Search NASA⌕ Search

SEARCH · Search NASA

Results for “distributed parallelization”

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 379 records · Page 21

The alignment-distribution graph

Implementing a data-parallel language such as Fortran 90 on a distributed-memory parallel computer requires distributing aggregate data objects (such as arrays) among the memory modules attached to the processors. The mapping of objects to the machine determines the amount of residual communication needed to bring operands of parallel operations into alignment with each other. We present a program representation called the alignment distribution graph that makes these communication requirements explicit. We describe the details of the representation, show how to model communication cost in this framework, and outline several algorithms for determining object mappings that approximately minimize residual communication.

Chatterjee, Siddhartha↗

The alignment-distribution graph

Implementing a data-parallel language such as Fortran 90 on a distributed-memory parallel computer requires distributing aggregate data objects (such as arrays) among the memory modules attached to the processors. The mapping of objects to the machine determines the amount of residual communication needed to bring operands of parallel operations into alignment with each other. We present a program representation called the alignment-distribution graph that makes these communication requirements explicit. We describe the details of the representation, show how to model communication cost in this framework, and outline several algorithms for determining object mappings that approximately minimize residual communication.

Chatterjee, Siddhartha↗

Scalable training of graph convolutional neural networks for fast and accurate predictions of HOMO-LUMO gap in molecules

Abstract Graph Convolutional Neural Network (GCNN) is a popular class of deep learning (DL) models in material science to predict material properties from the graph representation of molecular structures. Training an accurate and comprehensive GCNN surrogate for molecular design requires large-scale graph datasets and is usually a time-consuming process. Recent advances in GPUs and distributed computing open a path to reduce the computational cost for GCNN training effectively. However, efficient utilization of high performance computing (HPC) resources for training requires simultaneously optimizing large-scale data management and scalable stochastic batched optimization techniques. In this work, we focus on building GCNN models on HPC systems to predict material properties of millions of molecules. We use HydraGNN, our in-house library for large-scale GCNN training, leveraging distributed data parallelism in PyTorch. We use ADIOS, a high-performance data management framework for efficient storage and reading of large molecular graph data. We perform parallel training on two open-source large-scale graph datasets to build a GCNN predictor for an important quantum property known as the HOMO-LUMO gap. We measure the scalability, accuracy, and convergence of our approach on two DOE supercomputers: the Summit supercomputer at the Oak Ridge Leadership Computing Facility (OLCF) and the Perlmutter system at the National Energy Research Scientific Computing Center (NERSC). We present our experimental results with HydraGNN showing (i) reduction of data loading time up to 4.2 times compared with a conventional method and (ii) linear scaling performance for training up to 1024 GPUs on both Summit and Perlmutter.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

The mass of the Coma cluster

Mass estimate uncertainties for the Coma cluster in position and velocity data for the galaxies are examined under the assumption that mass and galaxy distribution are parallel, on the one hand, and on the other, that the form of the mass distribution is not constrained. While low mass models have all the mass in the cluster core, and require the galaxies to be on or near circular orbits, high mass models are noted to have weakly concentrated mass distributions and primarily radial orbits. It is concluded that galaxy data alone are unlikely to clearly distinguish between these possibilities.

The, Lih Sin↗

MEDUSA - An overset grid flow solver for network-based parallel computer systems

Continuing improvement in processing speed has made it feasible to solve the Reynolds-Averaged Navier-Stokes equations for simple three-dimensional flows on advanced workstations. Combining multiple workstations into a network-based heterogeneous parallel computer allows the application of programming principles learned on MIMD (Multiple Instruction Multiple Data) distributed memory parallel computers to the solution of larger problems. An overset-grid flow solution code has been developed which uses a cluster of workstations as a network-based parallel computer. Inter-process communication is provided by the Parallel Virtual Machine (PVM) software. Solution speed equivalent to one-third of a Cray-YMP processor has been achieved from a cluster of nine commonly used engineering workstation processors. Load imbalance and communication overhead are the principal impediments to parallel efficiency in this application.

Smith, Merritt H.↗

Triangle Counting with Cyclic Distributions

Triangles are the simplest non-trivial subgraphs and triangle counting is used in a number of different applications. The order in which vertices are processed in triangle counting strongly effects the amount of work that needs to be done (and thus the overall performance). Ordering vertices by degree has been shown to be one particularly effective ordering approach. However, for graphs with skewed degree distributions (such as power-law graphs), ordering by degree effects the distribution of work; parallelization must account for this distribution in order to balance work among workers. In this paper we provide an in- depth analysis of the ramifications of degree-based ordering on parallel triangle counting. We present approach for partitioning work in triangle counting, based on cyclic distribution and some surprisingly simple C++ implementations. Experimental results demonstrate the effectiveness of our approach, particularly for power-law (and social network) graphs.

