Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Parallel applications”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,351 records · Page 75

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↗

RIACS

Topics considered include: high-performance computing; cognitive and perceptual prostheses (computational aids designed to leverage human abilities); autonomous systems. Also included: development of a 3D unstructured grid code based on a finite volume formulation and applied to the Navier-stokes equations; Cartesian grid methods for complex geometry; multigrid methods for solving elliptic problems on unstructured grids; algebraic non-overlapping domain decomposition methods for compressible fluid flow problems on unstructured meshes; numerical methods for the compressible navier-stokes equations with application to aerodynamic flows; research in aerodynamic shape optimization; S-HARP: a parallel dynamic spectral partitioner; numerical schemes for the Hamilton-Jacobi and level set equations on triangulated domains; application of high-order shock capturing schemes to direct simulation of turbulence; multicast technology; network testbeds; supercomputer consolidation project.

Oliger, Joseph↗

Initial Performance Results on IBM POWER6

The POWER5+ processor has a faster memory bus than that of the previous generation POWER5 processor (533 MHz vs. 400 MHz), but the measured per-core memory bandwidth of the latter is better than that of the former (5.7 GB/s vs. 4.3 GB/s). The reason for this is that in the POWER5+, the two cores on the chip share the L2 cache, L3 cache and memory bus. The memory controller is also on the chip and is shared by the two cores. This serializes the path to memory. For consistently good performance on a wide range of applications, the performance of the processor, the memory subsystem, and the interconnects (both latency and bandwidth) should be balanced. Recognizing this, IBM has designed the Power6 processor so as to avoid the bottlenecks due to the L2 cache, memory controller and buffer chips of the POWER5+. Unlike the POWER5+, each core in the POWER6 has its own L2 cache (4 MB - double that of the Power5+), memory controller and buffer chips. Each core in the POWER6 runs at 4.7 GHz instead of 1.9 GHz in POWER5+. In this paper, we evaluate the performance of a dual-core Power6 based IBM p6-570 system, and we compare its performance with that of a dual-core Power5+ based IBM p575+ system. In this evaluation, we have used the High- Performance Computing Challenge (HPCC) benchmarks, NAS Parallel Benchmarks (NPB), and four real-world applications--three from computational fluid dynamics and one from climate modeling.

Saini, Subbash↗

The Role of Multiphysics Simulation in Multidisciplinary Analysis

This article describes the applications of the Spectrum(Tm) Solver in Multidisciplinary Analysis (MDA). Spectrum, a multiphysics simulation software based on the finite element method, addresses compressible and incompressible fluid flow, structural, and thermal modeling as well as the interaction between these disciplines. Multiphysics simulation is based on a single computational framework for the modeling of multiple interacting physical phenomena. Interaction constraints are enforced in a fully-coupled manner using the augmented-Lagrangian method. Within the multiphysics framework, the finite element treatment of fluids is based on Galerkin-Least-Squares (GLS) method with discontinuity capturing operators. The arbitrary-Lagrangian-Eulerian method is utilized to account for deformable fluid domains. The finite element treatment of solids and structures is based on the Hu-Washizu variational principle. The multiphysics architecture lends itself naturally to high-performance parallel computing. Aeroelastic, propulsion, thermal management and manufacturing applications are presented.

Rifai, Steven M.↗

VisIO: enabling interactive visualization of ultra-scale, time-series data via high-bandwidth distributed I/O systems

Petascale simulations compute at resolutions ranging into billions of cells and write terabytes of data for visualization and analysis. Interactive visuaUzation of this time series is a desired step before starting a new run. The I/O subsystem and associated network often are a significant impediment to interactive visualization of time-varying data; as they are not configured or provisioned to provide necessary I/O read rates. In this paper, we propose a new I/O library for visualization applications: VisIO. Visualization applications commonly use N-to-N reads within their parallel enabled readers which provides an incentive for a shared-nothing approach to I/O, similar to other data-intensive approaches such as Hadoop. However, unlike other data-intensive applications, visualization requires: (1) interactive performance for large data volumes, (2) compatibility with MPI and POSIX file system semantics for compatibility with existing infrastructure, and (3) use of existing file formats and their stipulated data partitioning rules. VisIO, provides a mechanism for using a non-POSIX distributed file system to provide linear scaling of 110 bandwidth. In addition, we introduce a novel scheduling algorithm that helps to co-locate visualization processes on nodes with the requested data. Testing using VisIO integrated into Para View was conducted using the Hadoop Distributed File System (HDFS) on TACC's Longhorn cluster. A representative dataset, VPIC, across 128 nodes showed a 64.4% read performance improvement compared to the provided Lustre installation. Also tested, was a dataset representing a global ocean salinity simulation that showed a 51.4% improvement in read performance over Lustre when using our VisIO system. VisIO, provides powerful high-performance I/O services to visualization applications, allowing for interactive performance with ultra-scale, time-series data.

