomposition of digital images.

El-Ghazawi, Tarek A.↗

A Step towards Energy Efficient Computing: Redesigning a Hydrodynamic Application on CPU-GPU, In: 2014 IEEE 28th International Parallel and Distributed Processing Symposium

Power and energy consumption are becoming an increasing concern in high performance computing. Compared to multi-core CPUs, GPUs have a much better performance per watt. In this paper we discuss efforts to redesign the most computation intensive parts of BLAST, an application that solves the equations for compressible hydrodynamics with high order finite elements, using GPUs BLAST, Dobrev. In order to exploit the hardware parallelism of GPUs and achieve high performance, we implemented custom linear algebra kernels. We intensively optimized our CUDA kernels by exploiting the memory hierarchy, which exceed the vendor's library routines substantially in performance. We proposed an auto tuning technique to adapt our CUDA kernels to the orders of the finite element method. Compared to a previous base implementation, our redesign and optimization lowered the energy consumption of the GPU in two aspects: 60% less time to solution and 10% less power required. Compared to the CPU-only solution, our GPU accelerated BLAST obtained a 2.5× overall speedup and 1.42× energy efficiency (green up) using 4th order (Q_4) finite elements, and a 1.9× speedup and 1.27× green up using 2nd order (Q2) finite elements.

Computer Science↗

eQE: An open‐source density functional embedding theory code for the condensed phase

Abstract In this work, we present the main features and algorithmic details of a novel implementation of the frozen density embedding (FDE) formulation of subsystem density functional theory (DFT) that is specifically designed to enable ab initio molecular dynamics (AIMD) simulations of large‐scale condensed‐phase systems containing 1000s of atoms. This code (available at http://eqe.rutgers.edu ) has been given the moniker of embedded Quantum ESPRESSO (eQE) as it is a generalization of the open‐source Quantum ESPRESSO (QE) suite of programs. The strengths of eQE reside in a hierarchical parallelization scheme that allows for an efficient and fully self‐consistent treatment of the electronic structure (via the addition of an additional DIIS extrapolation layer) while simultaneously exploiting the inherent symmetries and periodicities in the system (via sampling of subsystem‐specific first Brillouin zones and utilization of subsystem‐specific basis sets). While bulk liquids and molecular crystals are two classes of systems that exemplify the utility of the FDE approach (as these systems can be partitioned into weakly interacting subunits), we show that eQE has significantly extended this regime of applicability by outperforming standard semilocal Kohn–Sham DFT (KS‐DFT) for large‐scale heterogeneous catalysts with quite different layer‐specific electronic structure and intrinsic periodicities. eQE features very favorable strong parallel scaling for a model system of bulk liquid water composed of 256 water molecules, which allows for a significant decrease in the overall time to solution when compared to KS‐DFT. We show that eQE achieves speedups greater than one order of magnitude ( ) when performing AIMD simulations of such large‐scale condensed‐phase systems as: (1) molecular liquids via bulk liquid water represented by 1024 independent water molecules (3072 atoms with a 25.3× speedup over KS‐DFT), (2) polypeptide/biomolecule solvation via ( gly ) 6 solvated in (H 2 O) 395 (1230 atoms with a 38.6× speedup over KS‐DFT), and (3) molecular crystals via a 3 × 3 × 3 periodic supercell of pentacene (1940 atoms with a 12.0× speedup over KS‐DFT). These results represent a significant improvement over the current state‐of‐the‐art and now enable subsystem DFT‐based AIMD simulations of realistically sized condensed‐phase systems of interest throughout chemistry, physics, and materials science.

Genova, Alessandro↗

Progress in Computational Simulation of Earthquakes

GeoFEST(P) is a computer program written for use in the QuakeSim project, which is devoted to development and improvement of means of computational simulation of earthquakes. GeoFEST(P) models interacting earthquake fault systems from the fault-nucleation to the tectonic scale. The development of GeoFEST( P) has involved coupling of two programs: GeoFEST and the Pyramid Adaptive Mesh Refinement Library. GeoFEST is a message-passing-interface-parallel code that utilizes a finite-element technique to simulate evolution of stress, fault slip, and plastic/elastic deformation in realistic materials like those of faulted regions of the crust of the Earth. The products of such simulations are synthetic observable time-dependent surface deformations on time scales from days to decades. Pyramid Adaptive Mesh Refinement Library is a software library that facilitates the generation of computational meshes for solving physical problems. In an application of GeoFEST(P), a computational grid can be dynamically adapted as stress grows on a fault. Simulations on workstations using a few tens of thousands of stress and displacement finite elements can now be expanded to multiple millions of elements with greater than 98-percent scaled efficiency on over many hundreds of parallel processors (see figure).

