Results for “Parallelism” — Page 41

NASA NTRSDate not supplied

Parallelized Quadrupole Simulations of Thermographic Responses of Composites

Thermography has been shown to be a viable technique for inspection of composites. Model inversion of the thermography data requires a fast method for performing the forward problem. Viable numerical methods for the thermal response forward problem are finite element, finite difference and the quadrupole method. Normally both the finite element and finite difference methods solve for the thermal response in the time domain which limits one’s ability to increase the speed of the simulation by parallelization. In contrast, the quadrupole method solves for the Laplace transform of the thermal response. One of the features of the Laplace transform methodology is the solution at any discrete time is independent of the solution at all other times. Therefore, it is easy to separate into a set of independent calculations with each of the times of interest being performed in parallel. Additionally, the numeric inversion of the Laplace transform typically involves numerically solving for the Laplace transform at multiple Laplace frequencies. Each of those solutions are also independent of solutions at other frequencies and can be calculated in parallel. By parallelization of this method, it is possible to perform the simulations of three-dimensional configurations in seconds. When the input stimulus for thermal response is a delta function heat flux (a reasonable approximation for flash heating), the thermal response is smooth. For this case, it is possible to accurately estimate the thermal response at any time within a given time interval from a set of simulations separated by exponentially increasing time steps. From these simulations, it is possible to accurately interpolate to find the response at intermediate times by a spline interpolation of the logarithm of time versus logarithm of temperature. The thermal response with exponential time stepping is shown to produce values for the thermal response which are within 1% of values within the time interval. The simulations are compared to finite element simulations of the same inspection configurations. The simulations are also compared to the thermographic measurements on composites where shape and depth of the delaminations are obtained from other inspection methods.

Thermography↗

NASA NTRSDate not supplied

Parallelized Quadrupole Simulations of Thermographic Responses of Composites

Thermography has been shown to be a viable technique for inspection of composites. Model inversion of the thermography data requires a fast method for performing the forward problem. Viable numerical methods for the thermal response forward problem are finite element, finite difference and the quadrupole method. Normally both the finite element and finite difference methods solve for the thermal response in the time domain which limits one’s ability to increase the speed of the simulation by parallelization. In contrast, the quadrupole method solves for the Laplace transform of the thermal response. One of the features of the Laplace transform methodology is the solution at any discrete time is independent of the solution at all other times. Therefore, it is easy to separate into a set of independent calculations with each of the times of interest being performed in parallel. Additionally, the numeric inversion of the Laplace transform typically involves numerically solving for the Laplace transform at multiple Laplace frequencies. Each of those solutions are also independent of solutions at other frequencies and can be calculated in parallel. By parallelization of this method, it is possible to perform the simulations of three-dimensional configurations in seconds. When the input stimulus for thermal response is a delta function heat flux (a reasonable approximation for flash heating), the thermal response is smooth. For this case, it is possible to accurately estimate the thermal response at any time within a given time interval from a set of simulations separated by exponentially increasing time steps. From these simulations, it is possible to accurately interpolate to find the response at intermediate times by a spline interpolation of the logarithm of time versus logarithm of temperature. The thermal response with exponential time stepping is shown to produce values for the thermal response which are within 1% of values within the time interval. The simulations are compared to finite element simulations of the same inspection configurations. The simulations are also compared to the thermographic measurements on composites where shape and depth of the delaminations are obtained from other inspection methods.

Thermography↗

NASA NTRS2003

UCLA parallel PIC framework

The UCLA Parallel PIC Framework (UPIC) has been developed to provide trusted components for the rapid construction of new, parallel Particle-in-Cell (PIC) codes. The Framework uses object-based ideas in Fortran95, and is designed to provide support for various kinds of PIC codes on various kinds of hardware. The focus is on student programmers. The Framework supports multiple numerical methods, different physics approximations, different numerical optimizations and implementations for different hardware. It is designed with "defensive" programming in mind, meaning that it contains many error checks and debugging helps. Above all, it is designed to hide the complexity of parallel processing. It is currently being used in a number of new Parallel PIC codes.

