Visualizing Parallel Computer System Performance
NASA Technical Reports Server (NTRS)
Malony, Allen D.; Reed, Daniel A.
1988-01-01
Parallel computer systems are among the most complex of man's creations, making satisfactory performance characterization difficult. Despite this complexity, there are strong, indeed, almost irresistible, incentives to quantify parallel system performance using a single metric. The fallacy lies in succumbing to such temptations. A complete performance characterization requires not only an analysis of the system's constituent levels, it also requires both static and dynamic characterizations. Static or average behavior analysis may mask transients that dramatically alter system performance. Although the human visual system is remarkedly adept at interpreting and identifying anomalies in false color data, the importance of dynamic, visual scientific data presentation has only recently been recognized Large, complex parallel system pose equally vexing performance interpretation problems. Data from hardware and software performance monitors must be presented in ways that emphasize important events while eluding irrelevant details. Design approaches and tools for performance visualization are the subject of this paper.
High Performance Parallel Computational Nanotechnology
NASA Technical Reports Server (NTRS)
Saini, Subhash; Craw, James M. (Technical Monitor)
1995-01-01
At a recent press conference, NASA Administrator Dan Goldin encouraged NASA Ames Research Center to take a lead role in promoting research and development of advanced, high-performance computer technology, including nanotechnology. Manufacturers of leading-edge microprocessors currently perform large-scale simulations in the design and verification of semiconductor devices and microprocessors. Recently, the need for this intensive simulation and modeling analysis has greatly increased, due in part to the ever-increasing complexity of these devices, as well as the lessons of experiences such as the Pentium fiasco. Simulation, modeling, testing, and validation will be even more important for designing molecular computers because of the complex specification of millions of atoms, thousands of assembly steps, as well as the simulation and modeling needed to ensure reliable, robust and efficient fabrication of the molecular devices. The software for this capacity does not exist today, but it can be extrapolated from the software currently used in molecular modeling for other applications: semi-empirical methods, ab initio methods, self-consistent field methods, Hartree-Fock methods, molecular mechanics; and simulation methods for diamondoid structures. In as much as it seems clear that the application of such methods in nanotechnology will require powerful, highly powerful systems, this talk will discuss techniques and issues for performing these types of computations on parallel systems. We will describe system design issues (memory, I/O, mass storage, operating system requirements, special user interface issues, interconnects, bandwidths, and programming languages) involved in parallel methods for scalable classical, semiclassical, quantum, molecular mechanics, and continuum models; molecular nanotechnology computer-aided designs (NanoCAD) techniques; visualization using virtual reality techniques of structural models and assembly sequences; software required to
Interfacing Computer Aided Parallelization and Performance Analysis
NASA Technical Reports Server (NTRS)
Jost, Gabriele; Jin, Haoqiang; Labarta, Jesus; Gimenez, Judit; Biegel, Bryan A. (Technical Monitor)
2003-01-01
When porting sequential applications to parallel computer architectures, the program developer will typically go through several cycles of source code optimization and performance analysis. We have started a project to develop an environment where the user can jointly navigate through program structure and performance data information in order to make efficient optimization decisions. In a prototype implementation we have interfaced the CAPO computer aided parallelization tool with the Paraver performance analysis tool. We describe both tools and their interface and give an example for how the interface helps within the program development cycle of a benchmark code.
Misleading Performance Claims in Parallel Computations
Bailey, David H.
2009-05-29
In a previous humorous note entitled 'Twelve Ways to Fool the Masses,' I outlined twelve common ways in which performance figures for technical computer systems can be distorted. In this paper and accompanying conference talk, I give a reprise of these twelve 'methods' and give some actual examples that have appeared in peer-reviewed literature in years past. I then propose guidelines for reporting performance, the adoption of which would raise the level of professionalism and reduce the level of confusion, not only in the world of device simulation but also in the larger arena of technical computing.
Performance Evaluation in Network-Based Parallel Computing
NASA Technical Reports Server (NTRS)
Dezhgosha, Kamyar
1996-01-01
Network-based parallel computing is emerging as a cost-effective alternative for solving many problems which require use of supercomputers or massively parallel computers. The primary objective of this project has been to conduct experimental research on performance evaluation for clustered parallel computing. First, a testbed was established by augmenting our existing SUNSPARCs' network with PVM (Parallel Virtual Machine) which is a software system for linking clusters of machines. Second, a set of three basic applications were selected. The applications consist of a parallel search, a parallel sort, a parallel matrix multiplication. These application programs were implemented in C programming language under PVM. Third, we conducted performance evaluation under various configurations and problem sizes. Alternative parallel computing models and workload allocations for application programs were explored. The performance metric was limited to elapsed time or response time which in the context of parallel computing can be expressed in terms of speedup. The results reveal that the overhead of communication latency between processes in many cases is the restricting factor to performance. That is, coarse-grain parallelism which requires less frequent communication between processes will result in higher performance in network-based computing. Finally, we are in the final stages of installing an Asynchronous Transfer Mode (ATM) switch and four ATM interfaces (each 155 Mbps) which will allow us to extend our study to newer applications, performance metrics, and configurations.
Performance issues for engineering analysis on MIMD parallel computers
Fang, H.E.; Vaughan, C.T.; Gardner, D.R.
1994-08-01
We discuss how engineering analysts can obtain greater computational resolution in a more timely manner from applications codes running on MIMD parallel computers. Both processor speed and memory capacity are important to achieving better performance than a serial vector supercomputer. To obtain good performance, a parallel applications code must be scalable. In addition, the aspect ratios of the subdomains in the decomposition of the simulation domain onto the parallel computer should be of order 1. We demonstrate these conclusions using simulations conducted with the PCTH shock wave physics code running on a Cray Y-MP, a 1024-node nCUBE 2, and an 1840-node Paragon.
An introduction to performance debugging for parallel computers
Gropp, W.
1995-02-07
Programming parallel computers for performance is a difficult task that requires careful attention to both single-node performance and data exchange between processors. This paper discusses some of the sources of poor performance, ways to identify them in an application, and a few ways to address these issues.
Treveaven, P.
1989-01-01
This book presents an introduction to object-oriented, functional, and logic parallel computing on which the fifth generation of computer systems will be based. Coverage includes concepts for parallel computing languages, a parallel object-oriented system (DOOM) and its language (POOL), an object-oriented multilevel VLSI simulator using POOL, and implementation of lazy functional languages on parallel architectures.
Measuring performance of parallel computers. Progress report, 1989
Sullivan, F.
1994-07-01
Performance Measurement - the authors have developed a taxonomy of parallel algorithms based on data motion and example applications have been coded for each class of the taxonomy. Computational benchmark kernels have been extracted for several applications, and detailed measurements have been performed. Algorithms for Massively Parallel SIMD machines - measurement results and computational experiences indicate that top performance will be achieved by `iteration` type algorithms running on massively parallel SIMD machines. Reformulation as iteration may entail unorthodox approaches based on probabilistic methods. The authors have developed such methods for some applications. Here they discuss their approach to performance measurement, describe the taxonomy and measurements which have been made, and report on some general conclusions which can be drawn from the results of the measurements.
Routing performance analysis and optimization within a massively parallel computer
Archer, Charles Jens; Peters, Amanda; Pinnow, Kurt Walter; Swartz, Brent Allen
2013-04-16
An apparatus, program product and method optimize the operation of a massively parallel computer system by, in part, receiving actual performance data concerning an application executed by the plurality of interconnected nodes, and analyzing the actual performance data to identify an actual performance pattern. A desired performance pattern may be determined for the application, and an algorithm may be selected from among a plurality of algorithms stored within a memory, the algorithm being configured to achieve the desired performance pattern based on the actual performance data.
Energy Proportionality and Performance in Data Parallel Computing Clusters
Kim, Jinoh; Chou, Jerry; Rotem, Doron
2011-02-14
Energy consumption in datacenters has recently become a major concern due to the rising operational costs andscalability issues. Recent solutions to this problem propose the principle of energy proportionality, i.e., the amount of energy consumedby the server nodes must be proportional to the amount of work performed. For data parallelism and fault tolerancepurposes, most common file systems used in MapReduce-type clusters maintain a set of replicas for each data block. A coveringset is a group of nodes that together contain at least one replica of the data blocks needed for performing computing tasks. In thiswork, we develop and analyze algorithms to maintain energy proportionality by discovering a covering set that minimizesenergy consumption while placing the remaining nodes in lowpower standby mode. Our algorithms can also discover coveringsets in heterogeneous computing environments. In order to allow more data parallelism, we generalize our algorithms so that itcan discover k-covering sets, i.e., a set of nodes that contain at least k replicas of the data blocks. Our experimental results showthat we can achieve substantial energy saving without significant performance loss in diverse cluster configurations and workingenvironments.
Performance modeling of parallel computations in resource-constrained systems
Ghodsi, M.
1989-01-01
In this research, the author studies several problems related to the modeling and performance analysis of parallel computations. First, the author investigates the properties of a new task graph model, called the generalized task graphs, which can represent the nondeterminism involved in the parallel search algorithms. In parallel search, several possible solutions to a problem are carried out concurrently, with the intention of having only one successful result. The general task graphs are prone to problems such as deadlock, unboundedness, and unsafeness. The author defines the notion of well-formedness of task graphs by viewing them as high level Petri nets and provide necessary and sufficient conditions under which a generalized task graph is well formed. Next, the author proposes a new approximate iterative algorithm to predict the performance of a resource constrained queueing network, running a number of statistically identical jobs with internal concurrency. The jobs are assumed to be instances of an arbitrary task graph. The queueing network includes a limited number of identical passive resources. A task must acquire one unit of the passive resources before receiving service. Detailed experimental results are presented which show that the algorithm converges quite fast and is reasonably accurate.
Application Specific Performance Technology for Productive Parallel Computing
Malony, Allen D.; Shende, Sameer
2008-09-30
Our accomplishments over the last three years of the DOE project Application- Specific Performance Technology for Productive Parallel Computing (DOE Agreement: DE-FG02-05ER25680) are described below. The project will have met all of its objectives by the time of its completion at the end of September, 2008. Two extensive yearly progress reports were produced in in March 2006 and 2007 and were previously submitted to the DOE Office of Advanced Scientific Computing Research (OASCR). Following an overview of the objectives of the project, we summarize for each of the project areas the achievements in the first two years, and then describe in some more detail the project accomplishments this past year. At the end, we discuss the relationship of the proposed renewal application to the work done on the current project.
Performing an allreduce operation on a plurality of compute nodes of a parallel computer
Faraj, Ahmad
2012-04-17
Methods, apparatus, and products are disclosed for performing an allreduce operation on a plurality of compute nodes of a parallel computer. Each compute node includes at least two processing cores. Each processing core has contribution data for the allreduce operation. Performing an allreduce operation on a plurality of compute nodes of a parallel computer includes: establishing one or more logical rings among the compute nodes, each logical ring including at least one processing core from each compute node; performing, for each logical ring, a global allreduce operation using the contribution data for the processing cores included in that logical ring, yielding a global allreduce result for each processing core included in that logical ring; and performing, for each compute node, a local allreduce operation using the global allreduce results for each processing core on that compute node.
Performing a global barrier operation in a parallel computer
Archer, Charles J; Blocksome, Michael A; Ratterman, Joseph D; Smith, Brian E
2014-12-09
Executing computing tasks on a parallel computer that includes compute nodes coupled for data communications, where each compute node executes tasks, with one task on each compute node designated as a master task, including: for each task on each compute node until all master tasks have joined a global barrier: determining whether the task is a master task; if the task is not a master task, joining a single local barrier; if the task is a master task, joining the global barrier and the single local barrier only after all other tasks on the compute node have joined the single local barrier.
A Generic Scheduling Simulator for High Performance Parallel Computers
Yoo, B S; Choi, G S; Jette, M A
2001-08-01
It is well known that efficient job scheduling plays a crucial role in achieving high system utilization in large-scale high performance computing environments. A good scheduling algorithm should schedule jobs to achieve high system utilization while satisfying various user demands in an equitable fashion. Designing such a scheduling algorithm is a non-trivial task even in a static environment. In practice, the computing environment and workload are constantly changing. There are several reasons for this. First, the computing platforms constantly evolve as the technology advances. For example, the availability of relatively powerful commodity off-the-shelf (COTS) components at steadily diminishing prices have made it feasible to construct ever larger massively parallel computers in recent years [1, 4]. Second, the workload imposed on the system also changes constantly. The rapidly increasing compute resources have provided many applications developers with the opportunity to radically alter program characteristics and take advantage of these additional resources. New developments in software technology may also trigger changes in user applications. Finally, political climate change may alter user priorities or the mission of the organization. System designers in such dynamic environments must be able to accurately forecast the effect of changes in the hardware, software, and/or policies under consideration. If the environmental changes are significant, one must also reassess scheduling algorithms. Simulation has frequently been relied upon for this analysis, because other methods such as analytical modeling or actual measurements are usually too difficult or costly. A drawback of the simulation approach, however, is that developing a simulator is a time-consuming process. Furthermore, an existing simulator cannot be easily adapted to a new environment. In this research, we attempt to develop a generic job-scheduling simulator, which facilitates the evaluation of
Memory Benchmarks for SMP-Based High Performance Parallel Computers
Yoo, A B; de Supinski, B; Mueller, F; Mckee, S A
2001-11-20
As the speed gap between CPU and main memory continues to grow, memory accesses increasingly dominates the performance of many applications. The problem is particularly acute for symmetric multiprocessor (SMP) systems, where the shared memory may be accessed concurrently by a group of threads running on separate CPUs. Unfortunately, several key issues governing memory system performance in current systems are not well understood. Complex interactions between the levels of the memory hierarchy, buses or switches, DRAM back-ends, system software, and application access patterns can make it difficult to pinpoint bottlenecks and determine appropriate optimizations, and the situation is even more complex for SMP systems. To partially address this problem, we formulated a set of multi-threaded microbenchmarks for characterizing and measuring the performance of the underlying memory system in SMP-based high-performance computers. We report our use of these microbenchmarks on two important SMP-based machines. This paper has four primary contributions. First, we introduce a microbenchmark suite to systematically assess and compare the performance of different levels in SMP memory hierarchies. Second, we present a new tool based on hardware performance monitors to determine a wide array of memory system characteristics, such as cache sizes, quickly and easily; by using this tool, memory performance studies can be targeted to the full spectrum of performance regimes with many fewer data points than is otherwise required. Third, we present experimental results indicating that the performance of applications with large memory footprints remains largely constrained by memory. Fourth, we demonstrate that thread-level parallelism further degrades memory performance, even for the latest SMPs with hardware prefetching and switch-based memory interconnects.
NASA Astrophysics Data System (ADS)
Nishiura, Daisuke; Furuichi, Mikito; Sakaguchi, Hide
2015-09-01
The computational performance of a smoothed particle hydrodynamics (SPH) simulation is investigated for three types of current shared-memory parallel computer devices: many integrated core (MIC) processors, graphics processing units (GPUs), and multi-core CPUs. We are especially interested in efficient shared-memory allocation methods for each chipset, because the efficient data access patterns differ between compute unified device architecture (CUDA) programming for GPUs and OpenMP programming for MIC processors and multi-core CPUs. We first introduce several parallel implementation techniques for the SPH code, and then examine these on our target computer architectures to determine the most effective algorithms for each processor unit. In addition, we evaluate the effective computing performance and power efficiency of the SPH simulation on each architecture, as these are critical metrics for overall performance in a multi-device environment. In our benchmark test, the GPU is found to produce the best arithmetic performance as a standalone device unit, and gives the most efficient power consumption. The multi-core CPU obtains the most effective computing performance. The computational speed of the MIC processor on Xeon Phi approached that of two Xeon CPUs. This indicates that using MICs is an attractive choice for existing SPH codes on multi-core CPUs parallelized by OpenMP, as it gains computational acceleration without the need for significant changes to the source code.
Not Available
1991-10-23
An account of the Caltech Concurrent Computation Program (C{sup 3}P), a five year project that focused on answering the question: Can parallel computers be used to do large-scale scientific computations '' As the title indicates, the question is answered in the affirmative, by implementing numerous scientific applications on real parallel computers and doing computations that produced new scientific results. In the process of doing so, C{sup 3}P helped design and build several new computers, designed and implemented basic system software, developed algorithms for frequently used mathematical computations on massively parallel machines, devised performance models and measured the performance of many computers, and created a high performance computing facility based exclusively on parallel computers. While the initial focus of C{sup 3}P was the hypercube architecture developed by C. Seitz, many of the methods developed and lessons learned have been applied successfully on other massively parallel architectures.
Performing a local reduction operation on a parallel computer
Blocksome, Michael A.; Faraj, Daniel A.
2012-12-11
A parallel computer including compute nodes, each including two reduction processing cores, a network write processing core, and a network read processing core, each processing core assigned an input buffer. Copying, in interleaved chunks by the reduction processing cores, contents of the reduction processing cores' input buffers to an interleaved buffer in shared memory; copying, by one of the reduction processing cores, contents of the network write processing core's input buffer to shared memory; copying, by another of the reduction processing cores, contents of the network read processing core's input buffer to shared memory; and locally reducing in parallel by the reduction processing cores: the contents of the reduction processing core's input buffer; every other interleaved chunk of the interleaved buffer; the copied contents of the network write processing core's input buffer; and the copied contents of the network read processing core's input buffer.
Performing a local reduction operation on a parallel computer
Blocksome, Michael A; Faraj, Daniel A
2013-06-04
A parallel computer including compute nodes, each including two reduction processing cores, a network write processing core, and a network read processing core, each processing core assigned an input buffer. Copying, in interleaved chunks by the reduction processing cores, contents of the reduction processing cores' input buffers to an interleaved buffer in shared memory; copying, by one of the reduction processing cores, contents of the network write processing core's input buffer to shared memory; copying, by another of the reduction processing cores, contents of the network read processing core's input buffer to shared memory; and locally reducing in parallel by the reduction processing cores: the contents of the reduction processing core's input buffer; every other interleaved chunk of the interleaved buffer; the copied contents of the network write processing core's input buffer; and the copied contents of the network read processing core's input buffer.
High Performance Input/Output for Parallel Computer Systems
NASA Technical Reports Server (NTRS)
Ligon, W. B.
1996-01-01
The goal of our project is to study the I/O characteristics of parallel applications used in Earth Science data processing systems such as Regional Data Centers (RDCs) or EOSDIS. Our approach is to study the runtime behavior of typical programs and the effect of key parameters of the I/O subsystem both under simulation and with direct experimentation on parallel systems. Our three year activity has focused on two items: developing a test bed that facilitates experimentation with parallel I/O, and studying representative programs from the Earth science data processing application domain. The Parallel Virtual File System (PVFS) has been developed for use on a number of platforms including the Tiger Parallel Architecture Workbench (TPAW) simulator, The Intel Paragon, a cluster of DEC Alpha workstations, and the Beowulf system (at CESDIS). PVFS provides considerable flexibility in configuring I/O in a UNIX- like environment. Access to key performance parameters facilitates experimentation. We have studied several key applications fiom levels 1,2 and 3 of the typical RDC processing scenario including instrument calibration and navigation, image classification, and numerical modeling codes. We have also considered large-scale scientific database codes used to organize image data.
High-Performance Parallel Computing. Final report, 1 February 1984-31 January 1985
Browne, J.C.; Lipovski, G.J.
1986-01-22
The 1984/85 accomplishments of the research project High-Performance Parallel Computing included bringing the prototype of the Texas Reconfigurable Array Computer (TRAC) to a configuration and to a state of stability where it could support execution of simple assembly language programs; initial development of a unified model of parallel computation which is a basis for a programming environment uniting process and data flow models of parallel computation; bringing to operational status on an alternative host one of the two parallel programming languages (the Computation Structures Language, CSL) originally intended for use on TRAC; exploration of the expressive capabilities of this programming language; initiation of development of a graphical programming language based on the unified model of parallel computation mentioned previously; major progress on a graphically interfaced Petri Net-based performance modeling system for parallel computations and development of algorithms for scheduling of circuits to realize configurations in configurable Banyan network-based computer architectures.
NASA Technical Reports Server (NTRS)
Morgan, Philip E.
2004-01-01
This final report contains reports of research related to the tasks "Scalable High Performance Computing: Direct and Lark-Eddy Turbulent FLow Simulations Using Massively Parallel Computers" and "Devleop High-Performance Time-Domain Computational Electromagnetics Capability for RCS Prediction, Wave Propagation in Dispersive Media, and Dual-Use Applications. The discussion of Scalable High Performance Computing reports on three objectives: validate, access scalability, and apply two parallel flow solvers for three-dimensional Navier-Stokes flows; develop and validate a high-order parallel solver for Direct Numerical Simulations (DNS) and Large Eddy Simulation (LES) problems; and Investigate and develop a high-order Reynolds averaged Navier-Stokes turbulence model. The discussion of High-Performance Time-Domain Computational Electromagnetics reports on five objectives: enhancement of an electromagnetics code (CHARGE) to be able to effectively model antenna problems; utilize lessons learned in high-order/spectral solution of swirling 3D jets to apply to solving electromagnetics project; transition a high-order fluids code, FDL3DI, to be able to solve Maxwell's Equations using compact-differencing; develop and demonstrate improved radiation absorbing boundary conditions for high-order CEM; and extend high-order CEM solver to address variable material properties. The report also contains a review of work done by the systems engineer.
A high performance parallel computing architecture for robust image features
NASA Astrophysics Data System (ADS)
Zhou, Renyan; Liu, Leibo; Wei, Shaojun
2014-03-01
A design of parallel architecture for image feature detection and description is proposed in this article. The major component of this architecture is a 2D cellular network composed of simple reprogrammable processors, enabling the Hessian Blob Detector and Haar Response Calculation, which are the most computing-intensive stage of the Speeded Up Robust Features (SURF) algorithm. Combining this 2D cellular network and dedicated hardware for SURF descriptors, this architecture achieves real-time image feature detection with minimal software in the host processor. A prototype FPGA implementation of the proposed architecture achieves 1318.9 GOPS general pixel processing @ 100 MHz clock and achieves up to 118 fps in VGA (640 × 480) image feature detection. The proposed architecture is stand-alone and scalable so it is easy to be migrated into VLSI implementation.
Performing an allreduce operation on a plurality of compute nodes of a parallel computer
Faraj, Ahmad
2013-02-12
Methods, apparatus, and products are disclosed for performing an allreduce operation on a plurality of compute nodes of a parallel computer, each node including at least two processing cores, that include: performing, for each node, a local reduction operation using allreduce contribution data for the cores of that node, yielding, for each node, a local reduction result for one or more representative cores for that node; establishing one or more logical rings among the nodes, each logical ring including only one of the representative cores from each node; performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core included in that logical ring; and performing, for each node, a local broadcast operation using the global allreduce results for each representative core on that node.
Performance analysis of large scale parallel CFD computing based on Code_Saturne
NASA Astrophysics Data System (ADS)
Shang, Zhi
2013-02-01
In order to run computational fluid dynamics (CFD) codes on large scales, parallel computing has to be employed. For instance, on Petascale computing, general parallel computing without any optimization is not enough, especially for complex industrial issues that employ a large number of mesh cells to capture the details of the geometry. How to distribute these mesh cells among the multi-processors for Terascale and Petascale systems to obtain a good performance on parallel computing is really a challenge. Some mesh partitioning software packages, such as Metis, ParMetis, PT-Scotch and Zoltan, were chosen as the candidates ported into Code_Saturne to test if they can lead Code_Saturne towards Petascale and Exascale parallel CFD computing. Through the studies, it was found that mesh partitioning optimization software packages based on the graph mesh partitioning method can help the CFD code obtain good mesh distributions for high performance computing (HPC).
Parallel beam dynamics calculations on high performance computers
NASA Astrophysics Data System (ADS)
Ryne, Robert; Habib, Salman
1997-02-01
Faced with a backlog of nuclear waste and weapons plutonium, as well as an ever-increasing public concern about safety and environmental issues associated with conventional nuclear reactors, many countries are studying new, accelerator-driven technologies that hold the promise of providing safe and effective solutions to these problems. Proposed projects include accelerator transmutation of waste (ATW), accelerator-based conversion of plutonium (ABC), accelerator-driven energy production (ADEP), and accelerator production of tritium (APT). Also, next-generation spallation neutron sources based on similar technology will play a major role in materials science and biological science research. The design of accelerators for these projects will require a major advance in numerical modeling capability. For example, beam dynamics simulations with approximately 100 million particles will be needed to ensure that extremely stringent beam loss requirements (less than a nanoampere per meter) can be met. Compared with typical present-day modeling using 10,000-100,000 particles, this represents an increase of 3-4 orders of magnitude. High performance computing (HPC) platforms make it possible to perform such large scale simulations, which require 10's of GBytes of memory. They also make it possible to perform smaller simulations in a matter of hours that would require months to run on a single processor workstation. This paper will describe how HPC platforms can be used to perform the numerically intensive beam dynamics simulations required for development of these new accelerator-driven technologies.
Performing an allreduce operation on a plurality of compute nodes of a parallel computer
Faraj, Ahmad
2013-07-09
Methods, apparatus, and products are disclosed for performing an allreduce operation on a plurality of compute nodes of a parallel computer, each node including at least two processing cores, that include: establishing, for each node, a plurality of logical rings, each ring including a different set of at least one core on that node, each ring including the cores on at least two of the nodes; iteratively for each node: assigning each core of that node to one of the rings established for that node to which the core has not previously been assigned, and performing, for each ring for that node, a global allreduce operation using contribution data for the cores assigned to that ring or any global allreduce results from previous global allreduce operations, yielding current global allreduce results for each core; and performing, for each node, a local allreduce operation using the global allreduce results.
Geometrically nonlinear design sensitivity analysis on parallel-vector high-performance computers
NASA Technical Reports Server (NTRS)
Baddourah, Majdi A.; Nguyen, Duc T.
1993-01-01
Parallel-vector solution strategies for generation and assembly of element matrices, solution of the resulted system of linear equations, calculations of the unbalanced loads, displacements, stresses, and design sensitivity analysis (DSA) are all incorporated into the Newton Raphson (NR) procedure for nonlinear finite element analysis and DSA. Numerical results are included to show the performance of the proposed method for structural analysis and DSA in a parallel-vector computer environment.
NASA Technical Reports Server (NTRS)
Logan, Terry G.
1994-01-01
The purpose of this study is to investigate the performance of the integral equation computations using numerical source field-panel method in a massively parallel processing (MPP) environment. A comparative study of computational performance of the MPP CM-5 computer and conventional Cray-YMP supercomputer for a three-dimensional flow problem is made. A serial FORTRAN code is converted into a parallel CM-FORTRAN code. Some performance results are obtained on CM-5 with 32, 62, 128 nodes along with those on Cray-YMP with a single processor. The comparison of the performance indicates that the parallel CM-FORTRAN code near or out-performs the equivalent serial FORTRAN code for some cases.
High performance parallel architectures
Anderson, R.E. )
1989-09-01
In this paper the author describes current high performance parallel computer architectures. A taxonomy is presented to show computer architecture from the user programmer's point-of-view. The effects of the taxonomy upon the programming model are described. Some current architectures are described with respect to the taxonomy. Finally, some predictions about future systems are presented. 5 refs., 1 fig.
Performance Evaluation of Remote Memory Access (RMA) Programming on Shared Memory Parallel Computers
NASA Technical Reports Server (NTRS)
Jin, Hao-Qiang; Jost, Gabriele; Biegel, Bryan A. (Technical Monitor)
2002-01-01
The purpose of this study is to evaluate the feasibility of remote memory access (RMA) programming on shared memory parallel computers. We discuss different RMA based implementations of selected CFD application benchmark kernels and compare them to corresponding message passing based codes. For the message-passing implementation we use MPI point-to-point and global communication routines. For the RMA based approach we consider two different libraries supporting this programming model. One is a shared memory parallelization library (SMPlib) developed at NASA Ames, the other is the MPI-2 extensions to the MPI Standard. We give timing comparisons for the different implementation strategies and discuss the performance.
Parallel-vector unsymmetric Eigen-Solver on high performance computers
NASA Technical Reports Server (NTRS)
Nguyen, Duc T.; Jiangning, Qin
1993-01-01
The popular QR algorithm for solving all eigenvalues of an unsymmetric matrix is reviewed. Among the basic components in the QR algorithm, it was concluded from this study, that the reduction of an unsymmetric matrix to a Hessenberg form (before applying the QR algorithm itself) can be done effectively by exploiting the vector speed and multiple processors offered by modern high-performance computers. Numerical examples of several test cases have indicated that the proposed parallel-vector algorithm for converting a given unsymmetric matrix to a Hessenberg form offers computational advantages over the existing algorithm. The time saving obtained by the proposed methods is increased as the problem size increased.