Donnellan, Andrea↗

Perovskite superlattices with efficient carrier dynamics

Compared with their three-dimensional (3D) counterparts, low-dimensional metal halide perovskites (2D and quasi-2D; B 2 A n-1 M n X 3n+1 , such as B = R-NH 3 + , A = HC(NH 2 ) 2 + , Cs + ; M = Pb 2+ , Sn 2+ ; X = Cl - , Br - , I - ) with periodic inorganic-organic structures have shown promising stability and hysteresis-free electrical performance. However, their unique multiple-quantum-well structure limits the device efficiencies because of the grain boundaries and randomly oriented quantum wells in polycrystals. In single crystals, the carrier transport through the thickness direction is hindered by the layered insulating organic spacers. Furthermore, the strong quantum confinement from the organic spacers limits the generation and transport of free carriers. Also, lead-free metal halide perovskites have been developed but their device performance is limited by their low crystallinity and structural instability 11 . Here, in this paper, we report a low-dimensional metal halide perovskite BA 2 MA n-1 Sn n I 3n+1 (BA, butylammonium; MA, methylammonium; n = 1, 3, 5) superlattice by chemical epitaxy. The inorganic slabs are aligned vertical to the substrate and interconnected in a criss-cross 2D network parallel to the substrate, leading to efficient carrier transport in three dimensions. A lattice-mismatched substrate compresses the organic spacers, which weakens the quantum confinement. The performance of a superlattice solar cell has been certified under the quasi-steady state, showing a stable 12.36% photoelectric conversion efficiency. Moreover, an intraband exciton relaxation process may have yielded an unusually high open-circuit voltage (V OC ).

36 MATERIALS SCIENCE↗

A Parallel Pipelined Renderer for the Time-Varying Volume Data

This paper presents a strategy for efficiently rendering time-varying volume data sets on a distributed-memory parallel computer. Time-varying volume data take large storage space and visualizing them requires reading large files continuously or periodically throughout the course of the visualization process. Instead of using all the processors to collectively render one volume at a time, a pipelined rendering process is formed by partitioning processors into groups to render multiple volumes concurrently. In this way, the overall rendering time may be greatly reduced because the pipelined rendering tasks are overlapped with the I/O required to load each volume into a group of processors; moreover, parallelization overhead may be reduced as a result of partitioning the processors. We modify an existing parallel volume renderer to exploit various levels of rendering parallelism and to study how the partitioning of processors may lead to optimal rendering performance. Two factors which are important to the overall execution time are re-source utilization efficiency and pipeline startup latency. The optimal partitioning configuration is the one that balances these two factors. Tests on Intel Paragon computers show that in general optimal partitionings do exist for a given rendering task and result in 40-50% saving in overall rendering time.

Chiueh, Tzi-Cker↗

Full Core Reactor Analysis: Running Denovo on Jaguar

Fully-consistent, full-core, 3D, deterministic simulations using the orthogonal mesh code Denovo were run on the massively parallel computing architecture Jaguar XT5. Using energy and spatial parallelization schemes, Denovo was able to efficiently scale to over 160k processors. Cell-homogenized cross-sections were used with Step-Characteristics, Linear-Discontinuous Finite Element, and Tri-Linear-Discontinuous Finite Element spatial methods. It was determined that using the finite element methods gave considerably more accurate eigenvalue solutions for large aspect ratios meshes than those using Step-Characteristics.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Preheating after multifield inflation with nonminimal couplings. II. Resonance structure