Norton, Charles D.↗

DOE OSTI2013

Optimized Hypervisor Scheduler for Parallel Discrete Event Simulations on Virtual Machine Platforms

With the advent of virtual machine (VM)-based platforms for parallel computing, it is now possible to execute parallel discrete event simulations (PDES) over multiple virtual machines, in contrast to executing in native mode directly over hardware as is traditionally done over the past decades. While mature VM-based parallel systems now offer new, compelling benefits such as serviceability, dynamic reconfigurability and overall cost effectiveness, the runtime performance of parallel applications can be significantly affected. In particular, most VM-based platforms are optimized for general workloads, but PDES execution exhibits unique dynamics significantly different from other workloads. Here we first present results from experiments that highlight the gross deterioration of the runtime performance of VM-based PDES simulations when executed using traditional VM schedulers, quantitatively showing the bad scaling properties of the scheduler as the number of VMs is increased. The mismatch is fundamental in nature in the sense that any fairness-based VM scheduler implementation would exhibit this mismatch with PDES runs. We also present a new scheduler optimized specifically for PDES applications, and describe its design and implementation. Experimental results obtained from running PDES benchmarks (PHOLD and vehicular traffic simulations) over VMs show over an order of magnitude improvement in the run time of the PDES-optimized scheduler relative to the regular VM scheduler, with over 20 reduction in run time of simulations using up to 64 VMs. The observations and results are timely in the context of emerging systems such as cloud platforms and VM-based high performance computing installations, highlighting to the community the need for PDES-specific support, and the feasibility of significantly reducing the runtime overhead for scalable PDES on VM platforms.

Yoginath, Srikanth B↗

DOE OSTI2014

Parallel Heuristics for Scalable Community Detection

Community detection has become a fundamental operation in numerous graph-theoretic applications. It is used to reveal natural divisions that exist within real world networks without imposing prior size or cardinality constraints on the set of communities. Despite its potential for application, there is only limited support for community detection on large-scale parallel computers, largely owing to the irregular and inherently sequential nature of the underlying heuristics. In this paper, we present parallelization heuristics for fast community detection using the Louvain method as the serial template. The Louvain method is an iterative heuristic for modularity optimization. Originally developed by Blondel et al. in 2008, the method has become increasingly popular owing to its ability to detect high modularity community partitions in a fast and memory-efficient manner. However, the method is also inherently sequential, thereby limiting its scalability to problems that can be solved on desktops. Here, we observe certain key properties of this method that present challenges for its parallelization, and consequently propose multiple heuristics that are designed to break the sequential barrier. Our heuristics are agnostic to the underlying parallel architecture. For evaluation purposes, we implemented our heuristics on shared memory (OpenMP) and distributed memory (MapReduce-MPI) machines, and tested them over real world graphs derived from multiple application domains (internet, biological, natural language processing). Experimental results demonstrate the ability of our heuristics to converge to high modularity solutions comparable to those output by the serial algorithm in nearly the same number of iterations, while also drastically reducing time to solution.

clustering, graph algorithms↗

DOE OSTI2015

Memory Scalability and Efficiency Analysis of Parallel Codes

Memory scalability is an enduring problem and bottleneck that plagues many parallel codes. Parallel codes designed for High Performance Systems are typically designed over the span of several, and in some instances 10+, years. As a result, optimization practices which were appropriate for earlier systems may no longer be valid and thus require careful optimization consideration. Specifically, parallel codes whose memory footprint is a function of their scalability must be carefully considered for future exa-scale systems. In this paper we present a methodology and tool to study the memory scalability of parallel codes. Using our methodology we evaluate an application s memory footprint as a function of scalability, which we coined memory efficiency, and describe our results. In particular, using our in-house tools we can pinpoint the specific application components which contribute to the application s overall memory foot-print (application data- structures, libraries, etc.).