Massively parallel processor computer
NASA Technical Reports Server (NTRS)
Fung, L. W. (Inventor)
1983-01-01
An apparatus for processing multidimensional data with strong spatial characteristics, such as raw image data, characterized by a large number of parallel data streams in an ordered array is described. It comprises a large number (e.g., 16,384 in a 128 x 128 array) of parallel processing elements operating simultaneously and independently on single bit slices of a corresponding array of incoming data streams under control of a single set of instructions. Each of the processing elements comprises a bidirectional data bus in communication with a register for storing single bit slices together with a random access memory unit and associated circuitry, including a binary counter/shift register device, for performing logical and arithmetical computations on the bit slices, and an I/O unit for interfacing the bidirectional data bus with the data stream source. The massively parallel processor architecture enables very high speed processing of large amounts of ordered parallel data, including spatial translation by shifting or sliding of bits vertically or horizontally to neighboring processing elements.
High performance parallel computing of flows in complex geometries: II. Applications
NASA Astrophysics Data System (ADS)
Gourdain, N.; Gicquel, L.; Staffelbach, G.; Vermorel, O.; Duchaine, F.; Boussuge, J.-F.; Poinsot, T.
2009-01-01
Present regulations in terms of pollutant emissions, noise and economical constraints, require new approaches and designs in the fields of energy supply and transportation. It is now well established that the next breakthrough will come from a better understanding of unsteady flow effects and by considering the entire system and not only isolated components. However, these aspects are still not well taken into account by the numerical approaches or understood whatever the design stage considered. The main challenge is essentially due to the computational requirements inferred by such complex systems if it is to be simulated by use of supercomputers. This paper shows how new challenges can be addressed by using parallel computing platforms for distinct elements of a more complex systems as encountered in aeronautical applications. Based on numerical simulations performed with modern aerodynamic and reactive flow solvers, this work underlines the interest of high-performance computing for solving flow in complex industrial configurations such as aircrafts, combustion chambers and turbomachines. Performance indicators related to parallel computing efficiency are presented, showing that establishing fair criterions is a difficult task for complex industrial applications. Examples of numerical simulations performed in industrial systems are also described with a particular interest for the computational time and the potential design improvements obtained with high-fidelity and multi-physics computing methods. These simulations use either unsteady Reynolds-averaged Navier-Stokes methods or large eddy simulation and deal with turbulent unsteady flows, such as coupled flow phenomena (thermo-acoustic instabilities, buffet, etc). Some examples of the difficulties with grid generation and data analysis are also presented when dealing with these complex industrial applications.
Performance of a parallel code for the Euler equations on hypercube computers
NASA Technical Reports Server (NTRS)
Barszcz, Eric; Chan, Tony F.; Jesperson, Dennis C.; Tuminaro, Raymond S.
1990-01-01
The performance of hypercubes were evaluated on a computational fluid dynamics problem and the parallel environment issues were considered that must be addressed, such as algorithm changes, implementation choices, programming effort, and programming environment. The evaluation focuses on a widely used fluid dynamics code, FLO52, which solves the two dimensional steady Euler equations describing flow around the airfoil. The code development experience is described, including interacting with the operating system, utilizing the message-passing communication system, and code modifications necessary to increase parallel efficiency. Results from two hypercube parallel computers (a 16-node iPSC/2, and a 512-node NCUBE/ten) are discussed and compared. In addition, a mathematical model of the execution time was developed as a function of several machine and algorithm parameters. This model accurately predicts the actual run times obtained and is used to explore the performance of the code in interesting but yet physically realizable regions of the parameter space. Based on this model, predictions about future hypercubes are made.
A parallel-vector algorithm for rapid structural analysis on high-performance computers
NASA Technical Reports Server (NTRS)
Storaasli, Olaf O.; Nguyen, Duc T.; Agarwal, Tarun K.
1990-01-01
A fast, accurate Choleski method for the solution of symmetric systems of linear equations is presented. This direct method is based on a variable-band storage scheme and takes advantage of column heights to reduce the number of operations in the Choleski factorization. The method employs parallel computation in the outermost DO-loop and vector computation via the loop unrolling technique in the innermost DO-loop. The method avoids computations with zeros outside the column heights, and as an option, zeros inside the band. The close relationship between Choleski and Gauss elimination methods is examined. The minor changes required to convert the Choleski code to a Gauss code to solve non-positive-definite symmetric systems of equations are identified. The results for two large scale structural analyses performed on supercomputers, demonstrate the accuracy and speed of the method.
Orlando, Roberto; Delle Piane, Massimo; Bush, Ian J; Ugliengo, Piero; Ferrabone, Matteo; Dovesi, Roberto
2012-10-30
Fully ab initio treatment of complex solid systems needs computational software which is able to efficiently take advantage of the growing power of high performance computing (HPC) architectures. Recent improvements in CRYSTAL, a periodic ab initio code that uses a Gaussian basis set, allows treatment of very large unit cells for crystalline systems on HPC architectures with high parallel efficiency in terms of running time and memory requirements. The latter is a crucial point, due to the trend toward architectures relying on a very high number of cores with associated relatively low memory availability. An exhaustive performance analysis shows that density functional calculations, based on a hybrid functional, of low-symmetry systems containing up to 100,000 atomic orbitals and 8000 atoms are feasible on the most advanced HPC architectures available to European researchers today, using thousands of processors.
Parallel computers and parallel algorithms for CFD: An introduction
NASA Astrophysics Data System (ADS)
Roose, Dirk; Vandriessche, Rafael
1995-10-01
This text presents a tutorial on those aspects of parallel computing that are important for the development of efficient parallel algorithms and software for computational fluid dynamics. We first review the main architectural features of parallel computers and we briefly describe some parallel systems on the market today. We introduce some important concepts concerning the development and the performance evaluation of parallel algorithms. We discuss how work load imbalance and communication costs on distributed memory parallel computers can be minimized. We present performance results for some CFD test cases. We focus on applications using structured and block structured grids, but the concepts and techniques are also valid for unstructured grids.
Parallel-META: efficient metagenomic data analysis based on high-performance computation
2012-01-01
Background Metagenomics method directly sequences and analyses genome information from microbial communities. There are usually more than hundreds of genomes from different microbial species in the same community, and the main computational tasks for metagenomic data analyses include taxonomical and functional component examination of all genomes in the microbial community. Metagenomic data analysis is both data- and computation- intensive, which requires extensive computational power. Most of the current metagenomic data analysis softwares were designed to be used on a single computer or single computer clusters, which could not match with the fast increasing number of large metagenomic projects' computational requirements. Therefore, advanced computational methods and pipelines have to be developed to cope with such need for efficient analyses. Result In this paper, we proposed Parallel-META, a GPU- and multi-core-CPU-based open-source pipeline for metagenomic data analysis, which enabled the efficient and parallel analysis of multiple metagenomic datasets and the visualization of the results for multiple samples. In Parallel-META, the similarity-based database search was parallelized based on GPU computing and multi-core CPU computing optimization. Experiments have shown that Parallel-META has at least 15 times speed-up compared to traditional metagenomic data analysis method, with the same accuracy of the results http://www.computationalbioenergy.org/parallel-meta.html. Conclusion The parallel processing of current metagenomic data would be very promising: with current speed up of 15 times and above, binning would not be a very time-consuming process any more. Therefore, some deeper analysis of the metagenomic data, such as the comparison of different samples, would be feasible in the pipeline, and some of these functionalities have been included into the Parallel-META pipeline. PMID:23046922
Structural mechanics computations on parallel computing platforms
Kulak, R.F.; Plaskacz, E.J.; Pfeiffer, P.A.
1995-06-01
With recent advances in parallel supercomputers and network-connected workstations, the solution to large scale structural engineering problems has now become tractable. High-performance computer architectures, which are usually available at large universities and national laboratories, now can solve large nonlinear problems. At the other end of the spectrum, network connected workstations can be configured to become a distributed-parallel computer. This approach is attractive to small, medium and large engineering firms. This paper describes the development of a parallelized finite element computer program for the solution of static, nonlinear structural mechanics problems.
NASA Technical Reports Server (NTRS)
Katz, D.; Cwik, T.; Sterling, T.
1998-01-01
This paper uses the parallel calculation of the radiation integral for examination of performance and compiler issues on a Beowulf-class computer. This type of computer, built from mass-market, commodity, off-the-shelf components, has limited communications performance and therefore also has a limited regime of codes for which it is suitable.
Koniges, A.
1996-02-09
This project is a package of 11 individual CRADA`s plus hardware. This innovative project established a three-year multi-party collaboration that is significantly accelerating the availability of commercial massively parallel processing computing software technology to U.S. government, academic, and industrial end-users. This report contains individual presentations from nine principal investigators along with overall program information.
Turbomachinery CFD on parallel computers
NASA Technical Reports Server (NTRS)
Blech, Richard A.; Milner, Edward J.; Quealy, Angela; Townsend, Scott E.
1992-01-01
The role of multistage turbomachinery simulation in the development of propulsion system models is discussed. Particularly, the need for simulations with higher fidelity and faster turnaround time is highlighted. It is shown how such fast simulations can be used in engineering-oriented environments. The use of parallel processing to achieve the required turnaround times is discussed. Current work by several researchers in this area is summarized. Parallel turbomachinery CFD research at the NASA Lewis Research Center is then highlighted. These efforts are focused on implementing the average-passage turbomachinery model on MIMD, distributed memory parallel computers. Performance results are given for inviscid, single blade row and viscous, multistage applications on several parallel computers, including networked workstations.
NASA Technical Reports Server (NTRS)
Denning, Peter J.; Tichy, Walter F.
1990-01-01
Highly parallel computing architectures are the only means to achieve the computation rates demanded by advanced scientific problems. A decade of research has demonstrated the feasibility of such machines and current research focuses on which architectures designated as multiple instruction multiple datastream (MIMD) and single instruction multiple datastream (SIMD) have produced the best results to date; neither shows a decisive advantage for most near-homogeneous scientific problems. For scientific problems with many dissimilar parts, more speculative architectures such as neural networks or data flow may be needed.
pWeb: A High-Performance, Parallel-Computing Framework for Web-Browser-Based Medical Simulation.
Halic, Tansel; Ahn, Woojin; De, Suvranu
2014-01-01
This work presents a pWeb - a new language and compiler for parallelization of client-side compute intensive web applications such as surgical simulations. The recently introduced HTML5 standard has enabled creating unprecedented applications on the web. Low performance of the web browser, however, remains the bottleneck of computationally intensive applications including visualization of complex scenes, real time physical simulations and image processing compared to native ones. The new proposed language is built upon web workers for multithreaded programming in HTML5. The language provides fundamental functionalities of parallel programming languages as well as the fork/join parallel model which is not supported by web workers. The language compiler automatically generates an equivalent parallel script that complies with the HTML5 standard. A case study on realistic rendering for surgical simulations demonstrates enhanced performance with a compact set of instructions.
pWeb: A High-Performance, Parallel-Computing Framework for Web-Browser-Based Medical Simulation.
Halic, Tansel; Ahn, Woojin; De, Suvranu
2014-01-01
This work presents a pWeb - a new language and compiler for parallelization of client-side compute intensive web applications such as surgical simulations. The recently introduced HTML5 standard has enabled creating unprecedented applications on the web. Low performance of the web browser, however, remains the bottleneck of computationally intensive applications including visualization of complex scenes, real time physical simulations and image processing compared to native ones. The new proposed language is built upon web workers for multithreaded programming in HTML5. The language provides fundamental functionalities of parallel programming languages as well as the fork/join parallel model which is not supported by web workers. The language compiler automatically generates an equivalent parallel script that complies with the HTML5 standard. A case study on realistic rendering for surgical simulations demonstrates enhanced performance with a compact set of instructions. PMID:24732497
DeHart, Mark D; Williams, Mark L; Bowman, Stephen M
2010-01-01
The SCALE computational architecture has remained basically the same since its inception 30 years ago, although constituent modules and capabilities have changed significantly. This SCALE concept was intended to provide a framework whereby independent codes can be linked to provide a more comprehensive capability than possible with the individual programs - allowing flexibility to address a wide variety of applications. However, the current system was designed originally for mainframe computers with a single CPU and with significantly less memory than today's personal computers. It has been recognized that the present SCALE computation system could be restructured to take advantage of modern hardware and software capabilities, while retaining many of the modular features of the present system. Preliminary work is being done to define specifications and capabilities for a more advanced computational architecture. This paper describes the state of current SCALE development activities and plans for future development. With the release of SCALE 6.1 in 2010, a new phase of evolutionary development will be available to SCALE users within the TRITON and NEWT modules. The SCALE (Standardized Computer Analyses for Licensing Evaluation) code system developed by Oak Ridge National Laboratory (ORNL) provides a comprehensive and integrated package of codes and nuclear data for a wide range of applications in criticality safety, reactor physics, shielding, isotopic depletion and decay, and sensitivity/uncertainty (S/U) analysis. Over the last three years, since the release of version 5.1 in 2006, several important new codes have been introduced within SCALE, and significant advances applied to existing codes. Many of these new features became available with the release of SCALE 6.0 in early 2009. However, beginning with SCALE 6.1, a first generation of parallel computing is being introduced. In addition to near-term improvements, a plan for longer term SCALE enhancement
Trajectory optimization using parallel shooting method on parallel computer
Wirthman, D.J.; Park, S.Y.; Vadali, S.R.
1995-03-01
The efficiency of a parallel shooting method on a parallel computer for solving a variety of optimal control guidance problems is studied. Several examples are considered to demonstrate that a speedup of nearly 7 to 1 is achieved with the use of 16 processors. It is suggested that further improvements in performance can be achieved by parallelizing in the state domain. 10 refs.
High-Performance Computation of Distributed-Memory Parallel 3D Voronoi and Delaunay Tessellation
Peterka, Tom; Morozov, Dmitriy; Phillips, Carolyn
2014-11-14
Computing a Voronoi or Delaunay tessellation from a set of points is a core part of the analysis of many simulated and measured datasets: N-body simulations, molecular dynamics codes, and LIDAR point clouds are just a few examples. Such computational geometry methods are common in data analysis and visualization; but as the scale of simulations and observations surpasses billions of particles, the existing serial and shared-memory algorithms no longer suffice. A distributed-memory scalable parallel algorithm is the only feasible approach. The primary contribution of this paper is a new parallel Delaunay and Voronoi tessellation algorithm that automatically determines which neighbor points need to be exchanged among the subdomains of a spatial decomposition. Other contributions include periodic and wall boundary conditions, comparison of our method using two popular serial libraries, and application to numerous science datasets.
Parallel Computing Using Web Servers and "Servlets".
ERIC Educational Resources Information Center
Lo, Alfred; Bloor, Chris; Choi, Y. K.
2000-01-01
Describes parallel computing and presents inexpensive ways to implement a virtual parallel computer with multiple Web servers. Highlights include performance measurement of parallel systems; models for using Java and intranet technology including single server, multiple clients and multiple servers, single client; and a comparison of CGI (common…
Merlin - Massively parallel heterogeneous computing
NASA Technical Reports Server (NTRS)
Wittie, Larry; Maples, Creve
1989-01-01
Hardware and software for Merlin, a new kind of massively parallel computing system, are described. Eight computers are linked as a 300-MIPS prototype to develop system software for a larger Merlin network with 16 to 64 nodes, totaling 600 to 3000 MIPS. These working prototypes help refine a mapped reflective memory technique that offers a new, very general way of linking many types of computer to form supercomputers. Processors share data selectively and rapidly on a word-by-word basis. Fast firmware virtual circuits are reconfigured to match topological needs of individual application programs. Merlin's low-latency memory-sharing interfaces solve many problems in the design of high-performance computing systems. The Merlin prototypes are intended to run parallel programs for scientific applications and to determine hardware and software needs for a future Teraflops Merlin network.
Parallel computation and computers for artificial intelligence
Kowalik, J.S. )
1988-01-01
This book discusses Parallel Processing in Artificial Intelligence; Parallel Computing using Multilisp; Execution of Common Lisp in a Parallel Environment; Qlisp; Restricted AND-Parallel Execution of Logic Programs; PARLOG: Parallel Programming in Logic; and Data-driven Processing of Semantic Nets. Attention is also given to: Application of the Butterfly Parallel Processor in Artificial Intelligence; On the Range of Applicability of an Artificial Intelligence Machine; Low-level Vision on Warp and the Apply Programming Mode; AHR: A Parallel Computer for Pure Lisp; FAIM-1: An Architecture for Symbolic Multi-processing; and Overview of Al Application Oriented Parallel Processing Research in Japan.
Achieving high performance in numerical computations on RISC workstations and parallel systems
Goedecker, S.; Hoisie, A.
1997-08-20
The nominal peak speeds of both serial and parallel computers is raising rapidly. At the same time however it is becoming increasingly difficult to get out a significant fraction of this high peak speed from modern computer architectures. In this tutorial the authors give the scientists and engineers involved in numerically demanding calculations and simulations the necessary basic knowledge to write reasonably efficient programs. The basic principles are rather simple and the possible rewards large. Writing a program by taking into account optimization techniques related to the computer architecture can significantly speedup your program, often by factors of 10--100. As such, optimizing a program can for instance be a much better solution than buying a faster computer. If a few basic optimization principles are applied during program development, the additional time needed for obtaining an efficient program is practically negligible. In-depth optimization is usually only needed for a few subroutines or kernels and the effort involved is therefore also acceptable.
Parallel distributed computing using Python
NASA Astrophysics Data System (ADS)
Dalcin, Lisandro D.; Paz, Rodrigo R.; Kler, Pablo A.; Cosimo, Alejandro
2011-09-01
This work presents two software components aimed to relieve the costs of accessing high-performance parallel computing resources within a Python programming environment: MPI for Python and PETSc for Python. MPI for Python is a general-purpose Python package that provides bindings for the Message Passing Interface (MPI) standard using any back-end MPI implementation. Its facilities allow parallel Python programs to easily exploit multiple processors using the message passing paradigm. PETSc for Python provides access to the Portable, Extensible Toolkit for Scientific Computation (PETSc) libraries. Its facilities allow sequential and parallel Python applications to exploit state of the art algorithms and data structures readily available in PETSc for the solution of large-scale problems in science and engineering. MPI for Python and PETSc for Python are fully integrated to PETSc-FEM, an MPI and PETSc based parallel, multiphysics, finite elements code developed at CIMEC laboratory. This software infrastructure supports research activities related to simulation of fluid flows with applications ranging from the design of microfluidic devices for biochemical analysis to modeling of large-scale stream/aquifer interactions.
Parallel processing for scientific computations
NASA Technical Reports Server (NTRS)
Alkhatib, Hasan S.
1995-01-01
The scope of this project dealt with the investigation of the requirements to support distributed computing of scientific computations over a cluster of cooperative workstations. Various experiments on computations for the solution of simultaneous linear equations were performed in the early phase of the project to gain experience in the general nature and requirements of scientific applications. A specification of a distributed integrated computing environment, DICE, based on a distributed shared memory communication paradigm has been developed and evaluated. The distributed shared memory model facilitates porting existing parallel algorithms that have been designed for shared memory multiprocessor systems to the new environment. The potential of this new environment is to provide supercomputing capability through the utilization of the aggregate power of workstations cooperating in a cluster interconnected via a local area network. Workstations, generally, do not have the computing power to tackle complex scientific applications, making them primarily useful for visualization, data reduction, and filtering as far as complex scientific applications are concerned. There is a tremendous amount of computing power that is left unused in a network of workstations. Very often a workstation is simply sitting idle on a desk. A set of tools can be developed to take advantage of this potential computing power to create a platform suitable for large scientific computations. The integration of several workstations into a logical cluster of distributed, cooperative, computing stations presents an alternative to shared memory multiprocessor systems. In this project we designed and evaluated such a system.
NASA Technical Reports Server (NTRS)
Lyster, Peter M.; Guo, J.; Clune, T.; Larson, J. W.; Atlas, Robert (Technical Monitor)
2001-01-01
The computational complexity of algorithms for Four Dimensional Data Assimilation (4DDA) at NASA's Data Assimilation Office (DAO) is discussed. In 4DDA, observations are assimilated with the output of a dynamical model to generate best-estimates of the states of the system. It is thus a mapping problem, whereby scattered observations are converted into regular accurate maps of wind, temperature, moisture and other variables. The DAO is developing and using 4DDA algorithms that provide these datasets, or analyses, in support of Earth System Science research. Two large-scale algorithms are discussed. The first approach, the Goddard Earth Observing System Data Assimilation System (GEOS DAS), uses an atmospheric general circulation model (GCM) and an observation-space based analysis system, the Physical-space Statistical Analysis System (PSAS). GEOS DAS is very similar to global meteorological weather forecasting data assimilation systems, but is used at NASA for climate research. Systems of this size typically run at between 1 and 20 gigaflop/s. The second approach, the Kalman filter, uses a more consistent algorithm to determine the forecast error covariance matrix than does GEOS DAS. For atmospheric assimilation, the gridded dynamical fields typically have More than 10(exp 6) variables, therefore the full error covariance matrix may be in excess of a teraword. For the Kalman filter this problem can easily scale to petaflop/s proportions. We discuss the computational complexity of GEOS DAS and our implementation of the Kalman filter. We also discuss and quantify some of the technical issues and limitations in developing efficient, in terms of wall clock time, and scalable parallel implementations of the algorithms.
Parallel computing in enterprise modeling.
Goldsby, Michael E.; Armstrong, Robert C.; Shneider, Max S.; Vanderveen, Keith; Ray, Jaideep; Heath, Zach; Allan, Benjamin A.
2008-08-01
This report presents the results of our efforts to apply high-performance computing to entity-based simulations with a multi-use plugin for parallel computing. We use the term 'Entity-based simulation' to describe a class of simulation which includes both discrete event simulation and agent based simulation. What simulations of this class share, and what differs from more traditional models, is that the result sought is emergent from a large number of contributing entities. Logistic, economic and social simulations are members of this class where things or people are organized or self-organize to produce a solution. Entity-based problems never have an a priori ergodic principle that will greatly simplify calculations. Because the results of entity-based simulations can only be realized at scale, scalable computing is de rigueur for large problems. Having said that, the absence of a spatial organizing principal makes the decomposition of the problem onto processors problematic. In addition, practitioners in this domain commonly use the Java programming language which presents its own problems in a high-performance setting. The plugin we have developed, called the Parallel Particle Data Model, overcomes both of these obstacles and is now being used by two Sandia frameworks: the Decision Analysis Center, and the Seldon social simulation facility. While the ability to engage U.S.-sized problems is now available to the Decision Analysis Center, this plugin is central to the success of Seldon. Because Seldon relies on computationally intensive cognitive sub-models, this work is necessary to achieve the scale necessary for realistic results. With the recent upheavals in the financial markets, and the inscrutability of terrorist activity, this simulation domain will likely need a capability with ever greater fidelity. High-performance computing will play an important part in enabling that greater fidelity.
Parallel processing for scientific computations
NASA Technical Reports Server (NTRS)
Alkhatib, Hasan S.
1991-01-01
The main contribution of the effort in the last two years is the introduction of the MOPPS system. After doing extensive literature search, we introduced the system which is described next. MOPPS employs a new solution to the problem of managing programs which solve scientific and engineering applications on a distributed processing environment. Autonomous computers cooperate efficiently in solving large scientific problems with this solution. MOPPS has the advantage of not assuming the presence of any particular network topology or configuration, computer architecture, or operating system. It imposes little overhead on network and processor resources while efficiently managing programs concurrently. The core of MOPPS is an intelligent program manager that builds a knowledge base of the execution performance of the parallel programs it is managing under various conditions. The manager applies this knowledge to improve the performance of future runs. The program manager learns from experience.
Parallel algorithms for matrix computations
Plemmons, R.J.
1990-01-01
The present conference on parallel algorithms for matrix computations encompasses both shared-memory systems and distributed-memory systems, as well as combinations of the two, to provide an overall perspective on parallel algorithms for both dense and sparse matrix computations in solving systems of linear equations, dense or structured problems related to least-squares computations, eigenvalue computations, singular-value computations, and rapid elliptic solvers. Specific issues addressed include the influence of parallel and vector architectures on algorithm design, computations for distributed-memory architectures such as hypercubes, solutions for sparse symmetric positive definite linear systems, symbolic and numeric factorizations, and triangular solutions. Also addressed are reference sources for parallel and vector numerical algorithms, sources for machine architectures, and sources for programming languages.
Template based parallel checkpointing in a massively parallel computer system
Archer, Charles Jens; Inglett, Todd Alan
2009-01-13
A method and apparatus for a template based parallel checkpoint save for a massively parallel super computer system using a parallel variation of the rsync protocol, and network broadcast. In preferred embodiments, the checkpoint data for each node is compared to a template checkpoint file that resides in the storage and that was previously produced. Embodiments herein greatly decrease the amount of data that must be transmitted and stored for faster checkpointing and increased efficiency of the computer system. Embodiments are directed to a parallel computer system with nodes arranged in a cluster with a high speed interconnect that can perform broadcast communication. The checkpoint contains a set of actual small data blocks with their corresponding checksums from all nodes in the system. The data blocks may be compressed using conventional non-lossy data compression algorithms to further reduce the overall checkpoint size.
Parallel computation using limited resources
Sugla, B.
1985-01-01
This thesis addresses itself to the task of designing and analyzing parallel algorithms when the resources of processors, communication, and time are limited. The two parts of this thesis deal with multiprocessor systems and VLSI - the two important parallel processing environments that are prevalent today. In the first part a time-processor-communication tradeoff analysis is conducted for two kinds of problems - N input, 1 output, and N input, N output computations. In the class of problems of the second kind, the problem of prefix computation, an important problem due to the number of naturally occurring computations it can model, is studied. Finally, a general methodology is given for design of parallel algorithms that can be used to optimize a given design to a wide set of architectural variations. The second part of the thesis considers the design of parallel algorithms for the VLSI model of computation when the resource of time is severely restricted.
Chang, Justin; Karra, Satish; Nakshatrala, Kalyana B.
2016-07-26
It is well-known that the standard Galerkin formulation, which is often the formulation of choice under the finite element method for solving self-adjoint diffusion equations, does not meet maximum principles and the non-negative constraint for anisotropic diffusion equations. Recently, optimization-based methodologies that satisfy maximum principles and the non-negative constraint for steady-state and transient diffusion-type equations have been proposed. To date, these methodologies have been tested only on small-scale academic problems. The purpose of this paper is to systematically study the performance of the non-negative methodology in the context of high performance computing (HPC). PETSc and TAO libraries are, respectively, usedmore » for the parallel environment and optimization solvers. For large-scale problems, it is important for computational scientists to understand the computational performance of current algorithms available in these scientific libraries. The numerical experiments are conducted on the state-of-the-art HPC systems, and a single-core performance model is used to better characterize the efficiency of the solvers. Furthermore, our studies indicate that the proposed non-negative computational framework for diffusion-type equations exhibits excellent strong scaling for real-world large-scale problems.« less
Experimental Parallel-Processing Computer
NASA Technical Reports Server (NTRS)
Mcgregor, J. W.; Salama, M. A.
1986-01-01
Master processor supervises slave processors, each with its own memory. Computer with parallel processing serves as inexpensive tool for experimentation with parallel mathematical algorithms. Speed enhancement obtained depends on both nature of problem and structure of algorithm used. In parallel-processing architecture, "bank select" and control signals determine which one, if any, of N slave processor memories accessible to master processor at any given moment. When so selected, slave memory operates as part of master computer memory. When not selected, slave memory operates independently of main memory. Slave processors communicate with each other via input/output bus.
Parallel computation with the force
NASA Technical Reports Server (NTRS)
Jordan, H. F.
1985-01-01
A methodology, called the force, supports the construction of programs to be executed in parallel by a force of processes. The number of processes in the force is unspecified, but potentially very large. The force idea is embodied in a set of macros which produce multiproceossor FORTRAN code and has been studied on two shared memory multiprocessors of fairly different character. The method has simplified the writing of highly parallel programs within a limited class of parallel algorithms and is being extended to cover a broader class. The individual parallel constructs which comprise the force methodology are discussed. Of central concern are their semantics, implementation on different architectures and performance implications.
NASA Astrophysics Data System (ADS)
Alameda, J. C.