This is the second in a series of papers on preheating in inflationary models comprised of multiple scalar fields coupled nonminimally to gravity. We work in the rigid-spacetime approximation and consider field trajectories within the single-field attractor, which is a generic feature of these models. We construct the Floquet charts to find regions of parameter space in which particle production is efficient for both the adiabatic and isocurvature modes, and analyze the resonance structure using analytic and semianalytic techniques. Particle production in the adiabatic direction is characterized by the existence of an asymptotic scaling solution at large values of the nonminimal couplings, ${{\xi}}_{I}{\gg}1$, in which the dominant instability band arises in the long-wavelength limit, for comoving wave numbers $k{\rightarrow}0$. However, the large-${{\xi}}_{I}$ regime is not reached until ${{\xi}}_{I}{\ge}\mathcal{O}(100)$. In the intermediate regime, with ${{\xi}}_{I}{\sim}\mathcal{O}(1-10)$, the resonance structure depends strongly on wave number and couplings. The resonance structure for isocurvature perturbations is distinct and more complicated than its adiabatic counterpart. An intermediate regime, for ${{\xi}}_{I}{\sim}\mathcal{O}(1-10)$, is again evident. For large values of ${{\xi}}_{I}$, the Floquet chart consists of densely spaced, nearly parallel instability bands, suggesting a very efficient preheating behavior. The increased efficiency arises from features of the nontrivial field-space manifold in the Einstein frame, which itself arises from the fields’ nonminimal couplings in the Jordan frame, and has no analog in models with minimal couplings. Quantitatively, the approach to the large-${{\xi}}_{I}$ asymptotic solution for isocurvature modes is slower than in the case of the adiabatic modes.

79 ASTRONOMY AND ASTROPHYSICS↗

Diving into Petascale Production File Systems through Large Scale Profiling and Analysis

As leadership computing facilities grow their storage capacity into the multi- petabyte range, the number of files and directories leap into the scale of billions. A complete profiling of such a parallel file system in a production environment presents a unique challenge. On one hand, the time, resources, and negative performance impact on production users can make regular profiling difficult. On the other hand, the result of such profiling can yield much needed understanding of the file system's general characteristics, as well as provide insight to how users write and access their data on a grand scale. This paper presents a lightweight and scalable profiling solution that can efficiently walk, analyze, and profile multi-petabyte parallel file systems. This tool has been deployed and is in regular use on very large-scale production parallel file systems at both Oak Ridge National Lab's Oak Ridge Leadership Facility (OLCF) and Lawrence Livermore National Lab's Livermore Computing (LC) facilities. We present the results of our initial analysis on the data collected from these two large-scale production systems, organized into three use cases: (1) file system snapshot and composition, (2) striping pattern analysis for Lustre, and (3) simulated storage capacity utilization in preparation for future file systems. Our analysis shows that on the OLCF file system, over 96% of user files exhibit the default stripe width, potentially limiting performance on large files by underutilizing storage servers and disks. Our simulated block analysis quantitatively shows the space overhead when doing a forklift system migration. It also reveals that due to the difference in system compositions (OLCF vs. LC), we can achieve better performance and space trade-offs by employing different native file system block sizes.

Wang, Feiyi↗

Punctured Parallel and Serial Concatenated Convolutional Codes for BPSK/QPSK Channels

As available bandwidth for communication applications becomes scarce, bandwidth-efficient modulation and coding schemes become ever important. Since their discovery in 1993, turbo codes (parallel concatenated convolutional codes) have been the center of the attention in the coding community because of their bit error rate performance near the Shannon limit. Serial concatenated convolutional codes have also been shown to be as powerful as turbo codes. In this dissertation, we introduce algorithms for designing bandwidth-efficient rate r = k/(k + 1),k = 2, 3,..., 16, parallel and rate 3/4, 7/8, and 15/16 serial concatenated convolutional codes via puncturing for BPSK/QPSK (Binary Phase Shift Keying/Quadrature Phase Shift Keying) channels. Both parallel and serial concatenated convolutional codes have initially, steep bit error rate versus signal-to-noise ratio slope (called the -"cliff region"). However, this steep slope changes to a moderate slope with increasing signal-to-noise ratio, where the slope is characterized by the weight spectrum of the code. The region after the cliff region is called the "error rate floor" which dominates the behavior of these codes in moderate to high signal-to-noise ratios. Our goal is to design high rate parallel and serial concatenated convolutional codes while minimizing the error rate floor effect. The design algorithm includes an interleaver enhancement procedure and finds the polynomial sets (only for parallel concatenated convolutional codes) and the puncturing schemes that achieve the lowest bit error rate performance around the floor for the code rates of interest.