Janjusic, Tommy↗

DOE OSTI2016

Distributing an executable job load file to compute nodes in a parallel computer

Distributing an executable job load file to compute nodes in a parallel computer, the parallel computer comprising a plurality of compute nodes, including: determining, by a compute node in the parallel computer, whether the compute node is participating in a job; determining, by the compute node in the parallel computer, whether a descendant compute node is participating in the job; responsive to determining that the compute node is participating in the job or that the descendant compute node is participating in the job, communicating, by the compute node to a parent compute node, an identification of a data communications link over which the compute node receives data from the parent compute node; constructing a class route for the job, wherein the class route identifies all compute nodes participating in the job; and broadcasting the executable load file for the job along the class route for the job.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2016

Distributing an executable job load file to compute nodes in a parallel computer

Distributing an executable job load file to compute nodes in a parallel computer, the parallel computer comprising a plurality of compute nodes, including: determining, by a compute node in the parallel computer, whether the compute node is participating in a job; determining, by the compute node in the parallel computer, whether a descendant compute node is participating in the job; responsive to determining that the compute node is participating in the job or that the descendant compute node is participating in the job, communicating, by the compute node to a parent compute node, an identification of a data communications link over which the compute node receives data from the parent compute node; constructing a class route for the job, wherein the class route identifies all compute nodes participating in the job; and broadcasting the executable load file for the job along the class route for the job.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2017

Reduced-Order Structure-Preserving Model for Parallel-Connected Three-Phase Grid-Tied Inverters: Preprint

Given that next-generation infrastructures will contain large numbers of grid-connected inverters and these interfaces will be satisfying a growing fraction of system load, it is imperative to analyze the impacts of power electronics on such systems. However, since each inverter model has a relatively large number of dynamic states, it would be impractical to execute complex system models where the full dynamics of each inverter are retained. To address this challenge, we derive a reduced-order structure-preserving model for parallel-connected grid-tied three-phase inverters. Here, each inverter in the system is assumed to have a full-bridge topology, LCL filter at the point of common coupling, and the control architecture for each inverter includes a current controller, a power controller, and a phase-locked loop for grid synchronization. We outline a structure-preserving reduced-order inverter model for the setting where the parallel inverters are each designed such that the filter components and controller gains scale linearly with the power rating. By structure preserving, we mean that the reduced-order three-phase inverter model is also composed of an LCL filter, a power controller, current controller, and PLL. That is, we show that the system of parallel inverters can be modeled exactly as one aggregated inverter unit and this equivalent model has the same number of dynamical states as an individual inverter in the paralleled system. Numerical simulations validate the reduced-order models.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2017

Tuning HDF5 subfiling performance on parallel file systems

Subfiling is a technique used on parallel file systems to reduce locking and contention issues when multiple compute nodes interact with the same storage target node. Subfiling provides a compromise between the single shared file approach that instigates the lock contention problems on parallel file systems and having one file per process, which results in generating a massive and unmanageable number of files. In this paper, we evaluate and tune the performance of recently implemented subfiling feature in HDF5. In specific, we explain the implementation strategy of subfiling feature in HDF5, provide examples of using the feature, and evaluate and tune parallel I/O performance of this feature with parallel file systems of the Cray XC40 system at NERSC (Cori) that include a burst buffer storage and a Lustre disk-based storage. We also evaluate I/O performance on the Cray XC30 system, Edison, at NERSC. Our results show performance benefits of 1.2X to 6X performance advantage with subfiling compared to writing a single shared HDF5 file. We present our exploration of configurations, such as the number of subfiles and the number of Lustre storage targets to storing files, as optimization parameters to obtain superior I/O performance. Based on this exploration, we discuss recommendations for achieving good I/O performance as well as limitations with using the subfiling feature.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2018

Towards a theory of economic value for modeling and simulation: incremental cost of parallel simulation (wip)

A useful perspective on the economic value of simulation in engineering comes from considering the role simulation plays in constructing a system. We propose a conceptual model that combines an embryonic concept of the value of an information bit with cost estimation tools used in software engineering. The model provides an economic justification for the common conception that models should be built to address specific questions rather than to comprehensively represent a referent system. Extending these ideas, we offer a model for the incremental value of parallel execution to explain why simulation applications that take advantage of sophisticated parallel discrete event simulation tools form a relatively narrow niche despite the ubiquity of parallel computing hardware. The proposed model enables us to bring value of information criteria into model development and to make predictions about the future of parallel simulation applications.