Graph algorithms, parallel algorithms↗

Automatic data partitioning on distributed memory multicomputers

Distributed-memory parallel computers are increasingly being used to provide high levels of performance for scientific applications. Unfortunately, such machines are not very easy to program. A number of research efforts seek to alleviate this problem by developing compilers that take over the task of generating communication. The communication overheads and the extent of parallelism exploited in the resulting target program are determined largely by the manner in which data is partitioned across different processors of the machine. Most of the compilers provide no assistance to the programmer in the crucial task of determining a good data partitioning scheme. A novel approach is presented, the constraints-based approach, to the problem of automatic data partitioning for numeric programs. In this approach, the compiler identifies some desirable requirements on the distribution of various arrays being referenced in each statement, based on performance considerations. These desirable requirements are referred to as constraints. For each constraint, the compiler determines a quality measure that captures its importance with respect to the performance of the program. The quality measure is obtained through static performance estimation, without actually generating the target data-parallel program with explicit communication. Each data distribution decision is taken by combining all the relevant constraints. The compiler attempts to resolve any conflicts between constraints such that the overall execution time of the parallel program is minimized. This approach has been implemented as part of a compiler called Paradigm, that accepts Fortran 77 programs, and specifies the partitioning scheme to be used for each array in the program. We have obtained results on some programs taken from the Linpack and Eispack libraries, and the Perfect Benchmarks. These results are quite promising, and demonstrate the feasibility of automatic data partitioning for a significant class of scientific application programs with regular computations.

Gupta, Manish↗

T-FSM: A Scalable Distributed Task-Based System for Frequent Subgraph Pattern Mining from a Big Graph

Finding frequent subgraph patterns in a big graph is an important problem with many applications such as classifying chemical compounds and building indexes to speed up graph queries. Since this problem is NP-hard, some recent parallel and distributed systems have been developed to accelerate the mining. However, they often have a huge memory cost, very long running time, suboptimal load balancing, poor scale-out capability, and possibly inaccurate results. In this article, we propose an efficient system called T-FSM for parallel mining of frequent subgraph patterns in a big graph. T-FSM supports a new anti-monotonic frequentness measure called Fraction-Score, which is more accurate than the widely used MNI measure. The execution engine of T-FSM supports both intra-machine parallelism and inter-machine parallelism. For intra-machine parallelism, T-FSM adopts a novel task-based execution model to ensure high multithreading concurrency, bounded memory consumption, and effective load balancing. For inter-machine parallelism, T-FSM ensures good scale-out performance with a lightweight pattern rebalancing approach that reduces workload skewness of pattern evaluations among machines. To avoid recomputing the contexts for migrated patterns, we design a novel context cache table to support concurrent and asynchronous requesting and caching of remote context data, which can timely evict and garbage collect used pattern contexts that are no longer needed to keep memory consumption bounded. Extensive experiments show that T-FSM is orders of magnitude faster than existing state-of-the-art parallel systems (more than 10×, 51×, 131×, 55× speedup over ScaleMine, DistGraph, Pangolin and Peregrine, respectively) and distributed systems (more than 42× and 88× over ScaleMine and DistGraph, respectively) for frequent subgraph pattern mining, and it scales out satisfactorily to 512 CPU cores on the Polaris supercomputer at Argonne National Laboratory.

97 MATHEMATICS AND COMPUTING↗

Generation of Highly Oblique Lower Band Chorus Via Nonlinear Three-Wave Resonance

Chorus in the inner magnetosphere has been observed frequently at geomagnetically active times, typically exhibiting a two-band structure with a quasi-parallel lower band and an upper band with a broad range of wave normal angles. But recent observations by Van Allen Probes confirm another type of lower band chorus, which has a large wave normal angle close to the resonance cone angle. It has been proposed that these waves could be generated by a low-energy beam-like electron component or by temperature anisotropy of keV electrons in the presence of a low-energy plateau-like electron component. This paper, however, presents an alternative mechanism for generation of this highly oblique lower band chorus. Through a nonlinear three-wave resonance, a quasi-parallel lower band chorus wave can interact with a mildly oblique upper band chorus wave, producing a highly oblique quasi-electrostatic lower band chorus wave. This theoretical analysis is confirmed by 2-D electromagnetic particle-in-cell simulations. Furthermore, as the newly generated waves propagate away from the equator, their wave normal angle can further increase and they are able to scatter low-energy electrons to form a plateau-like structure in the parallel velocity distribution. As a result, the three-wave resonance mechanism may also explain the generation of quasi-parallel upper band chorus which has also been observed in the magnetosphere.