Acikel, Omer Fatih↗

HERCULES: A Pattern Driven Code Transformation System

New parallel computers are emerging, but developing efficient scientific code for them remains difficult. A scientist must manage not only the science-domain complexity but also the performance-optimization complexity. HERCULES is a code transformation system designed to help the scientist to separate the two concerns, which improves code maintenance, and facilitates performance optimization. The system combines three technologies, code patterns, transformation scripts and compiler plugins, to provide the scientist with an environment to quickly implement code transformations that suit his needs. Unlike existing code optimization tools, HERCULES is unique in its focus on user-level accessibility. In this paper we discuss the design, implementation and an initial evaluation of HERCULES.

99 GENERAL AND MISCELLANEOUS↗

POSTER: Automatic Differentiation of Parallel Loops with Formal Methods

The accompanying poster to this short paper presents a combination of reverse mode AD and formal methods to enable efficient differentiation of (or backpropagation through) shared-memory parallel code. Compared to the state of the art, our approach can more often avoid the need for atomic updates or private data copies during the parallel derivative computation, even in the presence of unstructured or data-dependent data access patterns. This is achieved by gathering information about the memory access patterns from the input program, which is assumed to be correctly parallelized. This information is then used to build a model of assertions in a theorem prover, which can be used to check the safety of shared memory accesses during the parallel derivative computation

Automatic Differentiation↗

ION ACCELERATION AT THE QUASI-PARALLEL BOW SHOCK: DECODING THE SIGNATURE OF INJECTION

Collisionless shocks are efficient particle accelerators. At Earth, ions with energies exceeding 100 keV are seen upstream of the bow shock when the magnetic geometry is quasi-parallel, and large-scale supernova remnant shocks can accelerate ions into cosmic-ray energies. This energization is attributed to diffusive shock acceleration; however, for this process to become active, the ions must first be sufficiently energized. How and where this initial acceleration takes place has been one of the key unresolved issues in shock acceleration theory. Using Cluster spacecraft observations, we study the signatures of ion reflection events in the turbulent transition layer upstream of the terrestrial bow shock, and with the support of a hybrid simulation of the shock, we show that these reflection signatures are characteristic of the first step in the ion injection process. These reflection events develop in particular in the region where the trailing edge of large-amplitude upstream waves intercept the local shock ramp and the upstream magnetic field changes from quasi-perpendicular to quasi-parallel. The dispersed ion velocity signature observed can be attributed to a rapid succession of ion reflections at this wave boundary. After the ions’ initial interaction with the shock, they flow upstream along the quasi-parallel magnetic field. Each subsequent wavefront in the upstream region will sweep the ions back toward the shock, where they gain energy with each transition between the upstream and the shock wave frames. Within three to five gyroperiods, some ions have gained enough parallel velocity to escape upstream, thus completing the injection process.

79 ASTRONOMY AND ASTROPHYSICS↗

An alternating direction implicit method for the Control Data STAR-100 vector computer

An implementation of the alternating direction implicit (ADI) method for the Control Data STAR-100 computer is presented and analyzed. Two parallel algorithms, both of which are most efficient when used to solve many independent tridiagonal systems of equations, are discussed relative to their usefulness in an ADI implementation on the STAR-100 computer. It is shown that it may be desirable to alternate between the parallel algorithms as the direction of implicitness is alternated in order to eliminate the data rearrangement which would otherwise be required. The applicability of the two parallel tridiagonal solvers to several other numerical algorithms is also discussed.

Lambiotte, J. J., Jr.↗

Is Monte Carlo embarrassingly parallel?

Monte Carlo is often stated as being embarrassingly parallel. However, running a Monte Carlo calculation, especially a reactor criticality calculation, in parallel using tens of processors shows a serious limitation in speedup and the execution time may even increase beyond a certain number of processors. In this paper the main causes of the loss of efficiency when using many processors are analyzed using a simple Monte Carlo program for criticality. The basic mechanism for parallel execution is MPI. One of the bottlenecks turn out to be the rendez-vous points in the parallel calculation used for synchronization and exchange of data between processors. This happens at least at the end of each cycle for fission source generation in order to collect the full fission source distribution for the next cycle and to estimate the effective multiplication factor, which is not only part of the requested results, but also input to the next cycle for population control. Basic improvements to overcome this limitation are suggested and tested. Also other time losses in the parallel calculation are identified. Moreover, the threading mechanism, which allows the parallel execution of tasks based on shared memory using OpenMP, is analyzed in detail. Recommendations are given to get the maximum efficiency out of a parallel Monte Carlo calculation. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