97 MATHEMATICS AND COMPUTING↗

Relationships among classes of self-oscillating transistor parallel inverters

It recently has been shown in the literature that many dc-to-square-wave parallel inverters which are widely used in power-conditioning applications can be grouped into one of two families. Each family is characterized by an equivalent RLC network. Based on this approach, a classification procedure is presented for self-oscillating parallel inverters which makes evident natural relationships which exist between various inverter configurations. By utilizing concepts from the basic theory of negative resistance oscillators and the principle of duality as applied to nonlinear networks, a chain of relationships is established which enables a methodical transfer of knowledge gained about one family of inverters to any of the other families in the classification array.

Wilson, T. G.↗

A Parallel Ghosting Algorithm for The Flexible Distributed Mesh Database

Critical to the scalability of parallel adaptive simulations are parallel control functions including load balancing, reduced inter-process communication and optimal data decomposition. In distributed meshes, many mesh-based applications frequently access neighborhood information for computational purposes which must be transmitted efficiently to avoid parallel performance degradation when the neighbors are on different processors. This article presents a parallel algorithm of creating and deleting data copies, referred to as ghost copies, which localize neighborhood data for computation purposes while minimizing inter-process communication. The key characteristics of the algorithm are: (1) It can create ghost copies of any permissible topological order in a 1D, 2D or 3D mesh based on selected adjacencies. (2) It exploits neighborhood communication patterns during the ghost creation process thus eliminating all-to-all communication. (3) For applications that need neighbors of neighbors, the algorithm can create n number of ghost layers up to a point where the whole partitioned mesh can be ghosted. Strong and weak scaling results are presented for the IBM BG/P and Cray XE6 architectures up to a core count of 32,768 processors. The algorithm also leads to scalable results when used in a parallel super-convergent patch recovery error estimator, an application that frequently accesses neighborhood data to carry out computation.

Mubarak, Misbah↗

An application of small-gap equations in sealing devices

The study of a thin, incompressible Newtonian fluid layer trapped between two almost parallel, sliding surfaces has been actively pursued in the last decades. This subject includes lubrication applications such as slider bearings or the sealing of non-pressurized fluids with rubber rotary shaft seals. In the present work we analyze numerically the flow of lubricant fluid through a micro-gap of sealing devices. The first stage of this study is carried out assuming that a 'small-gap' parameter delta attains an extreme value in the Navier-Stokes equations. The precise meaning of small-gap is achieved by the particular limit delta = 0 which, within the bounds of the hypotheses, predicts transport of lubricant through the sealed area by centrifugal instabilities. Numerical results obtained with the penalty function approximation in the finite element method are presented. In particular, the influence of inflow and outflow boundary conditions, and their impact in the simulated flow, are discussed.

Vionnet, Carlos A.↗

Overview of ICE Project: Integration of Computational Fluid Dynamics and Experiments

Researchers at the NASA Glenn Research Center have developed a prototype integrated environment for interactively exploring, analyzing, and validating information from computational fluid dynamics (CFD) computations and experiments. The Integrated CFD and Experiments (ICE) project is a first attempt at providing a researcher with a common user interface for control, manipulation, analysis, and data storage for both experiments and simulation. ICE can be used as a live, on-tine system that displays and archives data as they are gathered; as a postprocessing system for dataset manipulation and analysis; and as a control interface or "steering mechanism" for simulation codes while visualizing the results. Although the full capabilities of ICE have not been completely demonstrated, this report documents the current system. Various applications of ICE are discussed: a low-speed compressor, a supersonic inlet, real-time data visualization, and a parallel-processing simulation code interface. A detailed data model for the compressor application is included in the appendix.

Stegeman, James D.↗

SU-E-T-783: Using Matrixx to Determine Transit Dose Contribution Over Clinically Useful Limits of HDR Source Activity

Purpose: Most HDR brachytherapy treatment planning systems (TPS) use TG-43 formalism to calculate dose without including transit dose corrections. Historically, measurement of this contribution has required sophisticated apparatus unavailable in most hospitals. We use Matrixx to investigate several scenarios where transit dose contribution may effect a clinical treatment. Methods: Treatment plans were generated using Oncentra Brachy TPS (Version 4.3.0.410, Nucletron ) on a CT scan of a 24-catheter Freiburg applicator (Nucletron ) laid flat on the MatriXX (IBA) detector. This detector is an array of 1020 parallel plate ion chambers. All 24 catheters were digitized and dwells within a central square region of 5×5cm of the applicator were activated. Each of the active catheters had 6 dwells in increments of 1.0cm. The plans were normalized to 10mm. This places the 100% isodose line at the correct effective point of measurement, which lies half-way between the parallel plates of the ion chambers. It is also within the clinically relevant treatment depth for superficial applications. A total of 6 plans were delivered for 3 prescription doses, 1Gy, 2Gy and 4Gy using source activities of 2.9Ci and 11.2Ci. The MatriXX array was operated to capture dosimetric snaps every 500ms and yielded an integral dose at the end of treatment. Results: A comparison of integral dose from 2 different source activities shows that the transit dose contribution is larger when the source activity is higher. It is also observed that the relative transit dose contribution decreases as prescription dose increases. This is quantified by the Gamma analysis. Conclusion: We have demonstrated that the Matrixx detector can be used to evaluate the contribution for a HDR source during transit from the HDR afterloader to a dwell location, and between adjacent dwell locations.