Nutaro, James J↗

DOE OSTI2018

Single-pass parallel prefix scan with dynamic look back

One embodiment of the present invention performs a parallel prefix scan in a single pass that incorporates variable look-back. A parallel processing unit (PPU) subdivides a list of inputs into sequentially-ordered segments and assigns each segment to a streaming multiprocessor (SM) included in the PPU. Notably, the SMs may operate in parallel. Each SM executes write operations on a segment descriptor that includes the status, aggregate, and inclusive-prefix associated with the assigned segment. Further, each SM may execute read operations on segment descriptors associated with other segments. In operation, each SM may perform reduction operations to determine a segment-wide aggregate, may perform look-back operations across multiple preceding segments to determine an exclusive-prefix, and may perform a scan seeded with the exclusive prefix to generate output data. Advantageously, the PPU performs one read operation per input, thereby reducing the time required to execute the prefix scan relative to prior-art parallel implementations.

Merrill, Duane↗

DOE OSTI2019

Temperature control to reduce capacity mismatch in parallel-connected lithium ion cells

The temperature and capacity of individual cells affect the current distribution in a battery pack. Non uniform current distribution among parallel-connected cells can lead to capacity imbalance and premature aging. This paper develops models that calculate the current in parallel-connected cells and predict their capacity fade. The model is validated experimentally for a nonuniform battery pack at different temperatures. The paper also proposes and validates the hypothesis that temperature control can reduce capacity mismatch in parallel-connected cells. Three Lithium Iron Phosphate cells, two cells at higher initial capacity than the third cell, are connected in parallel. The pack is cycled for 1500 Hybrid Electric Vehicles cycles with the higher capacity cells regulated at 40 C and the lower capacity cell at 20 C. As predicted by the model, the higher capacity and temperature cells age faster, reducing the capacity mismatch by 48% over the 1500 cycles. A case study shows that cooling of low capacity cells can reduce capacity mismatch and extend pack life.

25 ENERGY STORAGE↗

DOE OSTI2019

In situ particle advection via parallelizing over particles

We extend the method for particle advection that parallelizes over particles to work in an in situ setting. We then compare our method with the typical method for in situ, parallelizing over data. Our experiments consist of parallelism at 512 cores, a data set with 67 million cells, and ten billion total advection steps. Our findings show that parallelizing over particles can be more than ten times faster for some workloads, for reasonable memory cost. Overall, the significance of these findings is to demonstrate that moving data can be worthwhile in some in situ settings.

Pugmire, Dave↗

DOE OSTI2020

Parallel Particle Advection Bake-Off for Scientific Visualization Workloads

There are multiple algorithms for parallelizing particle advection for scientific visualization workloads. While many previous studies have contributed to the understanding of individual algorithms, our study aims to provide a holistic understanding of how algorithms perform relative to each other on various workloads. To accomplish this, we consider four popular parallelization algorithms and run a “bake-off” study (i.e., an empirical study) to identify the best matches for each. The study includes 216 tests, going to a concurrency of up to 8192 cores and considering data sets as large as 34 billion cells with 300 million particles. Overall, our study informs three important research questions: (1) which parallelization algorithms perform best for a given workload?, (2) why?, and (3) what are the unsolved problems in parallel particle advection? In terms of findings, we find that the seeding box is the most important factor in choosing the best algorithm, and also that there is a significant opportunity for improvement in execution time, scalability, and efficiency.

Pugmire, Dave↗

DOE OSTI2022

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↗

DOE OSTI2012

A parallel in energy Krylov solution algorithm for the variational nodal method

To reduce the computational expense caused by the large number of energy groups in thermal and fast reactor calculations, we researched using parallelization in energy (i.e. group) of the multi-group response matrix equations derived from the Variational Nodal Method. This study focuses on solving steady-state eigenvalue problems in diffusion theory that are typically needed for depletion and reactivity coefficient calculations. To examine serial and parallel performance, GMRES, a Krylov algorithm, is applied to the multi-group system of equations at each fission source iteration. We also study the impact of using Wielandt acceleration on the eigenvalue/eigenvector problem. Numerical results based on a 72 group thermal and a 216 group fast reactor core problems are obtained for both serial and parallel calculations. The results show reduced CPU times when using GMRES and Wielandt acceleration in both serial and parallel calculations. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2016