2011-12-01
Development and optimization of computational science models, particularly on high performance computers, and with the advent of ubiquitous multicore processor systems, practically on every system, has been accomplished with basic software tools, typically, command-line based compilers, debuggers, performance tools that have not changed substantially from the days of serial and early vector computers. However, model complexity, including the complexity added by modern message passing libraries such as MPI, and the need for hybrid code models (such as openMP and MPI) to be able to take full advantage of high performance computers with an increasing core count per shared memory node, has made development and optimization of such codes an increasingly arduous task. Additional architectural developments, such as many-core processors, only complicate the situation further. In this paper, we describe how our NSF-funded project, "SI2-SSI: A Productive and Accessible Development Workbench for HPC Applications Using the Eclipse Parallel Tools Platform" (WHPC) seeks to improve the Eclipse Parallel Tools Platform, an environment designed to support scientific code development targeted at a diverse set of high performance computing systems. Our WHPC project to improve Eclipse PTP takes an application-centric view to improve PTP. We are using a set of scientific applications, each with a variety of challenges, and using PTP to drive further improvements to both the scientific application, as well as to understand shortcomings in Eclipse PTP from an application developer perspective, to drive our list of improvements we seek to make. We are also partnering with performance tool providers, to drive higher quality performance tool integration. We have partnered with the Cactus group at Louisiana State University to improve Eclipse's ability to work with computational frameworks and extremely complex build systems, as well as to develop educational materials to incorporate into
NASA Technical Reports Server (NTRS)
VanderWijngaart, Rob F.; Biswas, Rupak; Frumkin, Michael; Feng, Huiyu; Biegel, Bryan (Technical Monitor)
2001-01-01
The contents include: 1) A brief history of NPB; 2) What is (not) being measured by NPB; 3) Irregular dynamic applications (UA Benchmark); and 4) Wide area distributed computing (NAS Grid Benchmarks-NGB). This paper is presented in viewgraph form.
Parallel Computing Strategies for Irregular Algorithms
NASA Technical Reports Server (NTRS)
Biswas, Rupak; Oliker, Leonid; Shan, Hongzhang; Biegel, Bryan (Technical Monitor)
2002-01-01
Parallel computing promises several orders of magnitude increase in our ability to solve realistic computationally-intensive problems, but relies on their efficient mapping and execution on large-scale multiprocessor architectures. Unfortunately, many important applications are irregular and dynamic in nature, making their effective parallel implementation a daunting task. Moreover, with the proliferation of parallel architectures and programming paradigms, the typical scientist is faced with a plethora of questions that must be answered in order to obtain an acceptable parallel implementation of the solution algorithm. In this paper, we consider three representative irregular applications: unstructured remeshing, sparse matrix computations, and N-body problems, and parallelize them using various popular programming paradigms on a wide spectrum of computer platforms ranging from state-of-the-art supercomputers to PC clusters. We present the underlying problems, the solution algorithms, and the parallel implementation strategies. Smart load-balancing, partitioning, and ordering techniques are used to enhance parallel performance. Overall results demonstrate the complexity of efficiently parallelizing irregular algorithms.
Computational electromagnetics and parallel dense matrix computations
Forsman, K.; Kettunen, L.; Gropp, W.; Levine, D.
1995-06-01
We present computational results using CORAL, a parallel, three-dimensional, nonlinear magnetostatic code based on a volume integral equation formulation. A key feature of CORAL is the ability to solve, in parallel, the large, dense systems of linear equations that are inherent in the use of integral equation methods. Using the Chameleon and PSLES libraries ensures portability and access to the latest linear algebra solution technology.
NASA Astrophysics Data System (ADS)
Chugunov, Svyatoslav; Li, Changying
2015-09-01
Parallel implementation of two numerical tools popular in optical studies of biological materials-Inverse Adding-Doubling (IAD) program and Monte Carlo Multi-Layered (MCML) program-was developed and tested in this study. The implementation was based on Message Passing Interface (MPI) and standard C-language. Parallel versions of IAD and MCML programs were compared to their sequential counterparts in validation and performance tests. Additionally, the portability of the programs was tested using a local high performance computing (HPC) cluster, Penguin-On-Demand HPC cluster, and Amazon EC2 cluster. Parallel IAD was tested with up to 150 parallel cores using 1223 input datasets. It demonstrated linear scalability and the speedup was proportional to the number of parallel cores (up to 150x). Parallel MCML was tested with up to 1001 parallel cores using problem sizes of 104-109 photon packets. It demonstrated classical performance curves featuring communication overhead and performance saturation point. Optimal performance curve was derived for parallel MCML as a function of problem size. Typical speedup achieved for parallel MCML (up to 326x) demonstrated linear increase with problem size. Precision of MCML results was estimated in a series of tests - problem size of 106 photon packets was found optimal for calculations of total optical response and 108 photon packets for spatially-resolved results. The presented parallel versions of MCML and IAD programs are portable on multiple computing platforms. The parallel programs could significantly speed up the simulation for scientists and be utilized to their full potential in computing systems that are readily available without additional costs.
The science of computing - Parallel computation
NASA Technical Reports Server (NTRS)
Denning, P. J.
1985-01-01
Although parallel computation architectures have been known for computers since the 1920s, it was only in the 1970s that microelectronic components technologies advanced to the point where it became feasible to incorporate multiple processors in one machine. Concommitantly, the development of algorithms for parallel processing also lagged due to hardware limitations. The speed of computing with solid-state chips is limited by gate switching delays. The physical limit implies that a 1 Gflop operational speed is the maximum for sequential processors. A computer recently introduced features a 'hypercube' architecture with 128 processors connected in networks at 5, 6 or 7 points per grid, depending on the design choice. Its computing speed rivals that of supercomputers, but at a fraction of the cost. The added speed with less hardware is due to parallel processing, which utilizes algorithms representing different parts of an equation that can be broken into simpler statements and processed simultaneously. Present, highly developed computer languages like FORTRAN, PASCAL, COBOL, etc., rely on sequential instructions. Thus, increased emphasis will now be directed at parallel processing algorithms to exploit the new architectures.
Song, Baoxing; Xu, Jian; Ning, Kang
2014-01-01
The metagenomic method directly sequences and analyses genome information from microbial communities. The main computational tasks for metagenomic analyses include taxonomical and functional structure analysis for all genomes in a microbial community (also referred to as a metagenomic sample). With the advancement of Next Generation Sequencing (NGS) techniques, the number of metagenomic samples and the data size for each sample are increasing rapidly. Current metagenomic analysis is both data- and computation- intensive, especially when there are many species in a metagenomic sample, and each has a large number of sequences. As such, metagenomic analyses require extensive computational power. The increasing analytical requirements further augment the challenges for computation analysis. In this work, we have proposed Parallel-META 2.0, a metagenomic analysis software package, to cope with such needs for efficient and fast analyses of taxonomical and functional structures for microbial communities. Parallel-META 2.0 is an extended and improved version of Parallel-META 1.0, which enhances the taxonomical analysis using multiple databases, improves computation efficiency by optimized parallel computing, and supports interactive visualization of results in multiple views. Furthermore, it enables functional analysis for metagenomic samples including short-reads assembly, gene prediction and functional annotation. Therefore, it could provide accurate taxonomical and functional analyses of the metagenomic samples in high-throughput manner and on large scale. PMID:24595159
Su, Xiaoquan; Pan, Weihua; Song, Baoxing; Xu, Jian; Ning, Kang
2014-01-01
The metagenomic method directly sequences and analyses genome information from microbial communities. The main computational tasks for metagenomic analyses include taxonomical and functional structure analysis for all genomes in a microbial community (also referred to as a metagenomic sample). With the advancement of Next Generation Sequencing (NGS) techniques, the number of metagenomic samples and the data size for each sample are increasing rapidly. Current metagenomic analysis is both data- and computation- intensive, especially when there are many species in a metagenomic sample, and each has a large number of sequences. As such, metagenomic analyses require extensive computational power. The increasing analytical requirements further augment the challenges for computation analysis. In this work, we have proposed Parallel-META 2.0, a metagenomic analysis software package, to cope with such needs for efficient and fast analyses of taxonomical and functional structures for microbial communities. Parallel-META 2.0 is an extended and improved version of Parallel-META 1.0, which enhances the taxonomical analysis using multiple databases, improves computation efficiency by optimized parallel computing, and supports interactive visualization of results in multiple views. Furthermore, it enables functional analysis for metagenomic samples including short-reads assembly, gene prediction and functional annotation. Therefore, it could provide accurate taxonomical and functional analyses of the metagenomic samples in high-throughput manner and on large scale.
Locating hardware faults in a parallel computer
Archer, Charles J.; Megerian, Mark G.; Ratterman, Joseph D.; Smith, Brian E.
2010-04-13
Locating hardware faults in a parallel computer, including defining within a tree network of the parallel computer two or more sets of non-overlapping test levels of compute nodes of the network that together include all the data communications links of the network, each non-overlapping test level comprising two or more adjacent tiers of the tree; defining test cells within each non-overlapping test level, each test cell comprising a subtree of the tree including a subtree root compute node and all descendant compute nodes of the subtree root compute node within a non-overlapping test level; performing, separately on each set of non-overlapping test levels, an uplink test on all test cells in a set of non-overlapping test levels; and performing, separately from the uplink tests and separately on each set of non-overlapping test levels, a downlink test on all test cells in a set of non-overlapping test levels.
Zhang, Xiaohua; Wong, Sergio E; Lightstone, Felice C
2013-04-30
A mixed parallel scheme that combines message passing interface (MPI) and multithreading was implemented in the AutoDock Vina molecular docking program. The resulting program, named VinaLC, was tested on the petascale high performance computing (HPC) machines at Lawrence Livermore National Laboratory. To exploit the typical cluster-type supercomputers, thousands of docking calculations were dispatched by the master process to run simultaneously on thousands of slave processes, where each docking calculation takes one slave process on one node, and within the node each docking calculation runs via multithreading on multiple CPU cores and shared memory. Input and output of the program and the data handling within the program were carefully designed to deal with large databases and ultimately achieve HPC on a large number of CPU cores. Parallel performance analysis of the VinaLC program shows that the code scales up to more than 15K CPUs with a very low overhead cost of 3.94%. One million flexible compound docking calculations took only 1.4 h to finish on about 15K CPUs. The docking accuracy of VinaLC has been validated against the DUD data set by the re-docking of X-ray ligands and an enrichment study, 64.4% of the top scoring poses have RMSD values under 2.0 Å. The program has been demonstrated to have good enrichment performance on 70% of the targets in the DUD data set. An analysis of the enrichment factors calculated at various percentages of the screening database indicates VinaLC has very good early recovery of actives. PMID:23345155
Zhang, Xiaohua; Wong, Sergio E; Lightstone, Felice C
2013-04-30
A mixed parallel scheme that combines message passing interface (MPI) and multithreading was implemented in the AutoDock Vina molecular docking program. The resulting program, named VinaLC, was tested on the petascale high performance computing (HPC) machines at Lawrence Livermore National Laboratory. To exploit the typical cluster-type supercomputers, thousands of docking calculations were dispatched by the master process to run simultaneously on thousands of slave processes, where each docking calculation takes one slave process on one node, and within the node each docking calculation runs via multithreading on multiple CPU cores and shared memory. Input and output of the program and the data handling within the program were carefully designed to deal with large databases and ultimately achieve HPC on a large number of CPU cores. Parallel performance analysis of the VinaLC program shows that the code scales up to more than 15K CPUs with a very low overhead cost of 3.94%. One million flexible compound docking calculations took only 1.4 h to finish on about 15K CPUs. The docking accuracy of VinaLC has been validated against the DUD data set by the re-docking of X-ray ligands and an enrichment study, 64.4% of the top scoring poses have RMSD values under 2.0 Å. The program has been demonstrated to have good enrichment performance on 70% of the targets in the DUD data set. An analysis of the enrichment factors calculated at various percentages of the screening database indicates VinaLC has very good early recovery of actives.
Computing contingency statistics in parallel.
Bennett, Janine Camille; Thompson, David; Pebay, Philippe Pierre
2010-09-01
Statistical analysis is typically used to reduce the dimensionality of and infer meaning from data. A key challenge of any statistical analysis package aimed at large-scale, distributed data is to address the orthogonal issues of parallel scalability and numerical stability. Many statistical techniques, e.g., descriptive statistics or principal component analysis, are based on moments and co-moments and, using robust online update formulas, can be computed in an embarrassingly parallel manner, amenable to a map-reduce style implementation. In this paper we focus on contingency tables, through which numerous derived statistics such as joint and marginal probability, point-wise mutual information, information entropy, and {chi}{sup 2} independence statistics can be directly obtained. However, contingency tables can become large as data size increases, requiring a correspondingly large amount of communication between processors. This potential increase in communication prevents optimal parallel speedup and is the main difference with moment-based statistics where the amount of inter-processor communication is independent of data size. Here we present the design trade-offs which we made to implement the computation of contingency tables in parallel.We also study the parallel speedup and scalability properties of our open source implementation. In particular, we observe optimal speed-up and scalability when the contingency statistics are used in their appropriate context, namely, when the data input is not quasi-diffuse.
High Performance Parallel Architectures
NASA Technical Reports Server (NTRS)
El-Ghazawi, Tarek; Kaewpijit, Sinthop
1998-01-01
Traditional remote sensing instruments are multispectral, where observations are collected at a few different spectral bands. Recently, many hyperspectral instruments, that can collect observations at hundreds of bands, have been operational. Furthermore, there have been ongoing research efforts on ultraspectral instruments that can produce observations at thousands of spectral bands. While these remote sensing technology developments hold great promise for new findings in the area of Earth and space science, they present many challenges. These include the need for faster processing of such increased data volumes, and methods for data reduction. Dimension Reduction is a spectral transformation, aimed at concentrating the vital information and discarding redundant data. One such transformation, which is widely used in remote sensing, is the Principal Components Analysis (PCA). This report summarizes our progress on the development of a parallel PCA and its implementation on two Beowulf cluster configuration; one with fast Ethernet switch and the other with a Myrinet interconnection. Details of the implementation and performance results, for typical sets of multispectral and hyperspectral NASA remote sensing data, are presented and analyzed based on the algorithm requirements and the underlying machine configuration. It will be shown that the PCA application is quite challenging and hard to scale on Ethernet-based clusters. However, the measurements also show that a high- performance interconnection network, such as Myrinet, better matches the high communication demand of PCA and can lead to a more efficient PCA execution.
The new landscape of parallel computer architecture
NASA Astrophysics Data System (ADS)
Shalf, John
2007-07-01
The past few years has seen a sea change in computer architecture that will impact every facet of our society as every electronic device from cell phone to supercomputer will need to confront parallelism of unprecedented scale. Whereas the conventional multicore approach (2, 4, and even 8 cores) adopted by the computing industry will eventually hit a performance plateau, the highest performance per watt and per chip area is achieved using manycore technology (hundreds or even thousands of cores). However, fully unleashing the potential of the manycore approach to ensure future advances in sustained computational performance will require fundamental advances in computer architecture and programming models that are nothing short of reinventing computing. In this paper we examine the reasons behind the movement to exponentially increasing parallelism, and its ramifications for system design, applications and programming models.
NASA Astrophysics Data System (ADS)
Gowtham, S.
Small clusters of gallium oxide, technologically important high temperature ceramic, together with interaction of nucleic acid bases with graphene and small-diameter carbon nanotube are focus of first principles calculations in this work. A high performance parallel computing platform is also developed to perform these calculations at Michigan Tech. First principles calculations are based on density functional theory employing either local density or gradient-corrected approximation together with plane wave and Gaussian basis sets. The bulk Ga2O3 is known to be a very good candidate for fabricating electronic devices that operate at high temperatures. To explore the properties of Ga2O3 at nanoscale, we have performed a systematic theoretical study on the small polyatomic gallium oxide clusters. The calculated results find that all lowest energy isomers of GamO n clusters are dominated by the Ga-O bonds over the metal-metal or the oxygen-oxygen bonds. Analysis of atomic charges suggest the clusters to be highly ionic similar to the case of bulk Ga2O3. In the study of sequential oxidation of these clusters starting from Ga3O, it is found that the most stable isomers display up to four different backbones of constituent atoms. Furthermore, the predicted configuration of the ground state of Ga2O is recently confirmed by the experimental results of Neumark's group. Guided by the results of calculations the study of gallium oxide clusters, performance related challenge of computational simulations, of producing high performance computers/platforms, has been addressed. Several engineering aspects were thoroughly studied during the design, development and implementation of the high performance parallel computing platform, RAMA, at Michigan Tech. In an attempt to stay true to the principles of Beowulf revolution, the RAMA cluster was extensively customized to make it easy to understand, and use - for administrators as well as end-users. Following the results of benchmark
Computer-Aided Parallelizer and Optimizer
NASA Technical Reports Server (NTRS)
Jin, Haoqiang
2011-01-01
The Computer-Aided Parallelizer and Optimizer (CAPO) automates the insertion of compiler directives (see figure) to facilitate parallel processing on Shared Memory Parallel (SMP) machines. While CAPO currently is integrated seamlessly into CAPTools (developed at the University of Greenwich, now marketed as ParaWise), CAPO was independently developed at Ames Research Center as one of the components for the Legacy Code Modernization (LCM) project. The current version takes serial FORTRAN programs, performs interprocedural data dependence analysis, and generates OpenMP directives. Due to the widely supported OpenMP standard, the generated OpenMP codes have the potential to run on a wide range of SMP machines. CAPO relies on accurate interprocedural data dependence information currently provided by CAPTools. Compiler directives are generated through identification of parallel loops in the outermost level, construction of parallel regions around parallel loops and optimization of parallel regions, and insertion of directives with automatic identification of private, reduction, induction, and shared variables. Attempts also have been made to identify potential pipeline parallelism (implemented with point-to-point synchronization). Although directives are generated automatically, user interaction with the tool is still important for producing good parallel codes. A comprehensive graphical user interface is included for users to interact with the parallelization process.
Parallel algorithms for optical digital computers
Huang, A.
1983-01-01
Conventional computers suffer from several communication bottlenecks which fundamentally limit their performance. These bottlenecks are characterised by an address-dependent sequential transfer of information which arises from the need to time-multiplex information over a limited number of interconnections. An optical digital computer based on a classical finite state machine can be shown to be free of these bottlenecks. Such a processor would be unique since it would be capable of modifying its entire state space each cycle while conventional computers can only alter a few bits. New algorithms are needed to manage and use this capability. A technique based on recognising a particular symbol in parallel and replacing it in parallel with another symbol is suggested. Examples using this parallel symbolic substitution to perform binary addition and binary incrementation are presented. Applications involving Boolean logic, functional programming languages, production rule driven artificial intelligence, and molecular chemistry are also discussed. 12 references.
Finite element computation with parallel VLSI
NASA Technical Reports Server (NTRS)
Mcgregor, J.; Salama, M.
1983-01-01
This paper describes a parallel processing computer consisting of a 16-bit microcomputer as a master processor which controls and coordinates the activities of 8086/8087 VLSI chip set slave processors working in parallel. The hardware is inexpensive and can be flexibly configured and programmed to perform various functions. This makes it a useful research tool for the development of, and experimentation with parallel mathematical algorithms. Application of the hardware to computational tasks involved in the finite element analysis method is demonstrated by the generation and assembly of beam finite element stiffness matrices. A number of possible schemes for the implementation of N-elements on N- or n-processors (N is greater than n) are described, and the speedup factors of their time consumption are determined as a function of the number of available parallel processors.
Efficient communication in massively parallel computers
Cypher, R.E.
1989-01-01
A fundamental operation in parallel computation is sorting. Sorting is important not only because it is required by many algorithms, but also because it can be used to implement irregular, pointer-based communication. The author studies two algorithms for sorting in massively parallel computers. First, he examines Shellsort. Shellsort is a sorting algorithm that is based on a sequence of parameters called increments. Shellsort can be used to create a parallel sorting device known as a sorting network. Researchers have suggested that if the correct increment sequence is used, an optimal size sorting network can be obtained. All published increment sequences have been monotonically decreasing. He shows that no monotonically decreasing increment sequence will yield an optimal size sorting network. Second, he presents a sorting algorithm called Cubesort. Cubesort is the fastest known sorting algorithm for a variety of parallel computers aver a wide range of parameters. He also presents a paradigm for developing parallel algorithms that have efficient communication. The paradigm, called the data reduction paradigm, consists of using a divide-and-conquer strategy. Both the division and combination phases of the divide-and-conquer algorithm may require irregular, pointer-based communication between processors. However, the problem is divided so as to limit the amount of data that must be communicated. As a result the communication can be performed efficiently. He presents data reduction algorithms for the image component labeling problem, the closest pair problem and four versions of the parallel prefix problem.
Optics Program Modified for Multithreaded Parallel Computing
NASA Technical Reports Server (NTRS)
Lou, John; Bedding, Dave; Basinger, Scott
2006-01-01
A powerful high-performance computer program for simulating and analyzing adaptive and controlled optical systems has been developed by modifying the serial version of the Modeling and Analysis for Controlled Optical Systems (MACOS) program to impart capabilities for multithreaded parallel processing on computing systems ranging from supercomputers down to Symmetric Multiprocessing (SMP) personal computers. The modifications included the incorporation of OpenMP, a portable and widely supported application interface software, that can be used to explicitly add multithreaded parallelism to an application program under a shared-memory programming model. OpenMP was applied to parallelize ray-tracing calculations, one of the major computing components in MACOS. Multithreading is also used in the diffraction propagation of light in MACOS based on pthreads [POSIX Thread, (where "POSIX" signifies a portable operating system for UNIX)]. In tests of the parallelized version of MACOS, the speedup in ray-tracing calculations was found to be linear, or proportional to the number of processors, while the speedup in diffraction calculations ranged from 50 to 60 percent, depending on the type and number of processors. The parallelized version of MACOS is portable, and, to the user, its interface is basically the same as that of the original serial version of MACOS.
NASA Technical Reports Server (NTRS)
Sun, Xian-He; Moitra, Stuti
1996-01-01
Various tridiagonal solvers have been proposed in recent years for different parallel platforms. In this paper, the performance of three tridiagonal solvers, namely, the parallel partition LU algorithm, the parallel diagonal dominant algorithm, and the reduced diagonal dominant algorithm, is studied. These algorithms are designed for distributed-memory machines and are tested on an Intel Paragon and an IBM SP2 machines. Measured results are reported in terms of execution time and speedup. Analytical study are conducted for different communication topologies and for different tridiagonal systems. The measured results match the analytical results closely. In addition to address implementation issues, performance considerations such as problem sizes and models of speedup are also discussed.
Parallel Pascal - An extended Pascal for parallel computers
NASA Technical Reports Server (NTRS)
Reeves, A. P.
1984-01-01
Parallel Pascal is an extended version of the conventional serial Pascal programming language which includes a convenient syntax for specifying array operations. It is upward compatible with standard Pascal and involves only a small number of carefully chosen new features. Parallel Pascal was developed to reduce the semantic gap between standard Pascal and a large range of highly parallel computers. Two important design goals of Parallel Pascal were efficiency and portability. Portability is particularly difficult to achieve since different parallel computers frequently have very different capabilities.
Opportunities in computational mechanics: Advances in parallel computing
Lesar, R.A.
1999-02-01
In this paper, the authors will discuss recent advances in computing power and the prospects for using these new capabilities for studying plasticity and failure. They will first review the new capabilities made available with parallel computing. They will discuss how these machines perform and how well their architecture might work on materials issues. Finally, they will give some estimates on the size of problems possible using these computers.
Synchronizing compute node time bases in a parallel computer
Chen, Dong; Faraj, Daniel A; Gooding, Thomas M; Heidelberger, Philip
2015-01-27
Synchronizing time bases in a parallel computer that includes compute nodes organized for data communications in a tree network, where one compute node is designated as a root, and, for each compute node: calculating data transmission latency from the root to the compute node; configuring a thread as a pulse waiter; initializing a wakeup unit; and performing a local barrier operation; upon each node completing the local barrier operation, entering, by all compute nodes, a global barrier operation; upon all nodes entering the global barrier operation, sending, to all the compute nodes, a pulse signal; and for each compute node upon receiving the pulse signal: waking, by the wakeup unit, the pulse waiter; setting a time base for the compute node equal to the data transmission latency between the root node and the compute node; and exiting the global barrier operation.
Synchronizing compute node time bases in a parallel computer
Chen, Dong; Faraj, Daniel A; Gooding, Thomas M; Heidelberger, Philip
2014-12-30
Synchronizing time bases in a parallel computer that includes compute nodes organized for data communications in a tree network, where one compute node is designated as a root, and, for each compute node: calculating data transmission latency from the root to the compute node; configuring a thread as a pulse waiter; initializing a wakeup unit; and performing a local barrier operation; upon each node completing the local barrier operation, entering, by all compute nodes, a global barrier operation; upon all nodes entering the global barrier operation, sending, to all the compute nodes, a pulse signal; and for each compute node upon receiving the pulse signal: waking, by the wakeup unit, the pulse waiter; setting a time base for the compute node equal to the data transmission latency between the root node and the compute node; and exiting the global barrier operation.
Java Parallel Secure Stream for Grid Computing
Chen, Jie; Akers, Walter; Chen, Ying; Watson, William
2001-09-01
The emergence of high speed wide area networks makes grid computing a reality. However grid applications that need reliable data transfer still have difficulties to achieve optimal TCP performance due to network tuning of TCP window size to improve the bandwidth and to reduce latency on a high speed wide area network. This paper presents a pure Java package called JPARSS (Java Par-allel Secure Stream) that divides data into partitions that are sent over several parallel Java streams simultaneously and allows Java or Web applications to achieve optimal TCP performance in a gird environment without the necessity of tuning the TCP window size. Several experimental results are provided to show that using parallel stream is more effective than tuning TCP window size. In addi-tion X.509 certificate based single sign-on mechanism and SSL based connection establishment are integrated into this package. Finally a few applications using this package will be discussed.
Pronk, Sander; Pouya, Iman; Lundborg, Magnus; Rotskoff, Grant; Wesén, Björn; Kasson, Peter M; Lindahl, Erik
2015-06-01
Computational chemistry and other simulation fields are critically dependent on computing resources, but few problems scale efficiently to the hundreds of thousands of processors available in current supercomputers-particularly for molecular dynamics. This has turned into a bottleneck as new hardware generations primarily provide more processing units rather than making individual units much faster, which simulation applications are addressing by increasingly focusing on sampling with algorithms such as free-energy perturbation, Markov state modeling, metadynamics, or milestoning. All these rely on combining results from multiple simulations into a single observation. They are potentially powerful approaches that aim to predict experimental observables directly, but this comes at the expense of added complexity in selecting sampling strategies and keeping track of dozens to thousands of simulations and their dependencies. Here, we describe how the distributed execution framework Copernicus allows the expression of such algorithms in generic workflows: dataflow programs. Because dataflow algorithms explicitly state dependencies of each constituent part, algorithms only need to be described on conceptual level, after which the execution is maximally parallel. The fully automated execution facilitates the optimization of these algorithms with adaptive sampling, where undersampled regions are automatically detected and targeted without user intervention. We show how several such algorithms can be formulated for computational chemistry problems, and how they are executed efficiently with many loosely coupled simulations using either distributed or parallel resources with Copernicus. PMID:26575558
Pronk, Sander; Pouya, Iman; Lundborg, Magnus; Rotskoff, Grant; Wesén, Björn; Kasson, Peter M; Lindahl, Erik
2015-06-01
Computational chemistry and other simulation fields are critically dependent on computing resources, but few problems scale efficiently to the hundreds of thousands of processors available in current supercomputers-particularly for molecular dynamics. This has turned into a bottleneck as new hardware generations primarily provide more processing units rather than making individual units much faster, which simulation applications are addressing by increasingly focusing on sampling with algorithms such as free-energy perturbation, Markov state modeling, metadynamics, or milestoning. All these rely on combining results from multiple simulations into a single observation. They are potentially powerful approaches that aim to predict experimental observables directly, but this comes at the expense of added complexity in selecting sampling strategies and keeping track of dozens to thousands of simulations and their dependencies. Here, we describe how the distributed execution framework Copernicus allows the expression of such algorithms in generic workflows: dataflow programs. Because dataflow algorithms explicitly state dependencies of each constituent part, algorithms only need to be described on conceptual level, after which the execution is maximally parallel. The fully automated execution facilitates the optimization of these algorithms with adaptive sampling, where undersampled regions are automatically detected and targeted without user intervention. We show how several such algorithms can be formulated for computational chemistry problems, and how they are executed efficiently with many loosely coupled simulations using either distributed or parallel resources with Copernicus.
Parallel Algebraic Multigrid Methods - High Performance Preconditioners
Yang, U M
2004-11-11
The development of high performance, massively parallel computers and the increasing demands of computationally challenging applications have necessitated the development of scalable solvers and preconditioners. One of the most effective ways to achieve scalability is the use of multigrid or multilevel techniques. Algebraic multigrid (AMG) is a very efficient algorithm for solving large problems on unstructured grids. While much of it can be parallelized in a straightforward way, some components of the classical algorithm, particularly the coarsening process and some of the most efficient smoothers, are highly sequential, and require new parallel approaches. This chapter presents the basic principles of AMG and gives an overview of various parallel implementations of AMG, including descriptions of parallel coarsening schemes and smoothers, some numerical results as well as references to existing software packages.
Broadcasting a message in a parallel computer
Berg, Jeremy E.; Faraj, Ahmad A.