60 APPLIED LIFE SCIENCES↗

Multiprocessor graphics computation and display using transputers

A package of two-dimensional graphics routines was developed to run on a transputer-based parallel processing system. These routines were designed to enable applications programmers to easily generate and display results from the transputer network in a graphic format. The graphics procedures were designed for the lowest possible network communication overhead for increased performance. The routines were designed for ease of use and to present an intuitive approach to generating graphics on the transputer parallel processing system.

Ellis, Graham K.↗

A massively parallel method of characteristic neutral particle transport code for GPUs

Over the past 20 years, parallel computing has enabled computers to grow ever larger and more powerful while scientific applications have advanced in sophistication and resolution. This trend is being challenged, however, as the power consumption for conventional parallel computing architectures has risen to unsustainable levels and memory limitations have come to dominate compute performance. Heterogeneous computing platforms, such as Graphics Processing Units (GPUs), are an increasingly popular paradigm for solving these issues. This paper explores the applicability of GPUs for deterministic neutron transport. A 2D method of characteristics (MOC) code - OpenMOC - has been developed with solvers for both shared memory multi-core platforms as well as GPUs. The multi-threading and memory locality methodologies for the GPU solver are presented. Performance results for the 2D C5G7 benchmark demonstrate 25-35 x speedup for MOC on the GPU. The lessons learned from this case study will provide the basis for further exploration of MOC on GPUs as well as design decisions for hardware vendors exploring technologies for the next generation of machines for scientific computing. (authors)

97 MATHEMATICS AND COMPUTING↗

Concurrency-based approaches to parallel programming

The inevitable transition to parallel programming can be facilitated by appropriate tools, including languages and libraries. After describing the needs of applications developers, this paper presents three specific approaches aimed at development of efficient and reusable parallel software for irregular and dynamic-structured problems. A salient feature of all three approaches in their exploitation of concurrency within a processor. Benefits of individual approaches such as these can be leveraged by an interoperability environment which permits modules written using different approaches to co-exist in single applications.

Kale, L.V.↗

Trigger-based Incremental Data Processing with Unified Sync and Async Model

In recent years, more and more applications in the cloud have needs to process large-scale on-line datasets, which evolve over time as new entries are added and existing entries are modified. Several programming frameworks, such as Percolator and Oolong, are proposed for such incremental data processing and can achieve efficient processing with an event-driven abstraction. However, these frameworks are inherently asynchronous, leaving the heavy burden of managing synchronization to applications' developers, which further significantly restricts their usabilities. In this study, we propose a trigger-based incremental computing framework in the cloud, called Domino, with both synchronous and asynchronous mechanisms to coordinate parallel triggers. With this new framework, both synchronous and asynchronous applications can be seamlessly developed. Use cases and extensive evaluation results confirm that it can deliver sufficient performance, and also is easy to use for incremental applications in large-scale distributed computing.

97 MATHEMATICS AND COMPUTING↗

Adding Fault Tolerance to NPB Benchmarks Using ULFM

In the world of high-performance computing, fault tolerance and application resilience are becoming some of the primary concerns because of increasing hardware failures and memory corruptions. While the research community has been investigating various options, from system-level solutions to application-level solutions, standards such as the Message Passing Interface (MPI) are also starting to include such capabilities. The current proposal for MPI fault tolerant is centered around the User-Level Failure Mitigation (ULFM) concept, which provides means for fault detection and recovery of the MPI layer. This approach does not address application-level recovery, which is currently left to application developers. In this work, we present a modification of some of the benchmarks of the NAS parallel benchmark (NPB) to include support of the ULFM capabilities as well as application-level strategies and mechanisms for application-level failure recovery. As such, we present: (i) an application-level library to "checkpoint" and restore data, (ii) extensions of NPB benchmarks for fault tolerance based on different strategies, (iii) a fault injection tool, and (iv) some preliminary results that show the impact of such fault tolerant strategies on the application execution.

97 MATHEMATICS AND COMPUTING↗

The augmented saddle field discharge characteristics and its applications for plasma enhanced chemical vapour deposition

A high ion flux parallel electrode plasma is proposed and studied in its DC configuration. By cascading a diode source region which supplies electrons and a saddle field region where these seed electrons are energized and amplified, the energy of ion bombardment on the substrate can be decoupled from the plasma density. The sufficiently large density of electrons and holes in the vicinity of the substrate raises the possibility to perform plasma enhanced chemical vapour deposition on insulating materials, at low sheath voltages (around 40 V in the configuration studied), at low temperatures in which the surface mobility of film growth species may be provided by the bombardment of moderate energy ions. As a benchmarking exercise, experiments are carried out on silane discharge characteristics and deposition of hydrogenated amorphous silicon (a-Si:H) on both silicon wafer and glass. The films grown at low anode voltages have excellent microstructures with predominantly monohydride bonds, sharp band tails, but relatively high integrated defect density in the mid 10{sup 16}/cm{sup 3} range for the particular substrate temperature of 180 Degree-Sign C, indicating that further optimizations are necessary if the electrode configuration is to be used to create a-Si:H devices.

36 MATERIALS SCIENCE↗