On the Degradation of the Effectiveness of Nonlinear Diffusion Acceleration with Parallel Block Jacobi Splitting

MOOSE's radiation transport module Rattlesnake comprises a first order S N solver that can be accelerated with a flexible Nonlinear Diffusion Acceleration. At each NDA iteration, the streaming and collision operator of the S{sub N} equations are (approximately) inverted using a transport sweep. When executed on multiple processors parallel block Jacobi (PBJ) splitting is utilized lagging the processor sub-domain's incoming angular fluxes making the implemented method asynchronous, i.e. the convergence properties of NDA become dependent on the number of concurrent tasks and details of the partitioning. Within this work, we study the NDA convergence properties for the two-dimensional C5G7 benchmark problem PBJ with p = 1, 2, 4, 8, 12, 16, 24, 32, 48, 60, 72, 96 and 144 concurrent tasks. Rosa studied the convergence properties of PBJ splitting methods with and without transport synthetic acceleration (TSA) in slab geometry and two-dimensional Cartesian geometry, respectively. Rosa's iteration scheme lags both the incoming angular fluxes and the scattering source making it very similar to the method described here. Rosa found by means of Fourier analysis that convergence of unaccelerated PBJ can degrade for optically thin subdomains, i.e. for a large number of processors given a fixed domain size. Depending on the choice of parameters application of TSA can lead to an effective algorithm. In contrast to Rosa's work, this work focuses on PBJ splitting for NDA accelerated solution of the multigroup eigenvalue problem. Later, Zerr and Anistratov used their integral transport matrix method in conjunction with PBJ. This iteration scheme is different as it does not lag the scattering source. Nevertheless, Fourier analysis with Diamond Difference spatial discretization showed that the method degrades with increasing the overall size of the spatial domain while keeping the sub-domain size constant. In the case of an infinite medium, the observed spectral radius becomes unity. These references establish the iterative performance of PBJ for one-group fixed source problems discretized on rectangular cells, but for the Rattlesnake first order NDA method, an emphasis lies on multigroup eigenvalue problems. In this setting, the additional complication arises that the low order equations are not exactly inverted, but only to some iterative criterion. Further, as the low order problem is solved using Jacobian-Free Newton Krylov methods, a suitable preconditioner for the GMRES iterations is required. As the effectiveness of the preconditioners may decrease when more concurrent processes are used, we expect the low order solve to impact the overall effectiveness of the NDA method in parallel execution as well. Therefore, this work focuses on the question of how detrimental the asynchronicity of the high and low order iterative solutions is for the effectiveness of NDA solutions of realistic, unstructured grid, multigroup eigenvalue problems. The iterative convergence properties of Rattlesnake's NDA method for multigroup eigenvalue problems are studied for varying number of concurrent processes. Rattlesnake's NDA method is asynchronous because the high order transport sweep uses PBJ splitting and because the low order solve's preconditioners, algebraic multigrid and additive Schwarz, are asynchronous. We found that up to 144 processes the number of Picard iterations increases by a factor of 2-3 affecting but not destroying NDA's parallel performance for the C5G7 benchmark problem, a representative LWR (light water reactor) configuration. The solution of the NDA low order eigenvalue problem posed greater problems than expected for the solvers currently available in Rattlesnake mainly because of problems to effectively precondition the modified low-order problem. As a matter of fact, the low order problem becomes an advection-diffusion problem. Non-robustness was observed for the AMG preconditioned coarse mesh acceleration for more than 60 concurrent processes, which was fixed by increasing the number of permissible V-cycles per AMG call. For fine mesh acceleration AMG proved to be ineffective and we resorted to ASM. For both AMG and ASM, increasing number of linear iterations per nonlinear are observed when increasing the number of processes, but for ASM this problem appears to be worse. The results from this work also indicate that initial development efforts should be diverted from implementation of a high-order parallel sweeping algorithm to the effective and robust iterative solution of the low order NDA multigroup eigenvalue problem. (authors)

97 MATHEMATICS AND COMPUTING↗