2011-08-02
Methods, systems, and products are disclosed for broadcasting a message in a parallel computer. The parallel computer includes a plurality of compute nodes connected together using a data communications network. The data communications network optimized for point to point data communications and is characterized by at least two dimensions. The compute nodes are organized into at least one operational group of compute nodes for collective parallel operations of the parallel computer. One compute node of the operational group assigned to be a logical root. Broadcasting a message in a parallel computer includes: establishing a Hamiltonian path along all of the compute nodes in at least one plane of the data communications network and in the operational group; and broadcasting, by the logical root to the remaining compute nodes, the logical root's message along the established Hamiltonian path.
LEWICE droplet trajectory calculations on a parallel computer
NASA Technical Reports Server (NTRS)
Caruso, Steven C.
1993-01-01
A parallel computer implementation (128 processors) of LEWICE, a NASA Lewis code used to predict the time-dependent ice accretion process for two-dimensional aerodynamic bodies of simple geometries, is described. Two-dimensional parallel droplet trajectory calculations are performed to demonstrate the potential benefits of applying parallel processing to ice accretion analysis. Parallel performance is evaluated as a function of the number of trajectories and the number of processors. For comparison, similar trajectory calculations are performed on single-processor Cray computers, and the best parallel results are found to be 33 and 23 times faster, respectively, than those of the Cray XMP and YMP.
Efficient Parallel Engineering Computing on Linux Workstations
NASA Technical Reports Server (NTRS)
Lou, John Z.
2010-01-01
A C software module has been developed that creates lightweight processes (LWPs) dynamically to achieve parallel computing performance in a variety of engineering simulation and analysis applications to support NASA and DoD project tasks. The required interface between the module and the application it supports is simple, minimal and almost completely transparent to the user applications, and it can achieve nearly ideal computing speed-up on multi-CPU engineering workstations of all operating system platforms. The module can be integrated into an existing application (C, C++, Fortran and others) either as part of a compiled module or as a dynamically linked library (DLL).
An Expert Assistant for Computer Aided Parallelization
NASA Technical Reports Server (NTRS)
Jost, Gabriele; Chun, Robert; Jin, Haoqiang; Labarta, Jesus; Gimenez, Judit
2004-01-01
The prototype implementation of an expert system was developed to assist the user in the computer aided parallelization process. The system interfaces to tools for automatic parallelization and performance analysis. By fusing static program structure information and dynamic performance analysis data the expert system can help the user to filter, correlate, and interpret the data gathered by the existing tools. Sections of the code that show poor performance and require further attention are rapidly identified and suggestions for improvements are presented to the user. In this paper we describe the components of the expert system and discuss its interface to the existing tools. We present a case study to demonstrate the successful use in full scale scientific applications.
Efficient parallel global garbage collection on massively parallel computers
Kamada, Tomio; Matsuoka, Satoshi; Yonezawa, Akinori
1994-12-31
On distributed-memory high-performance MPPs where processors are interconnected by an asynchronous network, efficient Garbage Collection (GC) becomes difficult due to inter-node references and references within pending, unprocessed messages. The parallel global GC algorithm (1) takes advantage of reference locality, (2) efficiently traverses references over nodes, (3) admits minimum pause time of ongoing computations, and (4) has been shown to scale up to 1024 node MPPs. The algorithm employs a global weight counting scheme to substantially reduce message traffic. The two methods for confirming the arrival of pending messages are used: one counts numbers of messages and the other uses network `bulldozing.` Performance evaluation in actual implementations on a multicomputer with 32-1024 nodes, Fujitsu AP1000, reveals various favorable properties of the algorithm.
Implementing clips on a parallel computer
NASA Technical Reports Server (NTRS)
Riley, Gary
1987-01-01
The C language integrated production system (CLIPS) is a forward chaining rule based language to provide training and delivery for expert systems. Conceptually, rule based languages have great potential for benefiting from the inherent parallelism of the algorithms that they employ. During each cycle of execution, a knowledge base of information is compared against a set of rules to determine if any rules are applicable. Parallelism also can be employed for use with multiple cooperating expert systems. To investigate the potential benefits of using a parallel computer to speed up the comparison of facts to rules in expert systems, a parallel version of CLIPS was developed for the FLEX/32, a large grain parallel computer. The FLEX implementation takes a macroscopic approach in achieving parallelism by splitting whole sets of rules among several processors rather than by splitting the components of an individual rule among processors. The parallel CLIPS prototype demonstrates the potential advantages of integrating expert system tools with parallel computers.
A parallel Jacobson-Oksman optimization algorithm. [parallel processing (computers)
NASA Technical Reports Server (NTRS)
Straeter, T. A.; Markos, A. T.
1975-01-01
A gradient-dependent optimization technique which exploits the vector-streaming or parallel-computing capabilities of some modern computers is presented. The algorithm, derived by assuming that the function to be minimized is homogeneous, is a modification of the Jacobson-Oksman serial minimization method. In addition to describing the algorithm, conditions insuring the convergence of the iterates of the algorithm and the results of numerical experiments on a group of sample test functions are presented. The results of these experiments indicate that this algorithm will solve optimization problems in less computing time than conventional serial methods on machines having vector-streaming or parallel-computing capabilities.
A Computational Fluid Dynamics Algorithm on a Massively Parallel Computer
NASA Technical Reports Server (NTRS)
Jespersen, Dennis C.; Levit, Creon
1989-01-01
The discipline of computational fluid dynamics is demanding ever-increasing computational power to deal with complex fluid flow problems. We investigate the performance of a finite-difference computational fluid dynamics algorithm on a massively parallel computer, the Connection Machine. Of special interest is an implicit time-stepping algorithm; to obtain maximum performance from the Connection Machine, it is necessary to use a nonstandard algorithm to solve the linear systems that arise in the implicit algorithm. We find that the Connection Machine ran achieve very high computation rates on both explicit and implicit algorithms. The performance of the Connection Machine puts it in the same class as today's most powerful conventional supercomputers.
Parallel computing: One opportunity, four challenges
Gaudiot, J.-L.
1989-12-31
The author reviews briefly the area of parallel computer processing. This area has been expanding at a great rate in the past decade. Great strides have been made in the hardware area, and in the speed of performance of chips. However to some degree the hardware area is beginning to run into basic physical speed limits, which will slow the rate of advance of this area simply because of physical limitations. The author looks at ways that computer architecture, and software applications, can work to continue the rate of increase in computing power which has occurred over the past decade. Four particular areas are mentioned: programmability; communication network design; reliable operation; performance evaluation and benchmarking.
Reservoir Thermal Recover Simulation on Parallel Computers
NASA Astrophysics Data System (ADS)
Li, Baoyan; Ma, Yuanle
The rapid development of parallel computers has provided a hardware background for massive refine reservoir simulation. However, the lack of parallel reservoir simulation software has blocked the application of parallel computers on reservoir simulation. Although a variety of parallel methods have been studied and applied to black oil, compositional, and chemical model numerical simulations, there has been limited parallel software available for reservoir simulation. Especially, the parallelization study of reservoir thermal recovery simulation has not been fully carried out, because of the complexity of its models and algorithms. The authors make use of the message passing interface (MPI) standard communication library, the domain decomposition method, the block Jacobi iteration algorithm, and the dynamic memory allocation technique to parallelize their serial thermal recovery simulation software NUMSIP, which is being used in petroleum industry in China. The parallel software PNUMSIP was tested on both IBM SP2 and Dawn 1000A distributed-memory parallel computers. The experiment results show that the parallelization of I/O has great effects on the efficiency of parallel software PNUMSIP; the data communication bandwidth is also an important factor, which has an influence on software efficiency. Keywords: domain decomposition method, block Jacobi iteration algorithm, reservoir thermal recovery simulation, distributed-memory parallel computer
On evaluating parallel computer systems
NASA Technical Reports Server (NTRS)
Adams, George B., III; Brown, Robert L.; Denning, Peter J.
1985-01-01
A workshop was held in an attempt to program real problems on the MIT Static Data Flow Machine. Most of the architecture of the machine was specified but some parts were incomplete. The main purpose for the workshop was to explore principles for the evaluation of computer systems employing new architectures. Principles explored were: (1) evaluation must be an integral, ongoing part of a project to develop a computer of radically new architecture; (2) the evaluation should seek to measure the usability of the system as well as its performance; (3) users from the application domains must be an integral part of the evaluation process; and (4) evaluation results should be fed back into the design process. It is concluded that the general organizational principles are achievable in practice from this workshop.
Parallelized reliability estimation of reconfigurable computer networks
NASA Technical Reports Server (NTRS)
Nicol, David M.; Das, Subhendu; Palumbo, Dan
1990-01-01
A parallelized system, ASSURE, for computing the reliability of embedded avionics flight control systems which are able to reconfigure themselves in the event of failure is described. ASSURE accepts a grammar that describes a reliability semi-Markov state-space. From this it creates a parallel program that simultaneously generates and analyzes the state-space, placing upper and lower bounds on the probability of system failure. ASSURE is implemented on a 32-node Intel iPSC/860, and has achieved high processor efficiencies on real problems. Through a combination of improved algorithms, exploitation of parallelism, and use of an advanced microprocessor architecture, ASSURE has reduced the execution time on substantial problems by a factor of one thousand over previous workstation implementations. Furthermore, ASSURE's parallel execution rate on the iPSC/860 is an order of magnitude faster than its serial execution rate on a Cray-2 supercomputer. While dynamic load balancing is necessary for ASSURE's good performance, it is needed only infrequently; the particular method of load balancing used does not substantially affect performance.
Parallel Performance of a Combustion Chemistry Simulation
Skinner, Gregg; Eigenmann, Rudolf
1995-01-01
We used a description of a combustion simulation's mathematical and computational methods to develop a version for parallel execution. The result was a reasonable performance improvement on small numbers of processors. We applied several important programming techniques, which we describe, in optimizing the application. This work has implications for programming languages, compiler design, and software engineering.
Seismic imaging on massively parallel computers
Ober, C.C.; Oldfield, R.A.; Womble, D.E.; Mosher, C.C.
1997-07-01
A key to reducing the risks and costs associated with oil and gas exploration is the fast, accurate imaging of complex geologies, such as salt domes in the Gulf of Mexico and overthrust regions in US onshore regions. Pre-stack depth migration generally yields the most accurate images, and one approach to this is to solve the scalar-wave equation using finite differences. Current industry computational capabilities are insufficient for the application of finite-difference, 3-D, prestack, depth-migration algorithms. High performance computers and state-of-the-art algorithms and software are required to meet this need. As part of an ongoing ACTI project funded by the US Department of Energy, the authors have developed a finite-difference, 3-D prestack, depth-migration code for massively parallel computer systems. The goal of this work is to demonstrate that massively parallel computers (thousands of processors) can be used efficiently for seismic imaging, and that sufficient computing power exists (or soon will exist) to make finite-difference, prestack, depth migration practical for oil and gas exploration.
Parallel computations and control of adaptive structures
NASA Technical Reports Server (NTRS)
Park, K. C.; Alvin, Kenneth F.; Belvin, W. Keith; Chong, K. P. (Editor); Liu, S. C. (Editor); Li, J. C. (Editor)
1991-01-01
The equations of motion for structures with adaptive elements for vibration control are presented for parallel computations to be used as a software package for real-time control of flexible space structures. A brief introduction of the state-of-the-art parallel computational capability is also presented. Time marching strategies are developed for an effective use of massive parallel mapping, partitioning, and the necessary arithmetic operations. An example is offered for the simulation of control-structure interaction on a parallel computer and the impact of the approach presented for applications in other disciplines than aerospace industry is assessed.
Data-parallel algorithms for image computing
NASA Astrophysics Data System (ADS)
Carlotto, Mark J.
1990-11-01
Data-parallel algorithms for image computing on the Connection Machine are described. After a brief review of some basic programming concepts in *Lip, a parallel extension of Common Lisp, data-parallel programming paradigms based on a local (diffusion-like) model of computation, the scan model of computation, a general interprocessor communications model, and a region-based model are introduced. Algorithms for connected component labeling, distance transformation, Voronoi diagrams, finding minimum cost paths, local means, shape-from-shading, hidden surface calculations, affine transformation, oblique parallel projection, and spatial operations over regions are presented. An new algorithm for interpolating irregularly spaced data via Voronoi diagrams is also described.
Running Geant on T. Node parallel computer
Jejcic, A.; Maillard, J.; Silva, J. ); Mignot, B. )
1990-08-01
AnInmos transputer-based computer has been utilized to overcome the difficulties due to the limitations on the processing abilities of event parallelism and multiprocessor farms (i.e., the so called bus-crisis) and the concern regarding the growing sizes of databases typical in High Energy Physics. This study was done on the T.Node parallel computer manufactured by TELMAT. Detailed figures are reported concerning the event parallelization. (AIP)
Modified mesh-connected parallel computers
Carlson, D.A. )
1988-10-01
The mesh-connected parallel computer is an important parallel processing organization that has been used in the past for the design of supercomputing systems. In this paper, the authors explore modifications of a mesh-connected parallel computer for the purpose of increasing the efficiency of executing important application programs. These modifications are made by adding one or more global mesh structures to the processing array. They show how our modifications allow asymptotic improvements in the efficiency of executing computations having low to medium interprocessor communication requirements (e.g., tree computations, prefix computations, finding the connected components of a graph). For computations with high interprocessor communication requirements such as sorting, they show that they offer no speedup. They also compare the modified mesh-connected parallel computer to other similar organizations including the pyramid, the X-tree, and the mesh-of-trees.
Parallel algorithm for computing points on a computation front hyperplane
NASA Astrophysics Data System (ADS)
Krasnov, M. M.
2015-01-01
A parallel algorithm for computing points on a computation front hyperplane is described. This task arises in the computation of a quantity defined on a multidimensional rectangular domain. Three-dimensional domains are usually discussed, but the material is given in the general form when the number of measurements is at least two. When the values of a quantity at different points are internally independent (which is frequently the case), the corresponding computations are independent as well and can be performed in parallel. However, if there are internal dependences (as, for example, in the Gauss-Seidel method for systems of linear equations), then the order of scanning points of the domain is an important issue. A conventional approach in this case is to form a computation front hyperplane (a usual plane in the three-dimensional case and a line in the two-dimensional case) that moves linearly across the domain at a certain angle. At every step in the course of motion of this hyperplane, its intersection points with the domain can be treated independently and, hence, in parallel, but the steps themselves are executed sequentially. At different steps, the intersection of the hyperplane with the entire domain can have a rather complex geometry and the search for all points of the domain lying on the hyperplane at a given step is a nontrivial problem. This problem (i.e., the computation of the coordinates of points lying in the intersection of the domain with the hyperplane at a given step in the course of hyperplane motion) is addressed below. The computations over the points of the hyperplane can be executed in parallel.
Computation and parallel implementation for early vision
NASA Technical Reports Server (NTRS)
Gualtieri, J. Anthony
1990-01-01
The problem of early vision is to transform one or more retinal illuminance images-pixel arrays-to image representations built out of such primitive visual features such as edges, regions, disparities, and clusters. These transformed representations form the input to later vision stages that perform higher level vision tasks including matching and recognition. Researchers developed algorithms for: (1) edge finding in the scale space formulation; (2) correlation methods for computing matches between pairs of images; and (3) clustering of data by neural networks. These algorithms are formulated for parallel implementation of SIMD machines, such as the Massively Parallel Processor, a 128 x 128 array processor with 1024 bits of local memory per processor. For some cases, researchers can show speedups of three orders of magnitude over serial implementations.
Review of parallel computing methods and tools for FPGA technology
NASA Astrophysics Data System (ADS)
Cieszewski, Radosław; Linczuk, Maciej; Pozniak, Krzysztof; Romaniuk, Ryszard
2013-10-01
Parallel computing is emerging as an important area of research in computer architectures and software systems. Many algorithms can be greatly accelerated using parallel computing techniques. Specialized parallel computer architectures are used for accelerating speci c tasks. High-Energy Physics Experiments measuring systems often use FPGAs for ne-grained computation. FPGA combines many bene ts of both software and ASIC implementations. Like software, the mapped circuit is exible, and can be recon gured over the lifetime of the system. FPGAs therefore have the potential to achieve far greater performance than software as a result of bypassing the fetch-decode-execute operations of traditional processors, and possibly exploiting a greater level of parallelism. Creating parallel programs implemented in FPGAs is not trivial. This paper presents existing methods and tools for ne-grained computation implemented in FPGA using Behavioral Description and High Level Programming Languages.
Optimal dynamic remapping of parallel computations
NASA Technical Reports Server (NTRS)
Nicol, David M.; Reynolds, Paul F., Jr.
1987-01-01
A large class of computations are characterized by a sequence of phases, with phase changes occurring unpredictably. The decision problem was considered regarding the remapping of workload to processors in a parallel computation when the utility of remapping and the future behavior of the workload is uncertain, and phases exhibit stable execution requirements during a given phase, but requirements may change radically between phases. For these problems a workload assignment generated for one phase may hinder performance during the next phase. This problem is treated formally for a probabilistic model of computation with at most two phases. The fundamental problem of balancing the expected remapping performance gain against the delay cost was addressed. Stochastic dynamic programming is used to show that the remapping decision policy minimizing the expected running time of the computation has an extremely simple structure. Because the gain may not be predictable, the performance of a heuristic policy that does not require estimnation of the gain is examined. The heuristic method's feasibility is demonstrated by its use on an adaptive fluid dynamics code on a multiprocessor. The results suggest that except in extreme cases, the remapping decision problem is essentially that of dynamically determining whether gain can be achieved by remapping after a phase change. The results also suggest that this heuristic is applicable to computations with more than two phases.
Hypercluster - Parallel processing for computational mechanics
NASA Technical Reports Server (NTRS)
Blech, Richard A.
1988-01-01
An account is given of the development status, performance capabilities and implications for further development of NASA-Lewis' testbed 'hypercluster' parallel computer network, in which multiple processors communicate through a shared memory. Processors have local as well as shared memory; the hypercluster is expanded in the same manner as the hypercube, with processor clusters replacing the normal single processor node. The NASA-Lewis machine has three nodes with a vector personality and one node with a scalar personality. Each of the vector nodes uses four board-level vector processors, while the scalar node uses four general-purpose microcomputer boards.
Nonisothermal multiphase subsurface transport on parallel computers
Martinez, M.J.; Hopkins, P.L.; Shadid, J.N.
1997-10-01
We present a numerical method for nonisothermal, multiphase subsurface transport in heterogeneous porous media. The mathematical model considers nonisothermal two-phase (liquid/gas) flow, including capillary pressure effects, binary diffusion in the gas phase, conductive, latent, and sensible heat transport. The Galerkin finite element method is used for spatial discretization, and temporal integration is accomplished via a predictor/corrector scheme. Message-passing and domain decomposition techniques are used for implementing a scalable algorithm for distributed memory parallel computers. An illustrative application is shown to demonstrate capabilities and performance.
Parallel algorithms for mapping pipelined and parallel computations
NASA Technical Reports Server (NTRS)
Nicol, David M.
1988-01-01
Many computational problems in image processing, signal processing, and scientific computing are naturally structured for either pipelined or parallel computation. When mapping such problems onto a parallel architecture it is often necessary to aggregate an obvious problem decomposition. Even in this context the general mapping problem is known to be computationally intractable, but recent advances have been made in identifying classes of problems and architectures for which optimal solutions can be found in polynomial time. Among these, the mapping of pipelined or parallel computations onto linear array, shared memory, and host-satellite systems figures prominently. This paper extends that work first by showing how to improve existing serial mapping algorithms. These improvements have significantly lower time and space complexities: in one case a published O(nm sup 3) time algorithm for mapping m modules onto n processors is reduced to an O(nm log m) time complexity, and its space requirements reduced from O(nm sup 2) to O(m). Run time complexity is further reduced with parallel mapping algorithms based on these improvements, which run on the architecture for which they create the mappings.
Kepper, Nick; Ettig, Ramona; Dickmann, Frank; Stehr, Rene; Grosveld, Frank G; Wedemann, Gero; Knoch, Tobias A
2010-01-01
Especially in the life-science and the health-care sectors the huge IT requirements are imminent due to the large and complex systems to be analysed and simulated. Grid infrastructures play here a rapidly increasing role for research, diagnostics, and treatment, since they provide the necessary large-scale resources efficiently. Whereas grids were first used for huge number crunching of trivially parallelizable problems, increasingly parallel high-performance computing is required. Here, we show for the prime example of molecular dynamic simulations how the presence of large grid clusters including very fast network interconnects within grid infrastructures allows now parallel high-performance grid computing efficiently and thus combines the benefits of dedicated super-computing centres and grid infrastructures. The demands for this service class are the highest since the user group has very heterogeneous requirements: i) two to many thousands of CPUs, ii) different memory architectures, iii) huge storage capabilities, and iv) fast communication via network interconnects, are all needed in different combinations and must be considered in a highly dedicated manner to reach highest performance efficiency. Beyond, advanced and dedicated i) interaction with users, ii) the management of jobs, iii) accounting, and iv) billing, not only combines classic with parallel high-performance grid usage, but more importantly is also able to increase the efficiency of IT resource providers. Consequently, the mere "yes-we-can" becomes a huge opportunity like e.g. the life-science and health-care sectors as well as grid infrastructures by reaching higher level of resource efficiency.
Parallel computing techniques for rotorcraft aerodynamics
NASA Astrophysics Data System (ADS)
Ekici, Kivanc
The modification of unsteady three-dimensional Navier-Stokes codes for application on massively parallel and distributed computing environments is investigated. The Euler/Navier-Stokes code TURNS (Transonic Unsteady Rotor Navier-Stokes) was chosen as a test bed because of its wide use by universities and industry. For the efficient implementation of TURNS on parallel computing systems, two algorithmic changes are developed. First, main modifications to the implicit operator, Lower-Upper Symmetric Gauss Seidel (LU-SGS) originally used in TURNS, is performed. Second, application of an inexact Newton method, coupled with a Krylov subspace iterative method (Newton-Krylov method) is carried out. Both techniques have been tried previously for the Euler equations mode of the code. In this work, we have extended the methods to the Navier-Stokes mode. Several new implicit operators were tried because of convergence problems of traditional operators with the high cell aspect ratio (CAR) grids needed for viscous calculations on structured grids. Promising results for both Euler and Navier-Stokes cases are presented for these operators. For the efficient implementation of Newton-Krylov methods to the Navier-Stokes mode of TURNS, efficient preconditioners must be used. The parallel implicit operators used in the previous step are employed as preconditioners and the results are compared. The Message Passing Interface (MPI) protocol has been used because of its portability to various parallel architectures. It should be noted that the proposed methodology is general and can be applied to several other CFD codes (e.g. OVERFLOW).
Collectively loading an application in a parallel computer
Aho, Michael E.; Attinella, John E.; Gooding, Thomas M.; Miller, Samuel J.; Mundy, Michael B.
2016-01-05
Collectively loading an application in a parallel computer, the parallel computer comprising a plurality of compute nodes, including: identifying, by a parallel computer control system, a subset of compute nodes in the parallel computer to execute a job; selecting, by the parallel computer control system, one of the subset of compute nodes in the parallel computer as a job leader compute node; retrieving, by the job leader compute node from computer memory, an application for executing the job; and broadcasting, by the job leader to the subset of compute nodes in the parallel computer, the application for executing the job.
Parallel Proximity Detection for Computer Simulations
NASA Technical Reports Server (NTRS)
Steinman, Jeffrey S. (Inventor); Wieland, Frederick P. (Inventor)
1998-01-01
The present invention discloses a system for performing proximity detection in computer simulations on parallel processing architectures utilizing a distribution list which includes movers and sensor coverages which check in and out of grids. Each mover maintains a list of sensors that detect the mover's motion as the mover and sensor coverages check in and out of the grids. Fuzzy grids are included by fuzzy resolution parameters to allow movers and sensor coverages to check in and out of grids without computing exact grid crossings. The movers check in and out of grids while moving sensors periodically inform the grids of their coverage. In addition, a lookahead function is also included for providing a generalized capability without making any limiting assumptions about the particular application to which it is applied. The lookahead function is initiated so that risk-free synchronization strategies never roll back grid events. The lookahead function adds fixed delays as events are scheduled for objects on other nodes.
Parallel Proximity Detection for Computer Simulation
NASA Technical Reports Server (NTRS)
Steinman, Jeffrey S. (Inventor); Wieland, Frederick P. (Inventor)
1997-01-01
The present invention discloses a system for performing proximity detection in computer simulations on parallel processing architectures utilizing a distribution list which includes movers and sensor coverages which check in and out of grids. Each mover maintains a list of sensors that detect the mover's motion as the mover and sensor coverages check in and out of the grids. Fuzzy grids are includes by fuzzy resolution parameters to allow movers and sensor coverages to check in and out of grids without computing exact grid crossings. The movers check in and out of grids while moving sensors periodically inform the grids of their coverage. In addition, a lookahead function is also included for providing a generalized capability without making any limiting assumptions about the particular application to which it is applied. The lookahead function is initiated so that risk-free synchronization strategies never roll back grid events. The lookahead function adds fixed delays as events are scheduled for objects on other nodes.
Interconnections For Stacked Parallel Computer Modules
NASA Technical Reports Server (NTRS)
Johannesson, Richard T.
1996-01-01
Concept for interconnecting modules in parallel computers leads to cheaper, smaller, lighter, lower-power computing systems for aerospace, industrial, business, and consumer applications. Computer modules stacked and interconnected in various configurations. Connections among stacks controlled by switching within gateways and/or by addresses on buses.
Parallel unstructured grid generation for computational aerosciences
NASA Technical Reports Server (NTRS)
Shephard, Mark S.
1993-01-01
The objective of this research project is to develop efficient parallel automatic grid generation procedures for use in computational aerosciences. This effort is focused on a parallel version of the Finite Octree grid generator. Progress made during the first six months is reported.
Massively Parallel Computing: A Sandia Perspective
Dosanjh, Sudip S.; Greenberg, David S.; Hendrickson, Bruce; Heroux, Michael A.; Plimpton, Steve J.; Tomkins, James L.; Womble, David E.
1999-05-06
The computing power available to scientists and engineers has increased dramatically in the past decade, due in part to progress in making massively parallel computing practical and available. The expectation for these machines has been great. The reality is that progress has been slower than expected. Nevertheless, massively parallel computing is beginning to realize its potential for enabling significant break-throughs in science and engineering. This paper provides a perspective on the state of the field, colored by the authors' experiences using large scale parallel machines at Sandia National Laboratories. We address trends in hardware, system software and algorithms, and we also offer our view of the forces shaping the parallel computing industry.
Accurate modeling of parallel scientific computations
NASA Technical Reports Server (NTRS)
Nicol, David M.; Townsend, James C.
1988-01-01
Scientific codes are usually parallelized by partitioning a grid among processors. To achieve top performance it is necessary to partition the grid so as to balance workload and minimize communication/synchronization costs. This problem is particularly acute when the grid is irregular, changes over the course of the computation, and is not known until load time. Critical mapping and remapping decisions rest on the ability to accurately predict performance, given a description of a grid and its partition. This paper discusses one approach to this problem, and illustrates its use on a one-dimensional fluids code. The models constructed are shown to be accurate, and are used to find optimal remapping schedules.
HPC Infrastructure for Solid Earth Simulation on Parallel Computers
NASA Astrophysics Data System (ADS)
Nakajima, K.; Chen, L.; Okuda, H.
2004-12-01
Recently, various types of parallel computers with various types of architectures and processing elements (PE) have emerged, which include PC clusters and the Earth Simulator. Moreover, users can easily access to these computer resources through network on Grid environment. It is well-known that thorough tuning is required for programmers to achieve excellent performance on each computer. The method for tuning strongly depends on the type of PE and architecture. Optimization by tuning is a very tough work, especially for developers of applications. Moreover, parallel programming using message passing library such as MPI is another big task for application programmers. In GeoFEM project (http://gefeom.tokyo.rist.or.jp), authors have developed a parallel FEM platform for solid earth simulation on the Earth Simulator, which supports parallel I/O, parallel linear solvers and parallel visualization. This platform can efficiently hide complicated procedures for parallel programming and optimization on vector processors from application programmers. This type of infrastructure is very useful. Source codes developed on PC with single processor is easily optimized on massively parallel computer by linking the source code to the parallel platform installed on the target computer. This parallel platform, called HPC Infrastructure will provide dramatic efficiency, portability and reliability in development of scientific simulation codes. For example, line number of the source codes is expected to be less than 10,000 and porting legacy codes to parallel computer takes 2 or 3 weeks. Original GeoFEM platform supports only I/O, linear solvers and visualization. In the present work, further development for adaptive mesh refinement (AMR) and dynamic load-balancing (DLB) have been carried out. In this presentation, examples of large-scale solid earth simulation using the Earth Simulator will be demonstrated. Moreover, recent results of a parallel computational steering tool using an
Compute Server Performance Results
NASA Technical Reports Server (NTRS)
Stockdale, I. E.; Barton, John; Woodrow, Thomas (Technical Monitor)
1994-01-01
Parallel-vector supercomputers have been the workhorses of high performance computing. As expectations of future computing needs have risen faster than projected vector supercomputer performance, much work has been done investigating the feasibility of using Massively Parallel Processor systems as supercomputers. An even more recent development is the availability of high performance workstations which have the potential, when clustered together, to replace parallel-vector systems. We present a systematic comparison of floating point performance and price-performance for various compute server systems. A suite of highly vectorized programs was run on systems including traditional vector systems such as the Cray C90, and RISC workstations such as the IBM RS/6000 590 and the SGI R8000. The C90 system delivers 460 million floating point operations per second (FLOPS), the highest single processor rate of any vendor. However, if the price-performance ration (PPR) is considered to be most important, then the IBM and SGI processors are superior to the C90 processors. Even without code tuning, the IBM and SGI PPR's of 260 and 220 FLOPS per dollar exceed the C90 PPR of 160 FLOPS per dollar when running our highly vectorized suite,
Parallel hypergraph partitioning for scientific computing.