58 GEOSCIENCES↗

Distributed-Memory Computing With the Langley Aerothermodynamic Upwind Relaxation Algorithm (LAURA)

The Langley Aerothermodynamic Upwind Relaxation Algorithm (LAURA), a Navier-Stokes solver, has been modified for use in a parallel, distributed-memory environment using the Message-Passing Interface (MPI) standard. A standard domain decomposition strategy is used in which the computational domain is divided into subdomains with each subdomain assigned to a processor. Performance is examined on dedicated parallel machines and a network of desktop workstations. The effect of domain decomposition and frequency of boundary updates on performance and convergence is also examined for several realistic configurations and conditions typical of large-scale computational fluid dynamic analysis.

Riley, Christopher J.↗

Xyce parallel electronic simulator design.

This document is the Xyce Circuit Simulator developer guide. Xyce has been designed from the 'ground up' to be a SPICE-compatible, distributed memory parallel circuit simulator. While it is in many respects a research code, Xyce is intended to be a production simulator. As such, having software quality engineering (SQE) procedures in place to insure a high level of code quality and robustness are essential. Version control, issue tracking customer support, C++ style guildlines and the Xyce release process are all described. The Xyce Parallel Electronic Simulator has been under development at Sandia since 1999. Historically, Xyce has mostly been funded by ASC, the original focus of Xyce development has primarily been related to circuits for nuclear weapons. However, this has not been the only focus and it is expected that the project will diversify. Like many ASC projects, Xyce is a group development effort, which involves a number of researchers, engineers, scientists, mathmaticians and computer scientists. In addition to diversity of background, it is to be expected on long term projects for there to be a certain amount of staff turnover, as people move on to different projects. As a result, it is very important that the project maintain high software quality standards. The point of this document is to formally document a number of the software quality practices followed by the Xyce team in one place. Also, it is hoped that this document will be a good source of information for new developers.

99 GENERAL AND MISCELLANEOUS↗

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↗

Whistler anisotropy instability at low electron {beta}: Particle-in-cell simulations

The whistler anisotropy instability is studied in a magnetized, homogeneous, collisionless plasma model. The electrons (denoted by subscript e) are represented initially with a single bi-Maxwellian velocity distribution with a temperature anisotropy T{sub perpendiculare}/T{sub ||e}>1, where perpendicular and || denote directions perpendicular and parallel to the background magnetic field B{sub o}, respectively. Kinetic linear dispersion theory predicts that, if the ratio of the electron plasma frequency {omega}{sub e} to the electron cyclotron frequency {Omega}{sub e} is greater than unity and {beta}{sub ||e}{>=}0.025, the maximum growth rate of this instability is at parallel propagation, where the fluctuating fields are strictly electromagnetic. At smaller values of {beta}{sub ||e}, however, the maximum growth rate shifts to propagation oblique to B{sub o} and the fluctuating electric fields become predominantly electrostatic. Linear theory and two-dimensional particle-in-cell simulations are used to examine the consequences of this transition. Three simulations are carried out, with initial {beta}{sub ||e}=0.10, 0.03, and 0.01. The fluctuating fields of the {beta}{sub ||e}=0.10 run are predominantly electromagnetic, with nonlinear consequences similar to those of simulations already described in the literature. In contrast, the growth of fluctuations at oblique propagation in the low electron {beta} runs leads to a significant {delta}E{sub ||}, which heats the electrons leading to the formation of a substantial suprathermal component in the electron parallel velocity distribution.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Experiences with hypercube operating system instrumentation

The difficulties in conceptualizing the interactions among a large number of processors make it difficult both to identify the sources of inefficiencies and to determine how a parallel program could be made more efficient. This paper describes an instrumentation system that can trace the execution of distributed memory parallel programs by recording the occurrence of parallel program events. The resulting event traces can be used to compile summary statistics that provide a global view of program performance. In addition, visualization tools permit the graphic display of event traces. Visual presentation of performance data is particularly useful, indeed, necessary for large-scale parallel computers; the enormous volume of performance data mandates visual display.