Heaphy, Robert; Devine, Karen Dragon; Catalyurek, Umit; Bisseling, Robert; Hendrickson, Bruce Alan; Boman, Erik Gunnar
2005-07-01
Graph partitioning is often used for load balancing in parallel computing, but it is known that hypergraph partitioning has several advantages. First, hypergraphs more accurately model communication volume, and second, they are more expressive and can better represent nonsymmetric problems. Hypergraph partitioning is particularly suited to parallel sparse matrix-vector multiplication, a common kernel in scientific computing. We present a parallel software package for hypergraph (and sparse matrix) partitioning developed at Sandia National Labs. The algorithm is a variation on multilevel partitioning. Our parallel implementation is novel in that it uses a two-dimensional data distribution among processors. We present empirical results that show our parallel implementation achieves good speedup on several large problems (up to 33 million nonzeros) with up to 64 processors on a Linux cluster.
Parallelization of ARC3D with Computer-Aided Tools
NASA Technical Reports Server (NTRS)
Jin, Haoqiang; Hribar, Michelle; Yan, Jerry; Saini, Subhash (Technical Monitor)
1998-01-01
A series of efforts have been devoted to investigating methods of porting and parallelizing applications quickly and efficiently for new architectures, such as the SCSI Origin 2000 and Cray T3E. This report presents the parallelization of a CFD application, ARC3D, using the computer-aided tools, Cesspools. Steps of parallelizing this code and requirements of achieving better performance are discussed. The generated parallel version has achieved reasonably well performance, for example, having a speedup of 30 for 36 Cray T3E processors. However, this performance could not be obtained without modification of the original serial code. It is suggested that in many cases improving serial code and performing necessary code transformations are important parts for the automated parallelization process although user intervention in many of these parts are still necessary. Nevertheless, development and improvement of useful software tools, such as Cesspools, can help trim down many tedious parallelization details and improve the processing efficiency.
Identifying failure in a tree network of a parallel computer
Archer, Charles J.; Pinnow, Kurt W.; Wallenfelt, Brian P.
2010-08-24
Methods, parallel computers, and products are provided for identifying failure in a tree network of a parallel computer. The parallel computer includes one or more processing sets including an I/O node and a plurality of compute nodes. For each processing set embodiments include selecting a set of test compute nodes, the test compute nodes being a subset of the compute nodes of the processing set; measuring the performance of the I/O node of the processing set; measuring the performance of the selected set of test compute nodes; calculating a current test value in dependence upon the measured performance of the I/O node of the processing set, the measured performance of the set of test compute nodes, and a predetermined value for I/O node performance; and comparing the current test value with a predetermined tree performance threshold. If the current test value is below the predetermined tree performance threshold, embodiments include selecting another set of test compute nodes. If the current test value is not below the predetermined tree performance threshold, embodiments include selecting from the test compute nodes one or more potential problem nodes and testing individually potential problem nodes and links to potential problem nodes.
Simple, parallel virtual machines for extreme computations
NASA Astrophysics Data System (ADS)
Chokoufe Nejad, Bijan; Ohl, Thorsten; Reuter, Jürgen
2015-11-01
We introduce a virtual machine (VM) written in a numerically fast language like Fortran or C for evaluating very large expressions. We discuss the general concept of how to perform computations in terms of a VM and present specifically a VM that is able to compute tree-level cross sections for any number of external legs, given the corresponding byte-code from the optimal matrix element generator, O'MEGA. Furthermore, this approach allows to formulate the parallel computation of a single phase space point in a simple and obvious way. We analyze hereby the scaling behavior with multiple threads as well as the benefits and drawbacks that are introduced with this method. Our implementation of a VM can run faster than the corresponding native, compiled code for certain processes and compilers, especially for very high multiplicities, and has in general runtimes in the same order of magnitude. By avoiding the tedious compile and link steps, which may fail for source code files of gigabyte sizes, new processes or complex higher order corrections that are currently out of reach could be evaluated with a VM given enough computing power.
Parallel Performance Characterization of Columbia
NASA Technical Reports Server (NTRS)
Biswas, Rupak
2004-01-01
Using a collection of benchmark problems of increasing levels of realism and computational effort, we will characterize the strengths and limitations of the 10,240 processor Columbia system to deliver supercomputing value to application scientists. Scientists need to be able to determine if and how they can utilize Columbia to carry extreme workloads, either in terms of ultra-large applications that cannot be run otherwise (capability), or in terms of very large ensembles of medium-scale applications to populate response matrices (capacity). We select existing application benchmarks that scale from a small number of processors to the entire machine, and that highlight different issues in running supercomputing-calss applicaions, such as the various types of memory access, file I/O, inter- and intra-node communications and parallelization paradigms. http://www.nas.nasa.gov/Software/NPB/
Mapping Radiosity Computations to Parallel Processors.
NASA Astrophysics Data System (ADS)
Singh, Gautam Bir
The radiosity method for rendering scenes is gaining popularity because of its ability to accurately model the energy distribution in an environment. As this photonic energy distribution is independent of the viewer's position, generating scenes for different viewpoints only requires hidden surface removal and can be performed in real-time. This makes it more attractive than ray tracing as a technique for modeling illumination. It is quite conceivable that radiosity method will be used for applications in scientific visualization, lighting simulations, CAD/CAM, virtual reality, and medical imaging. Computing radiosity of a scene with moderate to high complexity is tantamount to solving a system of tens of thousands of linear equations. Iterative linear system solvers, such as Gauss-Seidel, Jacobi, or conjugate descent, are quite demanding for a system of equations this large. An alternate approach, known as progressive refinement, offers some computational tractability and delivers an approximate solution relatively quickly. This dissertation presents the results of partitioning the radiosity computation to suitably map on a variety of multiprocessor classes. The effect of problem decomposition on computation and communication components is studied for the shared memory, the message passing and the loosely coupled distributed memory multiprocessors. Kendall Square Research's KSR1 and Intel hypercube iPSC/860 were used for experimenting with the shared memory and message-passing algorithms respectively. A network of IBM RS/6000 was used for understanding coarse grain parallelization techniques. These experiments demonstrated that optimality of parallel algorithms must be considered as a < machine, algorithm > pair. Thus the notion of program portability must also take machine architecture in consideration beside allowing for software compatibility. As the number of polygons for processing complex scenes continues to grow, the subdivision in the object space become
A scalable parallel black oil simulator on distributed memory parallel computers
NASA Astrophysics Data System (ADS)
Wang, Kun; Liu, Hui; Chen, Zhangxin
2015-11-01
This paper presents our work on developing a parallel black oil simulator for distributed memory computers based on our in-house parallel platform. The parallel simulator is designed to overcome the performance issues of common simulators that are implemented for personal computers and workstations. The finite difference method is applied to discretize the black oil model. In addition, some advanced techniques are employed to strengthen the robustness and parallel scalability of the simulator, including an inexact Newton method, matrix decoupling methods, and algebraic multigrid methods. A new multi-stage preconditioner is proposed to accelerate the solution of linear systems from the Newton methods. Numerical experiments show that our simulator is scalable and efficient, and is capable of simulating extremely large-scale black oil problems with tens of millions of grid blocks using thousands of MPI processes on parallel computers.
Methods for operating parallel computing systems employing sequenced communications
Benner, Robert E.; Gustafson, John L.; Montry, Gary R.
1999-01-01
A parallel computing system and method having improved performance where a program is concurrently run on a plurality of nodes for reducing total processing time, each node having a processor, a memory, and a predetermined number of communication channels connected to the node and independently connected directly to other nodes. The present invention improves performance of performance of the parallel computing system by providing a system which can provide efficient communication between the processors and between the system and input and output devices. A method is also disclosed which can locate defective nodes with the computing system.
Parallel reservoir computing using optical amplifiers.
Vandoorne, Kristof; Dambre, Joni; Verstraeten, David; Schrauwen, Benjamin; Bienstman, Peter
2011-09-01
Reservoir computing (RC), a computational paradigm inspired on neural systems, has become increasingly popular in recent years for solving a variety of complex recognition and classification problems. Thus far, most implementations have been software-based, limiting their speed and power efficiency. Integrated photonics offers the potential for a fast, power efficient and massively parallel hardware implementation. We have previously proposed a network of coupled semiconductor optical amplifiers as an interesting test case for such a hardware implementation. In this paper, we investigate the important design parameters and the consequences of process variations through simulations. We use an isolated word recognition task with babble noise to evaluate the performance of the photonic reservoirs with respect to traditional software reservoir implementations, which are based on leaky hyperbolic tangent functions. Our results show that the use of coherent light in a well-tuned reservoir architecture offers significant performance benefits. The most important design parameters are the delay and the phase shift in the system's physical connections. With optimized values for these parameters, coherent semiconductor optical amplifier (SOA) reservoirs can achieve better results than traditional simulated reservoirs. We also show that process variations hardly degrade the performance, but amplifier noise can be detrimental. This effect must therefore be taken into account when designing SOA-based RC implementations.
CFD research, parallel computation and aerodynamic optimization
NASA Technical Reports Server (NTRS)
Ryan, James S.
1995-01-01
Over five years of research in Computational Fluid Dynamics and its applications are covered in this report. Using CFD as an established tool, aerodynamic optimization on parallel architectures is explored. The objective of this work is to provide better tools to vehicle designers. Submarine design requires accurate force and moment calculations in flow with thick boundary layers and large separated vortices. Low noise production is critical, so flow into the propulsor region must be predicted accurately. The High Speed Civil Transport (HSCT) has been the subject of recent work. This vehicle is to be a passenger vehicle with the capability of cutting overseas flight times by more than half. A successful design must surpass the performance of comparable planes. Fuel economy, other operational costs, environmental impact, and range must all be improved substantially. For all these reasons, improved design tools are required, and these tools must eventually integrate optimization, external aerodynamics, propulsion, structures, heat transfer and other disciplines.
NASA Astrophysics Data System (ADS)
Tramm, John R.; Gunow, Geoffrey; He, Tim; Smith, Kord S.; Forget, Benoit; Siegel, Andrew R.
2016-05-01
In this study we present and analyze a formulation of the 3D Method of Characteristics (MOC) technique applied to the simulation of full core nuclear reactors. Key features of the algorithm include a task-based parallelism model that allows independent MOC tracks to be assigned to threads dynamically, ensuring load balancing, and a wide vectorizable inner loop that takes advantage of modern SIMD computer architectures. The algorithm is implemented in a set of highly optimized proxy applications in order to investigate its performance characteristics on CPU, GPU, and Intel Xeon Phi architectures. Speed, power, and hardware cost efficiencies are compared. Additionally, performance bottlenecks are identified for each architecture in order to determine the prospects for continued scalability of the algorithm on next generation HPC architectures.
Link failure detection in a parallel computer
Archer, Charles J.; Blocksome, Michael A.; Megerian, Mark G.; Smith, Brian E.
2010-11-09
Methods, apparatus, and products are disclosed for link failure detection in a parallel computer including compute nodes connected in a rectangular mesh network, each pair of adjacent compute nodes in the rectangular mesh network connected together using a pair of links, that includes: assigning each compute node to either a first group or a second group such that adjacent compute nodes in the rectangular mesh network are assigned to different groups; sending, by each of the compute nodes assigned to the first group, a first test message to each adjacent compute node assigned to the second group; determining, by each of the compute nodes assigned to the second group, whether the first test message was received from each adjacent compute node assigned to the first group; and notifying a user, by each of the compute nodes assigned to the second group, whether the first test message was received.
Internode data communications in a parallel computer
Archer, Charles J.; Blocksome, Michael A.; Miller, Douglas R.; Parker, Jeffrey J.; Ratterman, Joseph D.; Smith, Brian E.
2013-09-03
Internode data communications in a parallel computer that includes compute nodes that each include main memory and a messaging unit, the messaging unit including computer memory and coupling compute nodes for data communications, in which, for each compute node at compute node boot time: a messaging unit allocates, in the messaging unit's computer memory, a predefined number of message buffers, each message buffer associated with a process to be initialized on the compute node; receives, prior to initialization of a particular process on the compute node, a data communications message intended for the particular process; and stores the data communications message in the message buffer associated with the particular process. Upon initialization of the particular process, the process establishes a messaging buffer in main memory of the compute node and copies the data communications message from the message buffer of the messaging unit into the message buffer of main memory.
Internode data communications in a parallel computer
Archer, Charles J; Blocksome, Michael A; Miller, Douglas R; Parker, Jeffrey J; Ratterman, Joseph D; Smith, Brian E
2014-02-11
Internode data communications in a parallel computer that includes compute nodes that each include main memory and a messaging unit, the messaging unit including computer memory and coupling compute nodes for data communications, in which, for each compute node at compute node boot time: a messaging unit allocates, in the messaging unit's computer memory, a predefined number of message buffers, each message buffer associated with a process to be initialized on the compute node; receives, prior to initialization of a particular process on the compute node, a data communications message intended for the particular process; and stores the data communications message in the message buffer associated with the particular process. Upon initialization of the particular process, the process establishes a messaging buffer in main memory of the compute node and copies the data communications message from the message buffer of the messaging unit into the message buffer of main memory.
NASA Technical Reports Server (NTRS)
Abineri, Adrian; Chen, Y. P.
1992-01-01
APTEC Computer Systems is a Portland, Oregon based manufacturer of I/O computers. APTEC's work in the context of high density storage media is on programs requiring real-time data capture with low latency processing and storage requirements. An example of APTEC's work in this area is the Loral/Space Telescope-Data Archival and Distribution System. This is an existing Loral AeroSys designed system, which utilizes an APTEC I/O computer. The key attributes of a system architecture that is suitable for this environment are as follows: (1) data acquisition alternatives; (2) a wide range of supported mass storage devices; (3) data processing options; (4) data availability through standard network connections; and (5) an overall system architecture (hardware and software designed for high bandwidth and low latency). APTEC's approach is outlined in this document.
Parallel aeroelastic computations for wing and wing-body configurations
NASA Technical Reports Server (NTRS)
Byun, Chansup
1994-01-01
The objective of this research is to develop computationally efficient methods for solving fluid-structural interaction problems by directly coupling finite difference Euler/Navier-Stokes equations for fluids and finite element dynamics equations for structures on parallel computers. This capability will significantly impact many aerospace projects of national importance such as Advanced Subsonic Civil Transport (ASCT), where the structural stability margin becomes very critical at the transonic region. This research effort will have direct impact on the High Performance Computing and Communication (HPCC) Program of NASA in the area of parallel computing.
PISCES: An environment for parallel scientific computation
NASA Technical Reports Server (NTRS)
Pratt, T. W.
1985-01-01
The parallel implementation of scientific computing environment (PISCES) is a project to provide high-level programming environments for parallel MIMD computers. Pisces 1, the first of these environments, is a FORTRAN 77 based environment which runs under the UNIX operating system. The Pisces 1 user programs in Pisces FORTRAN, an extension of FORTRAN 77 for parallel processing. The major emphasis in the Pisces 1 design is in providing a carefully specified virtual machine that defines the run-time environment within which Pisces FORTRAN programs are executed. Each implementation then provides the same virtual machine, regardless of differences in the underlying architecture. The design is intended to be portable to a variety of architectures. Currently Pisces 1 is implemented on a network of Apollo workstations and on a DEC VAX uniprocessor via simulation of the task level parallelism. An implementation for the Flexible Computing Corp. FLEX/32 is under construction. An introduction to the Pisces 1 virtual computer and the FORTRAN 77 extensions is presented. An example of an algorithm for the iterative solution of a system of equations is given. The most notable features of the design are the provision for several granularities of parallelism in programs and the provision of a window mechanism for distributed access to large arrays of data.
Wing-Body Aeroelasticity on Parallel Computers
NASA Technical Reports Server (NTRS)
Guruswamy, Guru P.; Byun, Chansup
1996-01-01
This article presents a procedure for computing the aeroelasticity of wing-body configurations on multiple-instruction, multiple-data parallel computers. In this procedure, fluids are modeled using Euler equations discretized by a finite difference method, and structures are modeled using finite element equations. The procedure is designed in such a way that each discipline can be developed and maintained independently by using a domain decomposition approach. A parallel integration scheme is used to compute aeroelastic responses by solving the coupled fluid and structural equations concurrently while keeping modularity of each discipline. The present procedure is validated by computing the aeroelastic response of a wing and comparing with experiment. Aeroelastic computations are illustrated for a high speed civil transport type wing-body configuration.
Methods for operating parallel computing systems employing sequenced communications
Benner, R.E.; Gustafson, J.L.; Montry, G.R.
1999-08-10
A parallel computing system and method are disclosed having improved performance where a program is concurrently run on a plurality of nodes for reducing total processing time, each node having a processor, a memory, and a predetermined number of communication channels connected to the node and independently connected directly to other nodes. The present invention improves performance of the parallel computing system by providing a system which can provide efficient communication between the processors and between the system and input and output devices. A method is also disclosed which can locate defective nodes with the computing system. 15 figs.
Chare kernel; A runtime support system for parallel computations
Shu, W. ); Kale, L.V. )
1991-03-01
This paper presents the chare kernel system, which supports parallel computations with irregular structure. The chare kernel is a collection of primitive functions that manage chares, manipulative messages, invoke atomic computations, and coordinate concurrent activities. Programs written in the chare kernel language can be executed on different parallel machines without change. Users writing such programs concern themselves with the creation of parallel actions but not with assigning them to specific processors. The authors describe the design and implementation of the chare kernel. Performance of chare kernel programs on two hypercube machines, the Intel iPSC/2 and the NCUBE, is also given.
Parallel computing using a Lagrangian formulation
NASA Technical Reports Server (NTRS)
Liou, May-Fun; Loh, Ching Yuen
1991-01-01
A new Lagrangian formulation of the Euler equation is adopted for the calculation of 2-D supersonic steady flow. The Lagrangian formulation represents the inherent parallelism of the flow field better than the common Eulerian formulation and offers a competitive alternative on parallel computers. The implementation of the Lagrangian formulation on the Thinking Machines Corporation CM-2 Computer is described. The program uses a finite volume, first-order Godunov scheme and exhibits high accuracy in dealing with multidimensional discontinuities (slip-line and shock). By using this formulation, a better than six times speed-up was achieved on a 8192-processor CM-2 over a single processor of a CRAY-2.
Temporal fringe pattern analysis with parallel computing
Tuck Wah Ng; Kar Tien Ang; Argentini, Gianluca
2005-11-20
Temporal fringe pattern analysis is invaluable in transient phenomena studies but necessitates long processing times. Here we describe a parallel computing strategy based on the single-program multiple-data model and hyperthreading processor technology to reduce the execution time. In a two-node cluster workstation configuration we found that execution periods were reduced by 1.6 times when four virtual processors were used. To allow even lower execution times with an increasing number of processors, the time allocated for data transfer, data read, and waiting should be minimized. Parallel computing is found here to present a feasible approach to reduce execution times in temporal fringe pattern analysis.
Highly parallel computer architecture for robotic computation
NASA Technical Reports Server (NTRS)
Fijany, Amir (Inventor); Bejczy, Anta K. (Inventor)
1991-01-01
In a computer having a large number of single instruction multiple data (SIMD) processors, each of the SIMD processors has two sets of three individual processor elements controlled by a master control unit and interconnected among a plurality of register file units where data is stored. The register files input and output data in synchronism with a minor cycle clock under control of two slave control units controlling the register file units connected to respective ones of the two sets of processor elements. Depending upon which ones of the register file units are enabled to store or transmit data during a particular minor clock cycle, the processor elements within an SIMD processor are connected in rings or in pipeline arrays, and may exchange data with the internal bus or with neighboring SIMD processors through interface units controlled by respective ones of the two slave control units.
Overview of ICLASS research: Reliable and parallel computing
NASA Technical Reports Server (NTRS)
Iyer, Ravi K.
1987-01-01
An overview of Illinois Computer Laboratory for Aerospace Systems and Software (ICLASS) Research: Reliable and Parallel Computing is presented. Topics covered include: reliable and fault tolerant computing; fault tolerant multiprocessor architectures; fault tolerant matrix computation; and parallel processing.
Efficient Parallel Kernel Solvers for Computational Fluid Dynamics Applications
NASA Technical Reports Server (NTRS)
Sun, Xian-He
1997-01-01
Distributed-memory parallel computers dominate today's parallel computing arena. These machines, such as Intel Paragon, IBM SP2, and Cray Origin2OO, have successfully delivered high performance computing power for solving some of the so-called "grand-challenge" problems. Despite initial success, parallel machines have not been widely accepted in production engineering environments due to the complexity of parallel programming. On a parallel computing system, a task has to be partitioned and distributed appropriately among processors to reduce communication cost and to attain load balance. More importantly, even with careful partitioning and mapping, the performance of an algorithm may still be unsatisfactory, since conventional sequential algorithms may be serial in nature and may not be implemented efficiently on parallel machines. In many cases, new algorithms have to be introduced to increase parallel performance. In order to achieve optimal performance, in addition to partitioning and mapping, a careful performance study should be conducted for a given application to find a good algorithm-machine combination. This process, however, is usually painful and elusive. The goal of this project is to design and develop efficient parallel algorithms for highly accurate Computational Fluid Dynamics (CFD) simulations and other engineering applications. The work plan is 1) developing highly accurate parallel numerical algorithms, 2) conduct preliminary testing to verify the effectiveness and potential of these algorithms, 3) incorporate newly developed algorithms into actual simulation packages. The work plan has well achieved. Two highly accurate, efficient Poisson solvers have been developed and tested based on two different approaches: (1) Adopting a mathematical geometry which has a better capacity to describe the fluid, (2) Using compact scheme to gain high order accuracy in numerical discretization. The previously developed Parallel Diagonal Dominant (PDD) algorithm
Archer, Charles J.; Blocksome, Michael A.; Ratterman, Joseph D.; Smith, Brian E.
2014-08-12
Endpoint-based parallel data processing in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints, each endpoint including a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, the compute nodes coupled for data communications through the PAMI, including establishing a data communications geometry, the geometry specifying, for tasks representing processes of execution of the parallel application, a set of endpoints that are used in collective operations of the PAMI including a plurality of endpoints for one of the tasks; receiving in endpoints of the geometry an instruction for a collective operation; and executing the instruction for a collective operation through the endpoints in dependence upon the geometry, including dividing data communications operations among the plurality of endpoints for one of the tasks.
Archer, Charles J; Blocksome, Michael E; Ratterman, Joseph D; Smith, Brian E
2014-02-11
Endpoint-based parallel data processing in a parallel active messaging interface ('PAMI') of a parallel computer, the PAMI composed of data communications endpoints, each endpoint including a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, the compute nodes coupled for data communications through the PAMI, including establishing a data communications geometry, the geometry specifying, for tasks representing processes of execution of the parallel application, a set of endpoints that are used in collective operations of the PAMI including a plurality of endpoints for one of the tasks; receiving in endpoints of the geometry an instruction for a collective operation; and executing the instruction for a collective opeartion through the endpoints in dependence upon the geometry, including dividing data communications operations among the plurality of endpoints for one of the tasks.
Parallel computation of Feynman loop integrals
NASA Astrophysics Data System (ADS)
de Doncker, E.; Yuasa, F.
2012-12-01
The need for large numbers of compute-intensive integrals, arising in quantum field theory perturbation calculations, justifies the parallelization of loop integrals. In earlier work, we devised effective multivariate methods by iterated (repeated) adaptive numerical integration and extrapolation, applicable for some problem classes where standard multivariate integration techniques fail through integrand singularities. In repeated integration, the function evaluations for the outer integral are independent integral computations for the next lower level and can be distributed to threads in a multi-core parallelization. The distribution level determines the number of dimensions in the inner integral below it and thus the granularity of the computation. We discuss a multi-threaded implementation over the OpenMP Application Program Interface.
Parallel Domain Decomposition Preconditioning for Computational Fluid Dynamics
NASA Technical Reports Server (NTRS)
Barth, Timothy J.; Chan, Tony F.; Tang, Wei-Pai; Kutler, Paul (Technical Monitor)
1998-01-01
This viewgraph presentation gives an overview of the parallel domain decomposition preconditioning for computational fluid dynamics. Details are given on some difficult fluid flow problems, stabilized spatial discretizations, and Newton's method for solving the discretized flow equations. Schur complement domain decomposition is described through basic formulation, simplifying strategies (including iterative subdomain and Schur complement solves, matrix element dropping, localized Schur complement computation, and supersparse computations), and performance evaluation.
Rectilinear partitioning of irregular data parallel computations
NASA Technical Reports Server (NTRS)
Nicol, David M.
1991-01-01
New mapping algorithms for domain oriented data-parallel computations, where the workload is distributed irregularly throughout the domain, but exhibits localized communication patterns are described. Researchers consider the problem of partitioning the domain for parallel processing in such a way that the workload on the most heavily loaded processor is minimized, subject to the constraint that the partition be perfectly rectilinear. Rectilinear partitions are useful on architectures that have a fast local mesh network. Discussed here is an improved algorithm for finding the optimal partitioning in one dimension, new algorithms for partitioning in two dimensions, and optimal partitioning in three dimensions. The application of these algorithms to real problems are discussed.
Solving tridiagonal linear systems on the Butterfly parallel computer
Kumar, S.P.
1989-01-01
A parallel block partitioning method to solve a tri-diagonal system of linear equations is adapted to the BBN Butterfly multiprocessor. A performance analysis of the programming experiments on the 32-node Butterfly is presented. An upper bound on the number of processors to achieve the best performance with this method is derived. The computational results verify the theoretical speedup and efficiency results of the parallel algorithm over its serial counterpart. Also included is a study comparing performance runs of the same code on the Butterfly processor with a hardware floating point unit and on one with a software floating point facility. The total parallel time of the given code is considerably reduced by making use of the hardware floating point facility whereas the speedup and efficiency of the parallel program considerably improve on the system with software floating point capability. The achieved results are shown to be within 82% to 90% of the predicted performance.
Parallel and Distributed Computational Fluid Dynamics: Experimental Results and Challenges
NASA Technical Reports Server (NTRS)
Djomehri, Mohammad Jahed; Biswas, R.; VanderWijngaart, R.; Yarrow, M.
2000-01-01
This paper describes several results of parallel and distributed computing using a large scale production flow solver program. A coarse grained parallelization based on clustering of discretization grids combined with partitioning of large grids for load balancing is presented. An assessment is given of its performance on distributed and distributed-shared memory platforms using large scale scientific problems. An experiment with this solver, adapted to a Wide Area Network execution environment is presented. We also give a comparative performance assessment of computation and communication times on both the tightly and loosely-coupled machines.
Numerical simulation of supersonic wake flow with parallel computers
Wong, C.C.; Soetrisno, M.
1995-07-01
Simulating a supersonic wake flow field behind a conical body is a computing intensive task. It requires a large number of computational cells to capture the dominant flow physics and a robust numerical algorithm to obtain a reliable solution. High performance parallel computers with unique distributed processing and data storage capability can provide this need. They have larger computational memory and faster computing time than conventional vector computers. We apply the PINCA Navier-Stokes code to simulate a wind-tunnel supersonic wake experiment on Intel Gamma, Intel Paragon, and IBM SP2 parallel computers. These simulations are performed to study the mean flow in the near wake region of a sharp, 7-degree half-angle, adiabatic cone at Mach number 4.3 and freestream Reynolds number of 40,600. Overall the numerical solutions capture the general features of the hypersonic laminar wake flow and compare favorably with the wind tunnel data. With a refined and clustering grid distribution in the recirculation zone, the calculated location of the rear stagnation point is consistent with the 2D axisymmetric and 3D experiments. In this study, we also demonstrate the importance of having a large local memory capacity within a computer node and the effective utilization of the number of computer nodes to achieve good parallel performance when simulating a complex, large-scale wake flow problem.
Applications of parallel supercomputers: Scientific results and computer science lessons
Fox, G.C.
1989-07-12
Parallel Computing has come of age with several commercial and inhouse systems that deliver supercomputer performance. We illustrate this with several major computations completed or underway at Caltech on hypercubes, transputer arrays and the SIMD Connection Machine CM-2 and AMT DAP. Applications covered are lattice gauge theory, computational fluid dynamics, subatomic string dynamics, statistical and condensed matter physics,theoretical and experimental astronomy, quantum chemistry, plasma physics, grain dynamics, computer chess, graphics ray tracing, and Kalman filters. We use these applications to compare the performance of several advanced architecture computers including the conventional CRAY and ETA-10 supercomputers. We describe which problems are suitable for which computers in the terms of a matching between problem and computer architecture. This is part of a set of lessons we draw for hardware, software, and performance. We speculate on the emergence of new academic disciplines motivated by the growing importance of computers. 138 refs., 23 figs., 10 tabs.
Parallel Computational Fluid Dynamics: Current Status and Future Requirements
NASA Technical Reports Server (NTRS)
Simon, Horst D.; VanDalsem, William R.; Dagum, Leonardo; Kutler, Paul (Technical Monitor)
1994-01-01
One or the key objectives of the Applied Research Branch in the Numerical Aerodynamic Simulation (NAS) Systems Division at NASA Allies Research Center is the accelerated introduction of highly parallel machines into a full operational environment. In this report we discuss the performance results obtained from the implementation of some computational fluid dynamics (CFD) applications on the Connection Machine CM-2 and the Intel iPSC/860. We summarize some of the experiences made so far with the parallel testbed machines at the NAS Applied Research Branch. Then we discuss the long term computational requirements for accomplishing some of the grand challenge problems in computational aerosciences. We argue that only massively parallel machines will be able to meet these grand challenge requirements, and we outline the computer science and algorithm research challenges ahead.
Intranode data communications in a parallel computer
Archer, Charles J; Blocksome, Michael A; Miller, Douglas R; Ratterman, Joseph D; Smith, Brian E
2014-01-07
Intranode data communications in a parallel computer that includes compute nodes configured to execute processes, where the data communications include: allocating, upon initialization of a first process of a computer node, a region of shared memory; establishing, by the first process, a predefined number of message buffers, each message buffer associated with a process to be initialized on the compute node; sending, to a second process on the same compute node, a data communications message without determining whether the second process has been initialized, including storing the data communications message in the message buffer of the second process; and upon initialization of the second process: retrieving, by the second process, a pointer to the second process's message buffer; and retrieving, by the second process from the second process's message buffer in dependence upon the pointer, the data communications message sent by the first process.
Parallel computing in atmospheric chemistry models
Rotman, D.
1996-02-01
Studies of atmospheric chemistry are of high scientific interest, involve computations that are complex and intense, and require enormous amounts of I/O. Current supercomputer computational capabilities are limiting the studies of stratospheric and tropospheric chemistry and will certainly not be able to handle the upcoming coupled chemistry/climate models. To enable such calculations, the authors have developed a computing framework that allows computations on a wide range of computational platforms, including massively parallel machines. Because of the fast paced changes in this field, the modeling framework and scientific modules have been developed to be highly portable and efficient. Here, the authors present the important features of the framework and focus on the atmospheric chemistry module, named IMPACT, and its capabilities. Applications of IMPACT to aircraft studies will be presented.
Intranode data communications in a parallel computer
Archer, Charles J; Blocksome, Michael A; Miller, Douglas R; Ratterman, Joseph D; Smith, Brian E
2013-07-23
Intranode data communications in a parallel computer that includes compute nodes configured to execute processes, where the data communications include: allocating, upon initialization of a first process of a compute node, a region of shared memory; establishing, by the first process, a predefined number of message buffers, each message buffer associated with a process to be initialized on the compute node; sending, to a second process on the same compute node, a data communications message without determining whether the second process has been initialized, including storing the data communications message in the message buffer of the second process; and upon initialization of the second process: retrieving, by the second process, a pointer to the second process's message buffer; and retrieving, by the second process from the second process's message buffer in dependence upon the pointer, the data communications message sent by the first process.
A compositional reservoir simulator on distributed memory parallel computers
Rame, M.; Delshad, M.
1995-12-31
This paper presents the application of distributed memory parallel computes to field scale reservoir simulations using a parallel version of UTCHEM, The University of Texas Chemical Flooding Simulator. The model is a general purpose highly vectorized chemical compositional simulator that can simulate a wide range of displacement processes at both field and laboratory scales. The original simulator was modified to run on both distributed memory parallel machines (Intel iPSC/960 and Delta, Connection Machine 5, Kendall Square 1 and 2, and CRAY T3D) and a cluster of workstations. A domain decomposition approach has been taken towards parallelization of the code. A portion of the discrete reservoir model is assigned to each processor by a set-up routine that attempts a data layout as even as possible from the load-balance standpoint. Each of these subdomains is extended so that data can be shared between adjacent processors for stencil computation. The added routines that make parallel execution possible are written in a modular fashion that makes the porting to new parallel platforms straight forward. Results of the distributed memory computing performance of Parallel simulator are presented for field scale applications such as tracer flood and polymer flood. A comparison of the wall-clock times for same problems on a vector supercomputer is also presented.
Problem solving on computers with parallel organization of computations
Glushkov, V.M.; Molchanov, I.N.
1981-07-01
In principle, parallelism can be implemented on different levels: the level of construction of physical models, objects or processes representing the phenomenon being studied, the level of the solution method, the level of the algorithm, the level of the program, the level of data exchange in the computer, and the level of data input and output. The authors consider some of the available parallelism techniques. 9 references.
Implementation and performance of parallelized elegant.
Wang, Y.; Borland, M.; Accelerator Systems Division
2008-01-01
The program elegant is widely used for design and modeling of linacs for free-electron lasers and energy recovery linacs, as well as storage rings and other applications. As part of a multi-year effort, we have parallelized many aspects of the code, including single-particle dynamics, wakefields, and coherent synchrotron radiation. We report on the approach used for gradual parallelization, which proved very beneficial in getting parallel features into the hands of users quickly. We also report details of parallelization of collective effects. Finally, we discuss performance of the parallelized code in various applications.
Method for implementation of recursive hierarchical segmentation on parallel computers
NASA Technical Reports Server (NTRS)
Tilton, James C. (Inventor)
2005-01-01
A method, computer readable storage, and apparatus for implementing a recursive hierarchical segmentation algorithm on a parallel computing platform. The method includes setting a bottom level of recursion that defines where a recursive division of an image into sections stops dividing, and setting an intermediate level of recursion where the recursive division changes from a parallel implementation into a serial implementation. The segmentation algorithm is implemented according to the set levels. The method can also include setting a convergence check level of recursion with which the first level of recursion communicates with when performing a convergence check.
Parallel Computing Environments and Methods for Power Distribution System Simulation
Lu, Ning; Taylor, Zachary T.; Chassin, David P.; Guttromson, Ross T.; Studham, Scott S.
2005-11-10
The development of cost-effective high-performance parallel computing on multi-processor super computers makes it attractive to port excessively time consuming simulation software from personal computers (PC) to super computes. The power distribution system simulator (PDSS) takes a bottom-up approach and simulates load at appliance level, where detailed thermal models for appliances are used. This approach works well for a small power distribution system consisting of a few thousand appliances. When the number of appliances increases, the simulation uses up the PC memory and its run time increases to a point where the approach is no longer feasible to model a practical large power distribution system. This paper presents an effort made to port a PC-based power distribution system simulator (PDSS) to a 128-processor shared-memory super computer. The paper offers an overview of the parallel computing environment and a description of the modification made to the PDSS model. The performances of the PDSS running on a standalone PC and on the super computer are compared. Future research direction of utilizing parallel computing in the power distribution system simulation is also addressed.
Solving unstructured grid problems on massively parallel computers
NASA Technical Reports Server (NTRS)
Hammond, Steven W.; Schreiber, Robert
1990-01-01
A highly parallel graph mapping technique that enables one to efficiently solve unstructured grid problems on massively parallel computers is presented. Many implicit and explicit methods for solving discretized partial differential equations require each point in the discretization to exchange data with its neighboring points every time step or iteration. The cost of this communication can negate the high performance promised by massively parallel computing. To eliminate this bottleneck, the graph of the irregular problem is mapped into the graph representing the interconnection topology of the computer such that the sum of the distances that the messages travel is minimized. It is shown that using the heuristic mapping algorithm significantly reduces the communication time compared to a naive assignment of processes to processors.
CFD Analysis and Design Optimization Using Parallel Computers
NASA Technical Reports Server (NTRS)
Martinelli, Luigi; Alonso, Juan Jose; Jameson, Antony; Reuther, James
1997-01-01
A versatile and efficient multi-block method is presented for the simulation of both steady and unsteady flow, as well as aerodynamic design optimization of complete aircraft configurations. The compressible Euler and Reynolds Averaged Navier-Stokes (RANS) equations are discretized using a high resolution scheme on body-fitted structured meshes. An efficient multigrid implicit scheme is implemented for time-accurate flow calculations. Optimum aerodynamic shape design is achieved at very low cost using an adjoint formulation. The method is implemented on parallel computing systems using the MPI message passing interface standard to ensure portability. The results demonstrate that, by combining highly efficient algorithms with parallel computing, it is possible to perform detailed steady and unsteady analysis as well as automatic design for complex configurations using the present generation of parallel computers.
Model of parallel performance. Final report, July-December 1988
Carmona, E.A.; Rice, M.D.
1989-04-01
This report introduces a general model of parallel performance. With the goal of developing conceptual and empirical methods for characterizing and understanding parallel algorithms, new definitions of speedup and efficiency were formulated. These definitions take into account the effects that problem size and the number of processors have on efficiency and speedup, and provide a natural and quantifiable measure of parallel performance. The terms introduced in the definitions provide new and improved interpretations of the 'serial' and 'parallel' fraction parameters commonly used in the literature (i.e., Amdahl's Law) to explain the behavior of parallel algorithms. The model provides a more complete characterization of parallel algorithm behavior and is used to correct apparent deficiencies in the formulation of speedup as expressed by Amdahl's Law. Chapter 2 of this report reviews the basic definitions of speedup and efficiency and introduces new definitions of these quantities in terms of work units and presents two illustrative examples. Chapter 3 discusses two models of parallel performance that appear in the parallel computing literature. The two formulations of speedup and efficiency are based on the notion of 'serial' and 'parallel' fractions and present an apparent dichotomy that must be resolved. Then, the 'serial' and 'parallel' fractions are expressed in terms of the new work quantities introduced in Chapter 2, providing a natural interpretation of the serial and parallel fractions of a task. The resulting equations are used to reconcile the (apparent) dichotomy between Amdahl's Law and the more recent formulation of speedup.
Parallelization of Finite Element Analysis Codes Using Heterogeneous Distributed Computing
NASA Technical Reports Server (NTRS)
Ozguner, Fusun
1996-01-01
Performance gains in computer design are quickly consumed as users seek to analyze larger problems to a higher degree of accuracy. Innovative computational methods, such as parallel and distributed computing, seek to multiply the power of existing hardware technology to satisfy the computational demands of large applications. In the early stages of this project, experiments were performed using two large, coarse-grained applications, CSTEM and METCAN. These applications were parallelized on an Intel iPSC/860 hypercube. It was found that the overall speedup was very low, due to large, inherently sequential code segments present in the applications. The overall execution time T(sub par), of the application is dependent on these sequential segments. If these segments make up a significant fraction of the overall code, the application will have a poor speedup measure.
Parallelism for quantum computation with qudits
O'Leary, Dianne P.; Brennen, Gavin K.; Bullock, Stephen S.
2006-09-15
Robust quantum computation with d-level quantum systems (qudits) poses two requirements: fast, parallel quantum gates and high-fidelity two-qudit gates. We first describe how to implement parallel single-qudit operations. It is by now well known that any single-qudit unitary can be decomposed into a sequence of Givens rotations on two-dimensional subspaces of the qudit state space. Using a coupling graph to represent physically allowed couplings between pairs of qudit states, we then show that the logical depth (time) of the parallel gate sequence is equal to the height of an associated tree. The implementation of a given unitary can then optimize the tradeoff between gate time and resources used. These ideas are illustrated for qudits encoded in the ground hyperfine states of the alkali-metal atoms {sup 87}Rb and {sup 133}Cs. Second, we provide a protocol for implementing parallelized nonlocal two-qudit gates using the assistance of entangled qubit pairs. Using known protocols for qubit entanglement purification, this offers the possibility of high-fidelity two-qudit gates.
Parallel algorithms for computing linked list prefix
Han, Y. )
1989-06-01
Given a linked list chi/sub 1/, chi/sub 2/, ....chi/sub n/ with chi/sub i/ following chi/sub i-1/ in the list and an associative operation O, the linked list prefix problem is to compute all prefixes O/sup j//sub i=1/chi/sub 1/, j=1,2,...,n. In this paper the authors study the linked list prefix problem on parallel computation models. A deterministic algorithm for computing a linked list prefix on a completely connected parallel computation model is obtained by applying vector balancing techniques. The time complexity of the algorithm is O(n/rho + rho log rho), where n is the number of elements in the linked list and rho is the number of processors used. Therefore their algorithm is optimal when n {ge}rho/sup 2/logrho. A PRAM linked list prefix algorithm is also presented. This PRAM algorithm has time complexity O(n/rho + log rho) with small multiplicative constant. It is optimal when n {ge}rho log rho.
National Combustion Code: Parallel Implementation and Performance
NASA Technical Reports Server (NTRS)
Quealy, A.; Ryder, R.; Norris, A.; Liu, N.-S.
2000-01-01
The National Combustion Code (NCC) is being developed by an industry-government team for the design and analysis of combustion systems. CORSAIR-CCD is the current baseline reacting flow solver for NCC. This is a parallel, unstructured grid code which uses a distributed memory, message passing model for its parallel implementation. The focus of the present effort has been to improve the performance of the NCC flow solver to meet combustor designer requirements for model accuracy and analysis turnaround time. Improving the performance of this code contributes significantly to the overall reduction in time and cost of the combustor design cycle. This paper describes the parallel implementation of the NCC flow solver and summarizes its current parallel performance on an SGI Origin 2000. Earlier parallel performance results on an IBM SP-2 are also included. The performance improvements which have enabled a turnaround of less than 15 hours for a 1.3 million element fully reacting combustion simulation are described.
A design methodology for portable software on parallel computers
NASA Technical Reports Server (NTRS)
Nicol, David M.; Miller, Keith W.; Chrisman, Dan A.
1993-01-01
This final report for research that was supported by grant number NAG-1-995 documents our progress in addressing two difficulties in parallel programming. The first difficulty is developing software that will execute quickly on a parallel computer. The second difficulty is transporting software between dissimilar parallel computers. In general, we expect that more hardware-specific information will be included in software designs for parallel computers than in designs for sequential computers. This inclusion is an instance of portability being sacrificed for high performance. New parallel computers are being introduced frequently. Trying to keep one's software on the current high performance hardware, a software developer almost continually faces yet another expensive software transportation. The problem of the proposed research is to create a design methodology that helps designers to more precisely control both portability and hardware-specific programming details. The proposed research emphasizes programming for scientific applications. We completed our study of the parallelizability of a subsystem of the NASA Earth Radiation Budget Experiment (ERBE) data processing system. This work is summarized in section two. A more detailed description is provided in Appendix A ('Programming Practices to Support Eventual Parallelism'). Mr. Chrisman, a graduate student, wrote and successfully defended a Ph.D. dissertation proposal which describes our research associated with the issues of software portability and high performance. The list of research tasks are specified in the proposal. The proposal 'A Design Methodology for Portable Software on Parallel Computers' is summarized in section three and is provided in its entirety in Appendix B. We are currently studying a proposed subsystem of the NASA Clouds and the Earth's Radiant Energy System (CERES) data processing system. This software is the proof-of-concept for the Ph.D. dissertation. We have implemented and measured
Element-topology-independent preconditioners for parallel finite element computations
NASA Technical Reports Server (NTRS)
Park, K. C.; Alexander, Scott
1992-01-01
A family of preconditioners for the solution of finite element equations are presented, which are element-topology independent and thus can be applicable to element order-free parallel computations. A key feature of the present preconditioners is the repeated use of element connectivity matrices and their left and right inverses. The properties and performance of the present preconditioners are demonstrated via beam and two-dimensional finite element matrices for implicit time integration computations.
Efficient Helicopter Aerodynamic and Aeroacoustic Predictions on Parallel Computers
NASA Technical Reports Server (NTRS)
Wissink, Andrew M.; Lyrintzis, Anastasios S.; Strawn, Roger C.; Oliker, Leonid; Biswas, Rupak
1996-01-01
This paper presents parallel implementations of two codes used in a combined CFD/Kirchhoff methodology to predict the aerodynamics and aeroacoustics properties of helicopters. The rotorcraft Navier-Stokes code, TURNS, computes the aerodynamic flowfield near the helicopter blades and the Kirchhoff acoustics code computes the noise in the far field, using the TURNS solution as input. The overall parallel strategy adds MPI message passing calls to the existing serial codes to allow for communication between processors. As a result, the total code modifications required for parallel execution are relatively small. The biggest bottleneck in running the TURNS code in parallel comes from the LU-SGS algorithm that solves the implicit system of equations. We use a new hybrid domain decomposition implementation of LU-SGS to obtain good parallel performance on the SP-2. TURNS demonstrates excellent parallel speedups for quasi-steady and unsteady three-dimensional calculations of a helicopter blade in forward flight. The execution rate attained by the code on 114 processors is six times faster than the same cases run on one processor of the Cray C-90. The parallel Kirchhoff code also shows excellent parallel speedups and fast execution rates. As a performance demonstration, unsteady acoustic pressures are computed at 1886 far-field observer locations for a sample acoustics problem. The calculation requires over two hundred hours of CPU time on one C-90 processor but takes only a few hours on 80 processors of the SP2. The resultant far-field acoustic field is analyzed with state of-the-art audio and video rendering of the propagating acoustic signals.
Parallel Analysis and Visualization on Cray Compute Node Linux
Pugmire, Dave; Ahern, Sean
2008-01-01
Capability computer systems are deployed to give researchers the computational power required to investigate and solve key challenges facing the scientific community. As the power of these computer systems increases, the computational problem domain typically increases in size, complexity and scope. These increases strain the ability of commodity analysis and visualization clusters to effectively perform post-processing tasks and provide critical insight and understanding to the computed results. An alternative to purchasing increasingly larger, separate analysis and visualization commodity clusters is to use the computational system itself to perform post-processing tasks. In this paper, the recent successful port of VisIt, a parallel, open source analysis and visualization tool, to compute node linux running on the Cray is detailed. Additionally, the unprecedented ability of this resource for analysis and visualization is discussed and a report on obtained results is presented.
Toward an automated parallel computing environment for geosciences
NASA Astrophysics Data System (ADS)
Zhang, Huai; Liu, Mian; Shi, Yaolin; Yuen, David A.; Yan, Zhenzhen; Liang, Guoping
2007-08-01
Software for geodynamic modeling has not kept up with the fast growing computing hardware and network resources. In the past decade supercomputing power has become available to most researchers in the form of affordable Beowulf clusters and other parallel computer platforms. However, to take full advantage of such computing power requires developing parallel algorithms and associated software, a task that is often too daunting for geoscience modelers whose main expertise is in geosciences. We introduce here an automated parallel computing environment built on open-source algorithms and libraries. Users interact with this computing environment by specifying the partial differential equations, solvers, and model-specific properties using an English-like modeling language in the input files. The system then automatically generates the finite element codes that can be run on distributed or shared memory parallel machines. This system is dynamic and flexible, allowing users to address different problems in geosciences. It is capable of providing web-based services, enabling users to generate source codes online. This unique feature will facilitate high-performance computing to be integrated with distributed data grids in the emerging cyber-infrastructures for geosciences. In this paper we discuss the principles of this automated modeling environment and provide examples to demonstrate its versatility.
Computational fluid dynamics on a massively parallel computer
NASA Technical Reports Server (NTRS)
Jespersen, Dennis C.; Levit, Creon
1989-01-01
A finite difference code was implemented for the compressible Navier-Stokes equations on the Connection Machine, a massively parallel computer. The code is based on the ARC2D/ARC3D program and uses the implicit factored algorithm of Beam and Warming. The codes uses odd-even elimination to solve linear systems. Timings and computation rates are given for the code, and a comparison is made with a Cray XMP.
Parallel Computing in Information Retrieval--An Updated Review.
ERIC Educational Resources Information Center
Macfarlane, A.; And Others
1997-01-01
Reviews the progress of parallel computing in information retrieval (IR) and stresses the importance of the motivation in using parallel computing for text retrieval. Analyzes parallel IR systems using a classification defined by Rasmussen; describes retrieval models used in parallel information processing; and suggests areas of needed research.…
Lattice gauge theory on the Intel parallel scientific computer
NASA Astrophysics Data System (ADS)
Gottlieb, Steven
1990-08-01
Intel Scientific Computers (ISC) has just started producing its third general of parallel computer, the iPSC/860. Based on the i860 chip that has a peak performance of 80 Mflops and with a current maximum of 128 nodes, this computer should achieve speeds in excess of those obtainable on conventional vector supercomputers. The hardware, software and computing techniques appropriate for lattice gauge theory calculations are described. The differences between a staggered fermion conjugate gradient program written under CANOPY and for the iPSC are detailed.
Hybrid parallel computing architecture for multiview phase shifting
NASA Astrophysics Data System (ADS)
Zhong, Kai; Li, Zhongwei; Zhou, Xiaohui; Shi, Yusheng; Wang, Congjun
2014-11-01
The multiview phase-shifting method shows its powerful capability in achieving high resolution three-dimensional (3-D) shape measurement. Unfortunately, this ability results in very high computation costs and 3-D computations have to be processed offline. To realize real-time 3-D shape measurement, a hybrid parallel computing architecture is proposed for multiview phase shifting. In this architecture, the central processing unit can co-operate with the graphic processing unit (GPU) to achieve hybrid parallel computing. The high computation cost procedures, including lens distortion rectification, phase computation, correspondence, and 3-D reconstruction, are implemented in GPU, and a three-layer kernel function model is designed to simultaneously realize coarse-grained and fine-grained paralleling computing. Experimental results verify that the developed system can perform 50 fps (frame per second) real-time 3-D measurement with 260 K 3-D points per frame. A speedup of up to 180 times is obtained for the performance of the proposed technique using a NVIDIA GT560Ti graphics card rather than a sequential C in a 3.4 GHZ Inter Core i7 3770.
IPython: components for interactive and parallel computing across disciplines. (Invited)
NASA Astrophysics Data System (ADS)
Perez, F.; Bussonnier, M.; Frederic, J. D.; Froehle, B. M.; Granger, B. E.; Ivanov, P.; Kluyver, T.; Patterson, E.; Ragan-Kelley, B.; Sailer, Z.
2013-12-01
Scientific computing is an inherently exploratory activity that requires constantly cycling between code, data and results, each time adjusting the computations as new insights and questions arise. To support such a workflow, good interactive environments are critical. The IPython project (http://ipython.org) provides a rich architecture for interactive computing with: 1. Terminal-based and graphical interactive consoles. 2. A web-based Notebook system with support for code, text, mathematical expressions, inline plots and other rich media. 3. Easy to use, high performance tools for parallel computing. Despite its roots in Python, the IPython architecture is designed in a language-agnostic way to facilitate interactive computing in any language. This allows users to mix Python with Julia, R, Octave, Ruby, Perl, Bash and more, as well as to develop native clients in other languages that reuse the IPython clients. In this talk, I will show how IPython supports all stages in the lifecycle of a scientific idea: 1. Individual exploration. 2. Collaborative development. 3. Production runs with parallel resources. 4. Publication. 5. Education. In particular, the IPython Notebook provides an environment for "literate computing" with a tight integration of narrative and computation (including parallel computing). These Notebooks are stored in a JSON-based document format that provides an "executable paper": notebooks can be version controlled, exported to HTML or PDF for publication, and used for teaching.
Advances in parallel computer technology for desktop atmospheric dispersion models
Bian, X.; Ionescu-Niscov, S.; Fast, J.D.; Allwine, K.J.
1996-12-31
Desktop models are those models used by analysts with varied backgrounds, for performing, for example, air quality assessment and emergency response activities. These models must be robust, well documented, have minimal and well controlled user inputs, and have clear outputs. Existing coarse-grained parallel computers can provide significant increases in computation speed in desktop atmospheric dispersion modeling without considerable increases in hardware cost. This increased speed will allow for significant improvements to be made in the scientific foundations of these applied models, in the form of more advanced diffusion schemes and better representation of the wind and turbulence fields. This is especially attractive for emergency response applications where speed and accuracy are of utmost importance. This paper describes one particular application of coarse-grained parallel computer technology to a desktop complex terrain atmospheric dispersion modeling system. By comparing performance characteristics of the coarse-grained parallel version of the model with the single-processor version, we will demonstrate that applying coarse-grained parallel computer technology to desktop atmospheric dispersion modeling systems will allow us to address critical issues facing future requirements of this class of dispersion models.
Variable-Complexity Multidisciplinary Optimization on Parallel Computers
NASA Technical Reports Server (NTRS)
Grossman, Bernard; Mason, William H.; Watson, Layne T.; Haftka, Raphael T.
1998-01-01
This report covers work conducted under grant NAG1-1562 for the NASA High Performance Computing and Communications Program (HPCCP) from December 7, 1993, to December 31, 1997. The objective of the research was to develop new multidisciplinary design optimization (MDO) techniques which exploit parallel computing to reduce the computational burden of aircraft MDO. The design of the High-Speed Civil Transport (HSCT) air-craft was selected as a test case to demonstrate the utility of our MDO methods. The three major tasks of this research grant included: development of parallel multipoint approximation methods for the aerodynamic design of the HSCT, use of parallel multipoint approximation methods for structural optimization of the HSCT, mathematical and algorithmic development including support in the integration of parallel computation for items (1) and (2). These tasks have been accomplished with the development of a response surface methodology that incorporates multi-fidelity models. For the aerodynamic design we were able to optimize with up to 20 design variables using hundreds of expensive Euler analyses together with thousands of inexpensive linear theory simulations. We have thereby demonstrated the application of CFD to a large aerodynamic design problem. For the predicting structural weight we were able to combine hundreds of structural optimizations of refined finite element models with thousands of optimizations based on coarse models. Computations have been carried out on the Intel Paragon with up to 128 nodes. The parallel computation allowed us to perform combined aerodynamic-structural optimization using state of the art models of a complex aircraft configurations.
CFD Research, Parallel Computation and Aerodynamic Optimization
NASA Technical Reports Server (NTRS)
Ryan, James S.
1995-01-01
During the last five years, CFD has matured substantially. Pure CFD research remains to be done, but much of the focus has shifted to integration of CFD into the design process. The work under these cooperative agreements reflects this trend. The recent work, and work which is planned, is designed to enhance the competitiveness of the US aerospace industry. CFD and optimization approaches are being developed and tested, so that the industry can better choose which methods to adopt in their design processes. The range of computer architectures has been dramatically broadened, as the assumption that only huge vector supercomputers could be useful has faded. Today, researchers and industry can trade off time, cost, and availability, choosing vector supercomputers, scalable parallel architectures, networked workstations, or heterogenous combinations of these to complete required computations efficiently.
Optimized data communications in a parallel computer
Faraj, Daniel A.
2014-08-19
A parallel computer includes nodes that include a network adapter that couples the node in a point-to-point network and supports communications in opposite directions of each dimension. Optimized communications include: receiving, by a network adapter of a receiving compute node, a packet--from a source direction--that specifies a destination node and deposit hints. Each hint is associated with a direction within which the packet is to be deposited. If a hint indicates the packet to be deposited in the opposite direction: the adapter delivers the packet to an application on the receiving node; forwards the packet to a next node in the opposite direction if the receiving node is not the destination; and forwards the packet to a node in a direction of a subsequent dimension if the hints indicate that the packet is to be deposited in the direction of the subsequent dimension.
Optimized data communications in a parallel computer
Faraj, Daniel A
2014-10-21
A parallel computer includes nodes that include a network adapter that couples the node in a point-to-point network and supports communications in opposite directions of each dimension. Optimized communications include: receiving, by a network adapter of a receiving compute node, a packet--from a source direction--that specifies a destination node and deposit hints. Each hint is associated with a direction within which the packet is to be deposited. If a hint indicates the packet to be deposited in the opposite direction: the adapter delivers the packet to an application on the receiving node; forwards the packet to a next node in the opposite direction if the receiving node is not the destination; and forwards the packet to a node in a direction of a subsequent dimension if the hints indicate that the packet is to be deposited in the direction of the subsequent dimension.
Pipeline and parallel architectures for computer communication systems
Reddi, A.V.
1983-01-01
Various existing communication precessor systems (CPSS) at different nodes in computer communication systems (CCSS) are reviewed for distributed processing systems. To meet the increasing load of messages, pipeline and parallel architectures are suggested in CPSS. Finally, pipeline, array, multi and multiple-processor architectures and their advantages in CPSS for CCSS are presented and analysed, and their performances are compared with the performance of uniprocessor architecture. 19 references.