Reed, Daniel A.↗

Parallel computation of three-dimensional flows using overlapping grids with adaptive mesh refinement

This paper describes an approach for the numerical solution of time-dependent partial differential equations in complex three-dimensional domains. The domains are represented by overlapping structured grids, and block-structured adaptive mesh refinement (AMR) is employed to locally increase the grid resolution. In addition, the numerical method is implemented on parallel distributed-memory computers using a domain-decomposition approach. The implementation is flexible so that each base grid within the overlapping grid structure and its associated refinement grids can be independently partitioned over a chosen set of processors. A modified bin-packing algorithm is used to specify the partition for each grid so that the computational work is evenly distributed amongst the processors. All components of the AMR algorithm such as error estimation, regridding, and interpolation are performed in parallel. The parallel time-stepping algorithm is illustrated for initial-boundary-value problems involving a linear advection-diffusion equation and the (nonlinear) reactive Euler equations. Numerical results are presented for both equations to demonstrate the accuracy and correctness of the parallel approach. Exact solutions of the advection-diffusion equation are constructed, and these are used to check the corresponding numerical solutions for a variety of tests involving different overlapping grids, different numbers of refinement levels and refinement ratios, and different numbers of processors. The problem of planar shock diffraction by a sphere is considered as an illustration of the numerical approach for the Euler equations, and a problem involving the initiation of a detonation from a hot spot in a T-shaped pipe is considered to demonstrate the numerical approach for the reactive case. For both problems, the accuracy of the numerical solutions is assessed quantitatively through an estimation of the errors from a grid convergence study. The parallel performance of the approach is examined for the shock diffraction problem.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Asynchronous In Situ Connected-Components Analysis for Complex Fluid flows, In: ISAV '16 Proceedings of the 2nd Workshop on In Situ Infrastructures for Enabling Extreme-scale Analysis and Visualization

The simulation of multiscale physics is an important challenge for scientific computing. For this class of problem, large three-dimensional simulations are performed to advance scientific inquiry. On massively parallel computing systems, the volume of data generated by such approaches can become a productivity bottleneck if the raw data generated from the simulation is analyzed in a post-processing step. To address this, we present a physics-based framework for in situ data reduction that is theoretically grounded in multiscale averaging theory. We show how task parallelism can be exploited to concurrently perform a variety of analysis tasks with data-dependent costs, including the generation of iso-surfaces, morphological analyses, and connected components analysis. All analyses are performed in parallel using distributed memory and use the same domain decomposition as the simulation. A task management framework is constructed to leverage available parallelism within a node for analysis. The capabilities of the framework are to launch asynchronous analysis threads, manage dependencies between different tasks, promote data locality and minimize the impact of data transfers. The framework is applied to analyze GPU-based simulations of two-fluid-phase flow in porous media, generating a set of averaged measures that represents the overall system behavior. We demonstrate how the approach can be applied to perform physically-consistent analysis over fluid sub-regions determined from connected components analysis. Simulations performed on Oak Ridge National Lab's Titan supercomputer are profiled to demonstrate the performance of the associated multi-threaded in situ analysis approach for typical production simulation of two-fluid-phase flow.

Computer Science↗

toolkit computational mesh conceptual model.

The Sierra Toolkit computational mesh is a software library intended to support massively parallel multi-physics computations on dynamically changing unstructured meshes. This domain of intended use is inherently complex due to distributed memory parallelism, parallel scalability, heterogeneity of physics, heterogeneous discretization of an unstructured mesh, and runtime adaptation of the mesh. Management of this inherent complexity begins with a conceptual analysis and modeling of this domain of intended use; i.e., development of a domain model. The Sierra Toolkit computational mesh software library is designed and implemented based upon this domain model. Software developers using, maintaining, or extending the Sierra Toolkit computational mesh library must be familiar with the concepts/domain model presented in this report.

97 MATHEMATICS AND COMPUTING↗

Electrostatic Particle-In-Cell Code For Hypercube Computer

Code simulates two-dimensional motions of plasma particles in self-consistent electrostatic field and externally applied magnetic field developed for execution on Mark IIIfp hypercube computer. Based on generalization of one-dimensional particle-in-cell algorithm applicable to many different parallel computing architectures. Intermediate product of continuing effort to speed particle-in-cell computations by taking advantage of distributed-memory parallel computers like those of hypercube class.

Ferraro, Robert D.↗