NASA Astrophysics Data System (ADS)
Mighell, Kenneth John
2010-10-01
The development of parallel-processing image-analysis codes is generally a challenging task that requires complicated choreography of interprocessor communications. If, however, the image-analysis algorithm is embarrassingly parallel, then the development of a parallel-processing implementation of that algorithm can be a much easier task to accomplish because, by definition, there is little need for communication between the compute processes. I describe the design, implementation, and performance of a parallel-processing image-analysis application, called crblaster, which does cosmic-ray rejection of CCD images using the embarrassingly parallel l.a.cosmic algorithm. crblaster is written in C using the high-performance computing industry standard Message Passing Interface (MPI) library. crblaster uses a two-dimensional image partitioning algorithm that partitions an input image into N rectangular subimages of nearly equal area; the subimages include sufficient additional pixels along common image partition edges such that the need for communication between computer processes is eliminated. The code has been designed to be used by research scientists who are familiar with C as a parallel-processing computational framework that enables the easy development of parallel-processing image-analysis programs based on embarrassingly parallel algorithms. The crblaster source code is freely available at the official application Web site at the National Optical Astronomy Observatory. Removing cosmic rays from a single 800 × 800 pixel Hubble Space Telescope WFPC2 image takes 44 s with the IRAF script lacos_im.cl running on a single core of an Apple Mac Pro computer with two 2.8 GHz quad-core Intel Xeon processors. crblaster is 7.4 times faster when processing the same image on a single core on the same machine. Processing the same image with crblaster simultaneously on all eight cores of the same machine takes 0.875 s—which is a speedup factor of 50.3 times faster than the
Computationally efficient implementation of combustion chemistry in parallel PDF calculations
NASA Astrophysics Data System (ADS)
Lu, Liuyan; Lantz, Steven R.; Ren, Zhuyin; Pope, Stephen B.
2009-08-01
In parallel calculations of combustion processes with realistic chemistry, the serial in situ adaptive tabulation (ISAT) algorithm [S.B. Pope, Computationally efficient implementation of combustion chemistry using in situ adaptive tabulation, Combustion Theory and Modelling, 1 (1997) 41-63; L. Lu, S.B. Pope, An improved algorithm for in situ adaptive tabulation, Journal of Computational Physics 228 (2009) 361-386] substantially speeds up the chemistry calculations on each processor. To improve the parallel efficiency of large ensembles of such calculations in parallel computations, in this work, the ISAT algorithm is extended to the multi-processor environment, with the aim of minimizing the wall clock time required for the whole ensemble. Parallel ISAT strategies are developed by combining the existing serial ISAT algorithm with different distribution strategies, namely purely local processing (PLP), uniformly random distribution (URAN), and preferential distribution (PREF). The distribution strategies enable the queued load redistribution of chemistry calculations among processors using message passing. They are implemented in the software x2f_mpi, which is a Fortran 95 library for facilitating many parallel evaluations of a general vector function. The relative performance of the parallel ISAT strategies is investigated in different computational regimes via the PDF calculations of multiple partially stirred reactors burning methane/air mixtures. The results show that the performance of ISAT with a fixed distribution strategy strongly depends on certain computational regimes, based on how much memory is available and how much overlap exists between tabulated information on different processors. No one fixed strategy consistently achieves good performance in all the regimes. Therefore, an adaptive distribution strategy, which blends PLP, URAN and PREF, is devised and implemented. It yields consistently good performance in all regimes. In the adaptive parallel
Analysis of multigrid methods on massively parallel computers: Architectural implications
NASA Technical Reports Server (NTRS)
Matheson, Lesley R.; Tarjan, Robert E.
1993-01-01
We study the potential performance of multigrid algorithms running on massively parallel computers with the intent of discovering whether presently envisioned machines will provide an efficient platform for such algorithms. We consider the domain parallel version of the standard V cycle algorithm on model problems, discretized using finite difference techniques in two and three dimensions on block structured grids of size 10(exp 6) and 10(exp 9), respectively. Our models of parallel computation were developed to reflect the computing characteristics of the current generation of massively parallel multicomputers. These models are based on an interconnection network of 256 to 16,384 message passing, 'workstation size' processors executing in an SPMD mode. The first model accomplishes interprocessor communications through a multistage permutation network. The communication cost is a logarithmic function which is similar to the costs in a variety of different topologies. The second model allows single stage communication costs only. Both models were designed with information provided by machine developers and utilize implementation derived parameters. With the medium grain parallelism of the current generation and the high fixed cost of an interprocessor communication, our analysis suggests an efficient implementation requires the machine to support the efficient transmission of long messages, (up to 1000 words) or the high initiation cost of a communication must be significantly reduced through an alternative optimization technique. Furthermore, with variable length message capability, our analysis suggests the low diameter multistage networks provide little or no advantage over a simple single stage communications network.
Implementation and performance of parallel Prolog interpreter
Wei, S.; Kale, L.V.; Balkrishna, R. . Dept. of Computer Science)
1988-01-01
In this paper, the authors discuss the implementation of a parallel Prolog interpreter on different parallel machines. The implementation is based on the REDUCE--OR process model which exploits both AND and OR parallelism in logic programs. It is machine independent as it runs on top of the chare-kernel--a machine-independent parallel programming system. The authors also give the performance of the interpreter running a diverse set of benchmark pargrams on parallel machines including shared memory systems: an Alliant FX/8, Sequent and a MultiMax, and a non-shared memory systems: Intel iPSC/32 hypercube, in addition to its performance on a multiprocessor simulation system.
Parallel distance matrix computation for Matlab data mining
NASA Astrophysics Data System (ADS)
Skurowski, Przemysław; Staniszewski, Michał
2016-06-01
The paper presents utility functions for computing of a distance matrix, which plays a crucial role in data mining. The goal in the design was to enable operating on relatively large datasets by overcoming basic shortcoming - computing time - with an interface easy to use. The presented solution is a set of functions, which were created with emphasis on practical applicability in real life. The proposed solution is presented along the theoretical background for the performance scaling. Furthermore, different approaches of the parallel computing are analyzed, including shared memory, which is uncommon in Matlab environment.
Parallel multithread computing for spectroscopic analysis in optical coherence tomography
NASA Astrophysics Data System (ADS)
Trojanowski, Michal; Kraszewski, Maciej; Strakowski, Marcin; Pluciński, Jerzy
2014-05-01
Spectroscopic Optical Coherence Tomography (SOCT) is an extension of Optical Coherence Tomography (OCT). It allows gathering spectroscopic information from individual scattering points inside the sample. It is based on time-frequency analysis of interferometric signals. Such analysis requires calculating hundreds of Fourier transforms while performing a single A-scan. Additionally, further processing of acquired spectroscopic information is needed. This significantly increases the time of required computations. During last years, application of graphical processing units (GPU's) was proposed to reduce computation time in OCT by using parallel computing algorithms. GPU technology can be also used to speed-up signal processing in SOCT. However, parallel algorithms used in classical OCT need to be revised because of different character of analyzed data. The classical OCT requires processing of long, independent interferometric signals for obtaining subsequent A-scans. The difference with SOCT is that it requires processing of multiple, shorter signals, which differ only in a small part of samples. We have developed new algorithms for parallel signal processing for usage in SOCT, implemented with NVIDIA CUDA (Compute Unified Device Architecture). We present details of the algorithms and performance tests for analyzing data from in-house SD-OCT system. We also give a brief discussion about usefulness of developed algorithm. Presented algorithms might be useful for researchers working on OCT, as they allow to reduce computation time and are step toward real-time signal processing of SOCT data.
Performance analysis of parallel supernodal sparse LU factorization
Grigori, Laura; Li, Xiaoye S.
2004-02-05
We investigate performance characteristics for the LU factorization of large matrices with various sparsity patterns. We consider supernodal right-looking parallel factorization on a bi-dimensional grid of processors, making use of static pivoting. We develop a performance model and we validate it using the implementation in SuperLU-DIST, the real matrices and the IBM Power3 machine at NERSC. We use this model to obtain performance bounds on parallel computers, to perform scalability analysis and to identify performance bottlenecks. We also discuss the role of load balance and data distribution in this approach.
New parallel vision environment in heterogeneous networked computing
NASA Astrophysics Data System (ADS)
Kim, Kyung N.; Choi, Tae-Sun; Ramakrishna, R. S.
1998-09-01
Many vision tasks are very complex and computationally intensive. Real time requirements further aggravate the situation. They usually involve both structured (low-level vision) and unstructured (high-level vision) computations. Parallel approaches offer hope in this context. Parallel approaches to vision tasks and scheduling schemes for their implementation receive special emphasis in this paper. Architectural issues are also addressed. The aim is to design algorithms which can be implemented on low cost heterogeneous networks running PVM. Issues connected with general purpose architectures also receive attention. The proposed ideas have been illustrated through a practical example (of eye location from an image sequence). Next generation multimedia environments are expected to routinely employ such high performance computing platforms.
QCMPI: A parallel environment for quantum computing
NASA Astrophysics Data System (ADS)
Tabakin, Frank; Juliá-Díaz, Bruno
2009-06-01
QCMPI is a quantum computer (QC) simulation package written in Fortran 90 with parallel processing capabilities. It is an accessible research tool that permits rapid evaluation of quantum algorithms for a large number of qubits and for various "noise" scenarios. The prime motivation for developing QCMPI is to facilitate numerical examination of not only how QC algorithms work, but also to include noise, decoherence, and attenuation effects and to evaluate the efficacy of error correction schemes. The present work builds on an earlier Mathematica code QDENSITY, which is mainly a pedagogic tool. In that earlier work, although the density matrix formulation was featured, the description using state vectors was also provided. In QCMPI, the stress is on state vectors, in order to employ a large number of qubits. The parallel processing feature is implemented by using the Message-Passing Interface (MPI) protocol. A description of how to spread the wave function components over many processors is provided, along with how to efficiently describe the action of general one- and two-qubit operators on these state vectors. These operators include the standard Pauli, Hadamard, CNOT and CPHASE gates and also Quantum Fourier transformation. These operators make up the actions needed in QC. Codes for Grover's search and Shor's factoring algorithms are provided as examples. A major feature of this work is that concurrent versions of the algorithms can be evaluated with each version subject to alternate noise effects, which corresponds to the idea of solving a stochastic Schrödinger equation. The density matrix for the ensemble of such noise cases is constructed using parallel distribution methods to evaluate its eigenvalues and associated entropy. Potential applications of this powerful tool include studies of the stability and correction of QC processes using Hamiltonian based dynamics. Program summaryProgram title: QCMPI Catalogue identifier: AECS_v1_0 Program summary URL
Serial multiplier arrays for parallel computation
NASA Technical Reports Server (NTRS)
Winters, Kel
1990-01-01
Arrays of systolic serial-parallel multiplier elements are proposed as an alternative to conventional SIMD mesh serial adder arrays for applications that are multiplication intensive and require few stored operands. The design and operation of a number of multiplier and array configurations featuring locality of connection, modularity, and regularity of structure are discussed. A design methodology combining top-down and bottom-up techniques is described to facilitate development of custom high-performance CMOS multiplier element arrays as well as rapid synthesis of simulation models and semicustom prototype CMOS components. Finally, a differential version of NORA dynamic circuits requiring a single-phase uncomplemented clock signal introduced for this application.
Design of a massively parallel computer using bit serial processing elements
NASA Technical Reports Server (NTRS)
Aburdene, Maurice F.; Khouri, Kamal S.; Piatt, Jason E.; Zheng, Jianqing
1995-01-01
A 1-bit serial processor designed for a parallel computer architecture is described. This processor is used to develop a massively parallel computational engine, with a single instruction-multiple data (SIMD) architecture. The computer is simulated and tested to verify its operation and to measure its performance for further development.
Broadcasting a message in a parallel computer
Archer, Charles J; Faraj, Ahmad A
2013-04-16
Methods, systems, and products are disclosed for broadcasting a message in a parallel computer that includes: transmitting, by the logical root to all of the nodes directly connected to the logical root, a message; and for each node except the logical root: receiving the message; if that node is the physical root, then transmitting the message to all of the child nodes except the child node from which the message was received; if that node received the message from a parent node and if that node is not a leaf node, then transmitting the message to all of the child nodes; and if that node received the message from a child node and if that node is not the physical root, then transmitting the message to all of the child nodes except the child node from which the message was received and transmitting the message to the parent node.
Broadcasting a message in a parallel computer
Archer, Charles J; Faraj, Daniel A
2014-11-18
Methods, systems, and products are disclosed for broadcasting a message in a parallel computer that includes: transmitting, by the logical root to all of the nodes directly connected to the logical root, a message; and for each node except the logical root: receiving the message; if that node is the physical root, then transmitting the message to all of the child nodes except the child node from which the message was received; if that node received the message from a parent node and if that node is not a leaf node, then transmitting the message to all of the child nodes; and if that node received the message from a child node and if that node is not the physical root, then transmitting the message to all of the child nodes except the child node from which the message was received and transmitting the message to the parent node.
Broadcasting collective operation contributions throughout a parallel computer
Faraj, Ahmad
2012-02-21
Methods, systems, and products are disclosed for broadcasting collective operation contributions throughout a parallel computer. The parallel computer includes a plurality of compute nodes connected together through a data communications network. Each compute node has a plurality of processors for use in collective parallel operations on the parallel computer. Broadcasting collective operation contributions throughout a parallel computer according to embodiments of the present invention includes: transmitting, by each processor on each compute node, that processor's collective operation contribution to the other processors on that compute node using intra-node communications; and transmitting on a designated network link, by each processor on each compute node according to a serial processor transmission sequence, that processor's collective operation contribution to the other processors on the other compute nodes using inter-node communications.
Parallel computing in genomic research: advances and applications.
Ocaña, Kary; de Oliveira, Daniel
2015-01-01
Today's genomic experiments have to process the so-called "biological big data" that is now reaching the size of Terabytes and Petabytes. To process this huge amount of data, scientists may require weeks or months if they use their own workstations. Parallelism techniques and high-performance computing (HPC) environments can be applied for reducing the total processing time and to ease the management, treatment, and analyses of this data. However, running bioinformatics experiments in HPC environments such as clouds, grids, clusters, and graphics processing unit requires the expertise from scientists to integrate computational, biological, and mathematical techniques and technologies. Several solutions have already been proposed to allow scientists for processing their genomic experiments using HPC capabilities and parallelism techniques. This article brings a systematic review of literature that surveys the most recently published research involving genomics and parallel computing. Our objective is to gather the main characteristics, benefits, and challenges that can be considered by scientists when running their genomic experiments to benefit from parallelism techniques and HPC capabilities.
Parallel computing in genomic research: advances and applications.
Ocaña, Kary; de Oliveira, Daniel
2015-01-01
Today's genomic experiments have to process the so-called "biological big data" that is now reaching the size of Terabytes and Petabytes. To process this huge amount of data, scientists may require weeks or months if they use their own workstations. Parallelism techniques and high-performance computing (HPC) environments can be applied for reducing the total processing time and to ease the management, treatment, and analyses of this data. However, running bioinformatics experiments in HPC environments such as clouds, grids, clusters, and graphics processing unit requires the expertise from scientists to integrate computational, biological, and mathematical techniques and technologies. Several solutions have already been proposed to allow scientists for processing their genomic experiments using HPC capabilities and parallelism techniques. This article brings a systematic review of literature that surveys the most recently published research involving genomics and parallel computing. Our objective is to gather the main characteristics, benefits, and challenges that can be considered by scientists when running their genomic experiments to benefit from parallelism techniques and HPC capabilities. PMID:26604801
Parallel computing in genomic research: advances and applications
Ocaña, Kary; de Oliveira, Daniel
2015-01-01
Today’s genomic experiments have to process the so-called “biological big data” that is now reaching the size of Terabytes and Petabytes. To process this huge amount of data, scientists may require weeks or months if they use their own workstations. Parallelism techniques and high-performance computing (HPC) environments can be applied for reducing the total processing time and to ease the management, treatment, and analyses of this data. However, running bioinformatics experiments in HPC environments such as clouds, grids, clusters, and graphics processing unit requires the expertise from scientists to integrate computational, biological, and mathematical techniques and technologies. Several solutions have already been proposed to allow scientists for processing their genomic experiments using HPC capabilities and parallelism techniques. This article brings a systematic review of literature that surveys the most recently published research involving genomics and parallel computing. Our objective is to gather the main characteristics, benefits, and challenges that can be considered by scientists when running their genomic experiments to benefit from parallelism techniques and HPC capabilities. PMID:26604801
Semi-coarsening multigrid methods for parallel computing
Jones, J.E.
1996-12-31
Standard multigrid methods are not well suited for problems with anisotropic coefficients which can occur, for example, on grids that are stretched to resolve a boundary layer. There are several different modifications of the standard multigrid algorithm that yield efficient methods for anisotropic problems. In the paper, we investigate the parallel performance of these multigrid algorithms. Multigrid algorithms which work well for anisotropic problems are based on line relaxation and/or semi-coarsening. In semi-coarsening multigrid algorithms a grid is coarsened in only one of the coordinate directions unlike standard or full-coarsening multigrid algorithms where a grid is coarsened in each of the coordinate directions. When both semi-coarsening and line relaxation are used, the resulting multigrid algorithm is robust and automatic in that it requires no knowledge of the nature of the anisotropy. This is the basic multigrid algorithm whose parallel performance we investigate in the paper. The algorithm is currently being implemented on an IBM SP2 and its performance is being analyzed. In addition to looking at the parallel performance of the basic semi-coarsening algorithm, we present algorithmic modifications with potentially better parallel efficiency. One modification reduces the amount of computational work done in relaxation at the expense of using multiple coarse grids. This modification is also being implemented with the aim of comparing its performance to that of the basic semi-coarsening algorithm.
Parallel solvers for reservoir simulation on MIMD computers
Piault, E.; Willien, F.; Roux, F.X.
1995-12-01
We have investigated parallel solvers for reservoir simulation. We compare different solvers and preconditioners using T3D and SP1 parallel computers. We use block diagonal domain decomposition preconditioner with non-overlapping sub-domains.
Performance of the Galley Parallel File System
NASA Technical Reports Server (NTRS)
Nieuwejaar, Nils; Kotz, David
1996-01-01
As the input/output (I/O) needs of parallel scientific applications increase, file systems for multiprocessors are being designed to provide applications with parallel access to multiple disks. Many parallel file systems present applications with a conventional Unix-like interface that allows the application to access multiple disks transparently. This interface conceals the parallism within the file system, which increases the ease of programmability, but makes it difficult or impossible for sophisticated programmers and libraries to use knowledge about their I/O needs to exploit that parallelism. Furthermore, most current parallel file systems are optimized for a different workload than they are being asked to support. We introduce Galley, a new parallel file system that is intended to efficiently support realistic parallel workloads. Initial experiments, reported in this paper, indicate that Galley is capable of providing high-performance 1/O to applications the applications that rely on them. In Section 3 we describe that access data in patterns that have been observed to be common.
SIAM Conference on Parallel Processing for Scientific Computing - March 12-14, 2008
2008-09-08
The themes of the 2008 conference included, but were not limited to: Programming languages, models, and compilation techniques; The transition to ubiquitous multicore/manycore processors; Scientific computing on special-purpose processors (Cell, GPUs, etc.); Architecture-aware algorithms; From scalable algorithms to scalable software; Tools for software development and performance evaluation; Global perspectives on HPC; Parallel computing in industry; Distributed/grid computing; Fault tolerance; Parallel visualization and large scale data management; and The future of parallel architectures.
Parallel CE/SE Computations via Domain Decomposition
NASA Technical Reports Server (NTRS)
Himansu, Ananda; Jorgenson, Philip C. E.; Wang, Xiao-Yen; Chang, Sin-Chung
2000-01-01
This paper describes the parallelization strategy and achieved parallel efficiency of an explicit time-marching algorithm for solving conservation laws. The Space-Time Conservation Element and Solution Element (CE/SE) algorithm for solving the 2D and 3D Euler equations is parallelized with the aid of domain decomposition. The parallel efficiency of the resultant algorithm on a Silicon Graphics Origin 2000 parallel computer is checked.
NASA Astrophysics Data System (ADS)
Shimizu, Futoshi; Kimizuka, Hajime; Kaburaki, Hideo
2002-08-01
A new parallel computing environment, called as ``Parallel Molecular Dynamics Stencil'', has been developed to carry out a large-scale short-range molecular dynamics simulation of solids. The stencil is written in C language using MPI for parallelization and designed successfully to separate and conceal parts of the programs describing cutoff schemes and parallel algorithms for data communication. This has been made possible by introducing the concept of image atoms. Therefore, only a sequential programming of the force calculation routine is required for executing the stencil in parallel environment. Typical molecular dynamics routines, such as various ensembles, time integration methods, and empirical potentials, have been implemented in the stencil. In the presentation, the performance of the stencil on parallel computers of Hitachi, IBM, SGI, and PC-cluster using the models of Lennard-Jones and the EAM type potentials for fracture problem will be reported.
Local rollback for fault-tolerance in parallel computing systems
Blumrich, Matthias A.; Chen, Dong; Gara, Alan; Giampapa, Mark E.; Heidelberger, Philip; Ohmacht, Martin; Steinmacher-Burow, Burkhard; Sugavanam, Krishnan
2012-01-24
A control logic device performs a local rollback in a parallel super computing system. The super computing system includes at least one cache memory device. The control logic device determines a local rollback interval. The control logic device runs at least one instruction in the local rollback interval. The control logic device evaluates whether an unrecoverable condition occurs while running the at least one instruction during the local rollback interval. The control logic device checks whether an error occurs during the local rollback. The control logic device restarts the local rollback interval if the error occurs and the unrecoverable condition does not occur during the local rollback interval.
PArallel Reacting Multiphase FLOw Computational Fluid Dynamic Analysis
2002-06-01
PARMFLO is a parallel multiphase reacting flow computational fluid dynamics (CFD) code. It can perform steady or unsteady simulations in three space dimensions. It is intended for use in engineering CFD analysis of industrial flow system components. Its parallel processing capabilities allow it to be applied to problems that use at least an order of magnitude more computational cells than the number that can be used on a typical single processor workstation (about 106 cellsmore » in parallel processing mode versus about io cells in serial processing mode). Alternately, by spreading the work of a CFD problem that could be run on a single workstation over a group of computers on a network, it can bring the runtime down by an order of magnitude or more (typically from many days to less than one day). The software was implemented using the industry standard Message-Passing Interface (MPI) and domain decomposition in one spatial direction. The phases of a flow problem may include an ideal gas mixture with an arbitrary number of chemical species, and dispersed droplet and particle phases. Regions of porous media may also be included within the domain. The porous media may be packed beds, foams, or monolith catalyst supports. With these features, the code is especially suited to analysis of mixing of reactants in the inlet chamber of catalytic reactors coupled to computation of product yields that result from the flow of the mixture through the catalyst coaled support structure.« less
A Simple Physical Optics Algorithm Perfect for Parallel Computing Architecture
NASA Technical Reports Server (NTRS)
Imbriale, W. A.; Cwik, T.
1994-01-01
A reflector antenna computer program based upon a simple discreet approximation of the radiation integral has proven to be extremely easy to adapt to the parallel computing architecture of the modest number of large-gain computing elements such as are used in the Intel iPSC and Touchstone Delta parallel machines.
A microeconomic scheduler for parallel computers
NASA Technical Reports Server (NTRS)
Stoica, Ion; Abdel-Wahab, Hussein; Pothen, Alex
1995-01-01
We describe a scheduler based on the microeconomic paradigm for scheduling on-line a set of parallel jobs in a multiprocessor system. In addition to the classical objectives of increasing the system throughput and reducing the response time, we consider fairness in allocating system resources among the users, and providing the user with control over the relative performances of his jobs. We associate with every user a savings account in which he receives money at a constant rate. When a user wants to run a job, he creates an expense account for that job to which he transfers money from his savings account. The job uses the funds in its expense account to obtain the system resources it needs for execution. The share of the system resources allocated to the user is directly related to the rate at which the user receives money; the rate at which the user transfers money into a job expense account controls the job's performance. We prove that starvation is not possible in our model. Simulation results show that our scheduler improves both system and user performances in comparison with two different variable partitioning policies. It is also shown to be effective in guaranteeing fairness and providing control over the performance of jobs.
TSE computers - A means for massively parallel computations
NASA Technical Reports Server (NTRS)
Strong, J. P., III
1976-01-01
A description is presented of hardware concepts for building a massively parallel processing system for two-dimensional data. The processing system is to use logic arrays of 128 x 128 elements which perform over 16 thousand operations simultaneously. Attention is given to image data, logic arrays, basic image logic functions, a prototype negator, an interleaver device, image logic circuits, and an image memory circuit.
Use Computer-Aided Tools to Parallelize Large CFD Applications
NASA Technical Reports Server (NTRS)
Jin, H.; Frumkin, M.; Yan, J.
2000-01-01
Porting applications to high performance parallel computers is always a challenging task. It is time consuming and costly. With rapid progressing in hardware architectures and increasing complexity of real applications in recent years, the problem becomes even more sever. Today, scalability and high performance are mostly involving handwritten parallel programs using message-passing libraries (e.g. MPI). However, this process is very difficult and often error-prone. The recent reemergence of shared memory parallel (SMP) architectures, such as the cache coherent Non-Uniform Memory Access (ccNUMA) architecture used in the SGI Origin 2000, show good prospects for scaling beyond hundreds of processors. Programming on an SMP is simplified by working in a globally accessible address space. The user can supply compiler directives, such as OpenMP, to parallelize the code. As an industry standard for portable implementation of parallel programs for SMPs, OpenMP is a set of compiler directives and callable runtime library routines that extend Fortran, C and C++ to express shared memory parallelism. It promises an incremental path for parallel conversion of existing software, as well as scalability and performance for a complete rewrite or an entirely new development. Perhaps the main disadvantage of programming with directives is that inserted directives may not necessarily enhance performance. In the worst cases, it can create erroneous results. While vendors have provided tools to perform error-checking and profiling, automation in directive insertion is very limited and often failed on large programs, primarily due to the lack of a thorough enough data dependence analysis. To overcome the deficiency, we have developed a toolkit, CAPO, to automatically insert OpenMP directives in Fortran programs and apply certain degrees of optimization. CAPO is aimed at taking advantage of detailed inter-procedural dependence analysis provided by CAPTools, developed by the University of
Parallel computation of three-dimensional nonlinear magnetostatic problems.
Levine, D.; Gropp, W.; Forsman, K.; Kettunen, L.; Mathematics and Computer Science; Tampere Univ. of Tech.
1999-02-01
We describe a general-purpose parallel electromagnetic code for computing accurate solutions to large computationally demanding, 3D, nonlinear magnetostatic problems. The code, CORAL, is based on a volume integral equation formulation. Using an IBM SP parallel computer and iterative solution methods, we successfully solved the dense linear systems inherent in such formulations. A key component of our work was the use of the PETSc library, which provides parallel portability and access to the latest linear algebra solution technology.
Rapid indirect trajectory optimization on highly parallel computing architectures
NASA Astrophysics Data System (ADS)
Antony, Thomas
Trajectory optimization is a field which can benefit greatly from the advantages offered by parallel computing. The current state-of-the-art in trajectory optimization focuses on the use of direct optimization methods, such as the pseudo-spectral method. These methods are favored due to their ease of implementation and large convergence regions while indirect methods have largely been ignored in the literature in the past decade except for specific applications in astrodynamics. It has been shown that the shortcomings conventionally associated with indirect methods can be overcome by the use of a continuation method in which complex trajectory solutions are obtained by solving a sequence of progressively difficult optimization problems. High performance computing hardware is trending towards more parallel architectures as opposed to powerful single-core processors. Graphics Processing Units (GPU), which were originally developed for 3D graphics rendering have gained popularity in the past decade as high-performance, programmable parallel processors. The Compute Unified Device Architecture (CUDA) framework, a parallel computing architecture and programming model developed by NVIDIA, is one of the most widely used platforms in GPU computing. GPUs have been applied to a wide range of fields that require the solution of complex, computationally demanding problems. A GPU-accelerated indirect trajectory optimization methodology which uses the multiple shooting method and continuation is developed using the CUDA platform. The various algorithmic optimizations used to exploit the parallelism inherent in the indirect shooting method are described. The resulting rapid optimal control framework enables the construction of high quality optimal trajectories that satisfy problem-specific constraints and fully satisfy the necessary conditions of optimality. The benefits of the framework are highlighted by construction of maximum terminal velocity trajectories for a hypothetical
Implementation of ADI: Schemes on MIMD parallel computers
NASA Technical Reports Server (NTRS)
Vanderwijngaart, Rob F.
1993-01-01
In order to simulate the effects of the impingement of hot exhaust jets of High Performance Aircraft on landing surfaces a multi-disciplinary computation coupling flow dynamics to heat conduction in the runway needs to be carried out. Such simulations, which are essentially unsteady, require very large computational power in order to be completed within a reasonable time frame of the order of an hour. Such power can be furnished by the latest generation of massively parallel computers. These remove the bottleneck of ever more congested data paths to one or a few highly specialized central processing units (CPU's) by having many off-the-shelf CPU's work independently on their own data, and exchange information only when needed. During the past year the first phase of this project was completed, in which the optimal strategy for mapping an ADI-algorithm for the three dimensional unsteady heat equation to a MIMD parallel computer was identified. This was done by implementing and comparing three different domain decomposition techniques that define the tasks for the CPU's in the parallel machine. These implementations were done for a Cartesian grid and Dirichlet boundary conditions. The most promising technique was then used to implement the heat equation solver on a general curvilinear grid with a suite of nontrivial boundary conditions. Finally, this technique was also used to implement the Scalar Penta-diagonal (SP) benchmark, which was taken from the NAS Parallel Benchmarks report. All implementations were done in the programming language C on the Intel iPSC/860 computer.
Numerical computation on massively parallel hypercubes. [Connection machine
McBryan, O.A.
1986-01-01
We describe numerical computations on the Connection Machine, a massively parallel hypercube architecture with 65,536 single-bit processors and 32 Mbytes of memory. A parallel extension of COMMON LISP, provides access to the processors and network. The rich software environment is further enhanced by a powerful virtual processor capability, which extends the degree of fine-grained parallelism beyond 1,000,000. We briefly describe the hardware and indicate the principal features of the parallel programming environment. We then present implementations of SOR, multigrid and pre-conditioned conjugate gradient algorithms for solving partial differential equations on the Connection Machine. Despite the lack of floating point hardware, computation rates above 100 megaflops have been achieved in PDE solution. Virtual processors prove to be a real advantage, easing the effort of software development while improving system performance significantly. The software development effort is also facilitated by the fact that hypercube communications prove to be fast and essentially independent of distance. 29 refs., 4 figs.
CFD Optimization on Network-Based Parallel Computer System
NASA Technical Reports Server (NTRS)
Cheung, Samson H.; VanDalsem, William (Technical Monitor)
1994-01-01
Combining multiple engineering workstations into a network-based heterogeneous parallel computer allows application of aerodynamic optimization with advance computational fluid dynamics codes, which is computationally expensive in mainframe supercomputer. This paper introduces a nonlinear quasi-Newton optimizer designed for this network-based heterogeneous parallel computer on a software called Parallel Virtual Machine. This paper will introduce the methodology behind coupling a Parabolized Navier-Stokes flow solver to the nonlinear optimizer. This parallel optimization package has been applied to reduce the wave drag of a body of revolution and a wing/body configuration with results of 5% to 6% drag reduction.
Parallel CFD design on network-based computer
NASA Technical Reports Server (NTRS)
Cheung, Samson
1995-01-01
Combining multiple engineering workstations into a network-based heterogeneous parallel computer allows application of aerodynamic optimization with advanced computational fluid dynamics codes, which can be computationally expensive on mainframe supercomputers. This paper introduces a nonlinear quasi-Newton optimizer designed for this network-based heterogeneous parallel computing environment utilizing a software called Parallel Virtual Machine. This paper will introduce the methodology behind coupling a Parabolized Navier-Stokes flow solver to the nonlinear optimizer. This parallel optimization package is applied to reduce the wave drag of a body of revolution and a wing/body configuration with results of 5% to 6% drag reduction.
National Combustion Code Parallel Performance Enhancements
NASA Technical Reports Server (NTRS)
Quealy, Angela; Benyo, Theresa (Technical Monitor)
2002-01-01
The National Combustion Code (NCC) is being developed by an industry-government team for the design and analysis of combustion systems. The unstructured grid, reacting flow code uses a distributed memory, message passing model for its parallel implementation. The focus of the present effort has been to improve the performance of the NCC code to meet combustor designer requirements for model accuracy and analysis turnaround time. Improving the performance of this code contributes significantly to the overall reduction in time and cost of the combustor design cycle. This report describes recent parallel processing modifications to NCC that have improved the parallel scalability of the code, enabling a two hour turnaround for a 1.3 million element fully reacting combustion simulation on an SGI Origin 2000.
Performance Characteristics of the Multi-Zone NAS Parallel Benchmarks
NASA Technical Reports Server (NTRS)
Jin, Haoqiang; VanderWijngaart, Rob F.
2003-01-01
We describe a new suite of computational benchmarks that models applications featuring multiple levels of parallelism. Such parallelism is often available in realistic flow computations on systems of grids, but had not previously been captured in bench-marks. The new suite, named NPB Multi-Zone, is extended from the NAS Parallel Benchmarks suite, and involves solving the application benchmarks LU, BT and SP on collections of loosely coupled discretization meshes. The solutions on the meshes are updated independently, but after each time step they exchange boundary value information. This strategy provides relatively easily exploitable coarse-grain parallelism between meshes. Three reference implementations are available: one serial, one hybrid using the Message Passing Interface (MPI) and OpenMP, and another hybrid using a shared memory multi-level programming model (SMP+OpenMP). We examine the effectiveness of hybrid parallelization paradigms in these implementations on three different parallel computers. We also use an empirical formula to investigate the performance characteristics of the multi-zone benchmarks.
Final Report: Super Instruction Architecture for Scalable Parallel Computations
Sanders, Beverly Ann; Bartlett, Rodney; Deumens, Erik
2013-12-23
The most advanced methods for reliable and accurate computation of the electronic structure of molecular and nano systems are the coupled-cluster techniques. These high-accuracy methods help us to understand, for example, how biological enzymes operate and contribute to the design of new organic explosives. The ACES III software provides a modern, high-performance implementation of these methods optimized for high performance parallel computer systems, ranging from small clusters typical in individual research groups, through larger clusters available in campus and regional computer centers, all the way to high-end petascale systems at national labs, including exploiting GPUs if available. This project enhanced the ACESIII software package and used it to study interesting scientific problems.
Massively parallel computing on an organic molecular layer
NASA Astrophysics Data System (ADS)
Bandyopadhyay, Anirban; Pati, Ranjit; Sahu, Satyajit; Peper, Ferdinand; Fujita, Daisuke
2010-05-01
Modern computers operate at enormous speeds-capable of executing in excess of 1013 instructions per second-but their sequential approach to processing, by which logical operations are performed one after another, has remained unchanged since the 1950s. In contrast, although individual neurons of the human brain fire at around just 103 times per second, the simultaneous collective action of millions of neurons enables them to complete certain tasks more efficiently than even the fastest supercomputer. Here we demonstrate an assembly of molecular switches that simultaneously interact to perform a variety of computational tasks including conventional digital logic, calculating Voronoi diagrams, and simulating natural phenomena such as heat diffusion and cancer growth. As well as representing a conceptual shift from serial-processing with static architectures, our parallel, dynamically reconfigurable approach could provide a means to solve otherwise intractable computational problems.
Center for Programming Models for Scalable Parallel Computing
John Mellor-Crummey
2008-02-29
Rice University's achievements as part of the Center for Programming Models for Scalable Parallel Computing include: (1) design and implemention of cafc, the first multi-platform CAF compiler for distributed and shared-memory machines, (2) performance studies of the efficiency of programs written using the CAF and UPC programming models, (3) a novel technique to analyze explicitly-parallel SPMD programs that facilitates optimization, (4) design, implementation, and evaluation of new language features for CAF, including communication topologies, multi-version variables, and distributed multithreading to simplify development of high-performance codes in CAF, and (5) a synchronization strength reduction transformation for automatically replacing barrier-based synchronization with more efficient point-to-point synchronization. The prototype Co-array Fortran compiler cafc developed in this project is available as open source software from http://www.hipersoft.rice.edu/caf.
Bustamam, Alhadi; Burrage, Kevin; Hamilton, Nicholas A
2012-01-01
Markov clustering (MCL) is becoming a key algorithm within bioinformatics for determining clusters in networks. However,with increasing vast amount of data on biological networks, performance and scalability issues are becoming a critical limiting factor in applications. Meanwhile, GPU computing, which uses CUDA tool for implementing a massively parallel computing environment in the GPU card, is becoming a very powerful, efficient, and low-cost option to achieve substantial performance gains over CPU approaches. The use of on-chip memory on the GPU is efficiently lowering the latency time, thus, circumventing a major issue in other parallel computing environments, such as MPI. We introduce a very fast Markov clustering algorithm using CUDA (CUDA-MCL) to perform parallel sparse matrix-matrix computations and parallel sparse Markov matrix normalizations, which are at the heart of MCL. We utilized ELLPACK-R sparse format to allow the effective and fine-grain massively parallel processing to cope with the sparse nature of interaction networks data sets in bioinformatics applications. As the results show, CUDA-MCL is significantly faster than the original MCL running on CPU. Thus, large-scale parallel computation on off-the-shelf desktop-machines, that were previously only possible on supercomputing architectures, can significantly change the way bioinformaticians and biologists deal with their data.
Parallel Computation of the Regional Ocean Modeling System (ROMS)
Wang, P; Song, Y T; Chao, Y; Zhang, H
2005-04-05
The Regional Ocean Modeling System (ROMS) is a regional ocean general circulation modeling system solving the free surface, hydrostatic, primitive equations over varying topography. It is free software distributed world-wide for studying both complex coastal ocean problems and the basin-to-global scale ocean circulation. The original ROMS code could only be run on shared-memory systems. With the increasing need to simulate larger model domains with finer resolutions and on a variety of computer platforms, there is a need in the ocean-modeling community to have a ROMS code that can be run on any parallel computer ranging from 10 to hundreds of processors. Recently, we have explored parallelization for ROMS using the MPI programming model. In this paper, an efficient parallelization strategy for such a large-scale scientific software package, based on an existing shared-memory computing model, is presented. In addition, scientific applications and data-performance issues on a couple of SGI systems, including Columbia, the world's third-fastest supercomputer, are discussed.
Computer Science Techniques Applied to Parallel Atomistic Simulation
NASA Astrophysics Data System (ADS)
Nakano, Aiichiro
1998-03-01
Recent developments in parallel processing technology and multiresolution numerical algorithms have established large-scale molecular dynamics (MD) simulations as a new research mode for studying materials phenomena such as fracture. However, this requires large system sizes and long simulated times. We have developed: i) Space-time multiresolution schemes; ii) fuzzy-clustering approach to hierarchical dynamics; iii) wavelet-based adaptive curvilinear-coordinate load balancing; iv) multilevel preconditioned conjugate gradient method; and v) spacefilling-curve-based data compression for parallel I/O. Using these techniques, million-atom parallel MD simulations are performed for the oxidation dynamics of nanocrystalline Al. The simulations take into account the effect of dynamic charge transfer between Al and O using the electronegativity equalization scheme. The resulting long-range Coulomb interaction is calculated efficiently with the fast multipole method. Results for temperature and charge distributions, residual stresses, bond lengths and bond angles, and diffusivities of Al and O will be presented. The oxidation of nanocrystalline Al is elucidated through immersive visualization in virtual environments. A unique dual-degree education program at Louisiana State University will also be discussed in which students can obtain a Ph.D. in Physics & Astronomy and a M.S. from the Department of Computer Science in five years. This program fosters interdisciplinary research activities for interfacing High Performance Computing and Communications with large-scale atomistic simulations of advanced materials. This work was supported by NSF (CAREER Program), ARO, PRF, and Louisiana LEQSF.
Solving the Cauchy-Riemann equations on parallel computers
NASA Technical Reports Server (NTRS)
Fatoohi, Raad A.; Grosch, Chester E.
1987-01-01
Discussed is the implementation of a single algorithm on three parallel-vector computers. The algorithm is a relaxation scheme for the solution of the Cauchy-Riemann equations; a set of coupled first order partial differential equations. The computers were chosen so as to encompass a variety of architectures. They are: the MPP, and SIMD machine with 16K bit serial processors; FLEX/32, an MIMD machine with 20 processors; and CRAY/2, an MIMD machine with four vector processors. The machine architectures are briefly described. The implementation of the algorithm is discussed in relation to these architectures and measures of the performance on each machine are given. Simple performance models are used to describe the performance. These models highlight the bottlenecks and limiting factors for this algorithm on these architectures. Conclusions are presented.
Data communications in a parallel active messaging interface of a parallel computer
Archer, Charles J; Blocksome, Michael A; Ratterman, Joseph D; Smith, Brian E
2013-11-12
Data communications in a parallel active messaging interface (`PAMI`) of a parallel computer composed of compute nodes that execute a parallel application, each compute node including application processors that execute the parallel application and at least one management processor dedicated to gathering information regarding data communications. The PAMI is composed of data communications endpoints, each endpoint composed of a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, the compute nodes and the endpoints coupled for data communications through the PAMI and through data communications resources. Embodiments function by gathering call site statistics describing data communications resulting from execution of data communications instructions and identifying in dependence upon the call cite statistics a data communications algorithm for use in executing a data communications instruction at a call site in the parallel application.
Zhang, Xiaohua; Wong, Sergio E; Lightstone, Felice C
2014-01-27
In this work we announce and evaluate a high throughput virtual screening pipeline for in-silico screening of virtual compound databases using high performance computing (HPC). Notable features of this pipeline are an automated receptor preparation scheme with unsupervised binding site identification. The pipeline includes receptor/target preparation, ligand preparation, VinaLC docking calculation, and molecular mechanics/generalized Born surface area (MM/GBSA) rescoring using the GB model by Onufriev and co-workers [J. Chem. Theory Comput. 2007, 3, 156-169]. Furthermore, we leverage HPC resources to perform an unprecedented, comprehensive evaluation of MM/GBSA rescoring when applied to the DUD-E data set (Directory of Useful Decoys: Enhanced), in which we selected 38 protein targets and a total of ∼0.7 million actives and decoys. The computer wall time for virtual screening has been reduced drastically on HPC machines, which increases the feasibility of extremely large ligand database screening with more accurate methods. HPC resources allowed us to rescore 20 poses per compound and evaluate the optimal number of poses to rescore. We find that keeping 5-10 poses is a good compromise between accuracy and computational expense. Overall the results demonstrate that MM/GBSA rescoring has higher average receiver operating characteristic (ROC) area under curve (AUC) values and consistently better early recovery of actives than Vina docking alone. Specifically, the enrichment performance is target-dependent. MM/GBSA rescoring significantly out performs Vina docking for the folate enzymes, kinases, and several other enzymes. The more accurate energy function and solvation terms of the MM/GBSA method allow MM/GBSA to achieve better enrichment, but the rescoring is still limited by the docking method to generate the poses with the correct binding modes. PMID:24358939
Zhang, Xiaohua; Wong, Sergio E; Lightstone, Felice C
2014-01-27
In this work we announce and evaluate a high throughput virtual screening pipeline for in-silico screening of virtual compound databases using high performance computing (HPC). Notable features of this pipeline are an automated receptor preparation scheme with unsupervised binding site identification. The pipeline includes receptor/target preparation, ligand preparation, VinaLC docking calculation, and molecular mechanics/generalized Born surface area (MM/GBSA) rescoring using the GB model by Onufriev and co-workers [J. Chem. Theory Comput. 2007, 3, 156-169]. Furthermore, we leverage HPC resources to perform an unprecedented, comprehensive evaluation of MM/GBSA rescoring when applied to the DUD-E data set (Directory of Useful Decoys: Enhanced), in which we selected 38 protein targets and a total of ∼0.7 million actives and decoys. The computer wall time for virtual screening has been reduced drastically on HPC machines, which increases the feasibility of extremely large ligand database screening with more accurate methods. HPC resources allowed us to rescore 20 poses per compound and evaluate the optimal number of poses to rescore. We find that keeping 5-10 poses is a good compromise between accuracy and computational expense. Overall the results demonstrate that MM/GBSA rescoring has higher average receiver operating characteristic (ROC) area under curve (AUC) values and consistently better early recovery of actives than Vina docking alone. Specifically, the enrichment performance is target-dependent. MM/GBSA rescoring significantly out performs Vina docking for the folate enzymes, kinases, and several other enzymes. The more accurate energy function and solvation terms of the MM/GBSA method allow MM/GBSA to achieve better enrichment, but the rescoring is still limited by the docking method to generate the poses with the correct binding modes.
Event parallelism: Distributed memory parallel computing for high energy physics experiments
Nash, T.
1989-05-01
This paper describes the present and expected future development of distributed memory parallel computers for high energy physics experiments. It covers the use of event parallel microprocessor farms, particularly at Fermilab, including both ACP multiprocessors and farms of MicroVAXES. These systems have proven very cost effective in the past. A case is made for moving to the more open environment of UNIX and RISC processors. The 2nd Generation ACP Multiprocessor System, which is based on powerful RISC systems, is described. Given the promise of still more extraordinary increases in processor performance, a new emphasis on point to point, rather than bussed, communication will be required. Developments in this direction are described. 6 figs.
Access and visualization using clusters and other parallel computers
NASA Technical Reports Server (NTRS)
Katz, D. S.; Bergou, A.; Berriman, B.; Block, G.; Collier, J.; Curkendall, D.; Good, J.; Husman, L.; Jacob, J.; Laity, A.; Li, P.; Miller, C.; Plesea, L.; Prince, T.; Siegel, H.; Williams, R.
2003-01-01
JPL's Parallel Applications Technologies Group has been exploring the issues of data access and visualization of very large data sets over the past 10 years. this work has used a number of types of parallel computers, and today includes the use of commodity clusters. This talk will highlight some of the applications and tools we have developed, including how they use parallel computing resources, and specifically how we are using modern clusters.
Improving parallel I/O autotuning with performance modeling
Behzad, Babak; Byna, Surendra; Wild, Stefan M.; Prabhat, -; Snir, Marc
2014-01-01
Various layers of the parallel I/O subsystem offer tunable parameters for improving I/O performance on large-scale computers. However, searching through a large parameter space is challenging. We are working towards an autotuning framework for determining the parallel I/O parameters that can achieve good I/O performance for different data write patterns. In this paper, we characterize parallel I/O and discuss the development of predictive models for use in effectively reducing the parameter space. Furthermore, applying our technique on tuning an I/O kernel derived from a large-scale simulation code shows that the search time can be reduced from 12 hours to 2more » hours, while achieving 54X I/O performance speedup.« less
Improving parallel I/O autotuning with performance modeling
Behzad, Babak; Byna, Surendra; Wild, Stefan M.; Prabhat, -; Snir, Marc
2014-01-01
Various layers of the parallel I/O subsystem offer tunable parameters for improving I/O performance on large-scale computers. However, searching through a large parameter space is challenging. We are working towards an autotuning framework for determining the parallel I/O parameters that can achieve good I/O performance for different data write patterns. In this paper, we characterize parallel I/O and discuss the development of predictive models for use in effectively reducing the parameter space. Furthermore, applying our technique on tuning an I/O kernel derived from a large-scale simulation code shows that the search time can be reduced from 12 hours to 2 hours, while achieving 54X I/O performance speedup.
Parallel Object-Oriented Computation Applied to a Finite Element Problem
NASA Technical Reports Server (NTRS)
Weissman, Jon B.; Grimshaw, Andrew S.; Ferraro, Robert
1993-01-01
The conventional wisdom in the scientific computing community is that the best way to solve large-scale numerically intensive scientific problems on today's parallel MIMD computers is to use Fortran or C programmed in a data-parallel style using low-level message-passing primitives. This approach inevitably leads to nonportable codes, extensive development time, and restricts parallel programming to the domain of the expert programmer. We believe that these problems are not inherent to parallel computing but are the result of the tools used. We will show that comparable performance can be achieved with little effort if better tools that present higher level abstractions are used.
Parallel Computation of Persistent Homology using the Blowup Complex
Lewis, Ryan; Morozov, Dmitriy
2015-04-27
We describe a parallel algorithm that computes persistent homology, an algebraic descriptor of a filtered topological space. Our algorithm is distinguished by operating on a spatial decomposition of the domain, as opposed to a decomposition with respect to the filtration. We rely on a classical construction, called the Mayer--Vietoris blowup complex, to glue global topological information about a space from its disjoint subsets. We introduce an efficient algorithm to perform this gluing operation, which may be of independent interest, and describe how to process the domain hierarchically. We report on a set of experiments that help assess the strengths and identify the limitations of our method.
Semi-automatic process partitioning for parallel computation
NASA Technical Reports Server (NTRS)
Koelbel, Charles; Mehrotra, Piyush; Vanrosendale, John
1988-01-01
On current multiprocessor architectures one must carefully distribute data in memory in order to achieve high performance. Process partitioning is the operation of rewriting an algorithm as a collection of tasks, each operating primarily on its own portion of the data, to carry out the computation in parallel. A semi-automatic approach to process partitioning is considered in which the compiler, guided by advice from the user, automatically transforms programs into such an interacting task system. This approach is illustrated with a picture processing example written in BLAZE, which is transformed into a task system maximizing locality of memory reference.
Signal processing applications of massively parallel charge domain computing devices
NASA Technical Reports Server (NTRS)
Fijany, Amir (Inventor); Barhen, Jacob (Inventor); Toomarian, Nikzad (Inventor)
1999-01-01
The present invention is embodied in a charge coupled device (CCD)/charge injection device (CID) architecture capable of performing a Fourier transform by simultaneous matrix vector multiplication (MVM) operations in respective plural CCD/CID arrays in parallel in O(1) steps. For example, in one embodiment, a first CCD/CID array stores charge packets representing a first matrix operator based upon permutations of a Hartley transform and computes the Fourier transform of an incoming vector. A second CCD/CID array stores charge packets representing a second matrix operator based upon different permutations of a Hartley transform and computes the Fourier transform of an incoming vector. The incoming vector is applied to the inputs of the two CCD/CID arrays simultaneously, and the real and imaginary parts of the Fourier transform are produced simultaneously in the time required to perform a single MVM operation in a CCD/CID array.
Parallel image computation in clusters with task-distributor.
Baun, Christian
2016-01-01
Distributed systems, especially clusters, can be used to execute ray tracing tasks in parallel for speeding up the image computation. Because ray tracing is a computational expensive and memory consuming task, ray tracing can also be used to benchmark clusters. This paper introduces task-distributor, a free software solution for the parallel execution of ray tracing tasks in distributed systems. The ray tracing solution used for this work is the Persistence Of Vision Raytracer (POV-Ray). Task-distributor does not require any modification of the POV-Ray source code or the installation of an additional message passing library like the Message Passing Interface or Parallel Virtual Machine to allow parallel image computation, in contrast to various other projects. By analyzing the runtime of the sequential and parallel program parts of task-distributor, it becomes clear how the problem size and available hardware resources influence the scaling of the parallel application.
Parallel image computation in clusters with task-distributor.
Baun, Christian
2016-01-01
Distributed systems, especially clusters, can be used to execute ray tracing tasks in parallel for speeding up the image computation. Because ray tracing is a computational expensive and memory consuming task, ray tracing can also be used to benchmark clusters. This paper introduces task-distributor, a free software solution for the parallel execution of ray tracing tasks in distributed systems. The ray tracing solution used for this work is the Persistence Of Vision Raytracer (POV-Ray). Task-distributor does not require any modification of the POV-Ray source code or the installation of an additional message passing library like the Message Passing Interface or Parallel Virtual Machine to allow parallel image computation, in contrast to various other projects. By analyzing the runtime of the sequential and parallel program parts of task-distributor, it becomes clear how the problem size and available hardware resources influence the scaling of the parallel application. PMID:27330898
Distributing an executable job load file to compute nodes in a parallel computer
Gooding, Thomas M.
2016-09-13
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.
Distributing an executable job load file to compute nodes in a parallel computer
Gooding, Thomas M.
2016-08-09
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.
Parallel 3-D Electromagnetic Particle Code Using High Performance FORTRAN: Parallel TRISTAN
NASA Astrophysics Data System (ADS)
Cai, D.; Li, Y.; Nishikawa, K.-I.; et al.
A three-dimensional full electromagnetic particle-in-cell (PIC ) code, TRISTAN (Tridimensional Stanford) code, has been parallelized using High Performance Fortran (HPF) as a RPM (Real Parallel Machine). In the parallelized HPF code, the simulation domain is decomposed in one-dimension, and both the particle and field data located in each domain that we call the sub-domain are distributed on each processor. Both the particle and field data on a sub-domain are needed by the neighbor sub-domains and thus communications between the sub-domains are inevitable. Our simulation results using HPF exhibit the promising applicability of the HPF communications to a large scale scientific computing such as solar wind-magnetosphere interactions.
Shen, Wenfeng; Wei, Daming; Xu, Weimin; Zhu, Xin; Yuan, Shizhong
2010-10-01
Biological computations like electrocardiological modelling and simulation usually require high-performance computing environments. This paper introduces an implementation of parallel computation for computer simulation of electrocardiograms (ECGs) in a personal computer environment with an Intel CPU of Core (TM) 2 Quad Q6600 and a GPU of Geforce 8800GT, with software support by OpenMP and CUDA. It was tested in three parallelization device setups: (a) a four-core CPU without a general-purpose GPU, (b) a general-purpose GPU plus 1 core of CPU, and (c) a four-core CPU plus a general-purpose GPU. To effectively take advantage of a multi-core CPU and a general-purpose GPU, an algorithm based on load-prediction dynamic scheduling was developed and applied to setting (c). In the simulation with 1600 time steps, the speedup of the parallel computation as compared to the serial computation was 3.9 in setting (a), 16.8 in setting (b), and 20.0 in setting (c). This study demonstrates that a current PC with a multi-core CPU and a general-purpose GPU provides a good environment for parallel computations in biological modelling and simulation studies.
Shen, Wenfeng; Wei, Daming; Xu, Weimin; Zhu, Xin; Yuan, Shizhong
2010-10-01
Biological computations like electrocardiological modelling and simulation usually require high-performance computing environments. This paper introduces an implementation of parallel computation for computer simulation of electrocardiograms (ECGs) in a personal computer environment with an Intel CPU of Core (TM) 2 Quad Q6600 and a GPU of Geforce 8800GT, with software support by OpenMP and CUDA. It was tested in three parallelization device setups: (a) a four-core CPU without a general-purpose GPU, (b) a general-purpose GPU plus 1 core of CPU, and (c) a four-core CPU plus a general-purpose GPU. To effectively take advantage of a multi-core CPU and a general-purpose GPU, an algorithm based on load-prediction dynamic scheduling was developed and applied to setting (c). In the simulation with 1600 time steps, the speedup of the parallel computation as compared to the serial computation was 3.9 in setting (a), 16.8 in setting (b), and 20.0 in setting (c). This study demonstrates that a current PC with a multi-core CPU and a general-purpose GPU provides a good environment for parallel computations in biological modelling and simulation studies. PMID:20674066
Scalable Performance Environments for Parallel Systems
NASA Technical Reports Server (NTRS)
Reed, Daniel A.; Olson, Robert D.; Aydt, Ruth A.; Madhyastha, Tara M.; Birkett, Thomas; Jensen, David W.; Nazief, Bobby A. A.; Totty, Brian K.
1991-01-01
As parallel systems expand in size and complexity, the absence of performance tools for these parallel systems exacerbates the already difficult problems of application program and system software performance tuning. Moreover, given the pace of technological change, we can no longer afford to develop ad hoc, one-of-a-kind performance instrumentation software; we need scalable, portable performance analysis tools. We describe an environment prototype based on the lessons learned from two previous generations of performance data analysis software. Our environment prototype contains a set of performance data transformation modules that can be interconnected in user-specified ways. It is the responsibility of the environment infrastructure to hide details of module interconnection and data sharing. The environment is written in C++ with the graphical displays based on X windows and the Motif toolkit. It allows users to interconnect and configure modules graphically to form an acyclic, directed data analysis graph. Performance trace data are represented in a self-documenting stream format that includes internal definitions of data types, sizes, and names. The environment prototype supports the use of head-mounted displays and sonic data presentation in addition to the traditional use of visual techniques.
Toward a science of parallel computation
Worlton, W.J.
1986-01-01
The evolution of parallel processing over the past several decades can be viewed as the development of a new scientific discipline. Parallel processing has been, and is, undergoing the same evolutionary stages that are common to the development of scientific disciplines in general: exploration, focusing, and maturity. That parallel processing is not yet a science can readily be appreciated by its lack of some of the characteristics typical of mature sciences, such as prescriptive terminology, comprehensive taxonomies, and authoritative fundamental principles. A great deal of outstanding work has been done and the field is experiencing the beginnings of its ''focusing'' phase, i.e., support is being concentrated in a set of the more promising approaches selected from among the larger set of exploratory projects. However, the possible set of parallel-processing concepts is so extensive that exploratory work will probably continue for one or two more decades. In the meantime, the growing maturity of the field will be reflected in the increasing clarity and precision of the terminology, the development of systematic classification of the domain of discourse, the development of basic principles, and the growing number of commercial products that are the outcome of the research and development projects on which support is being focused. In this paper we develop some generalizations of taxonomies and use basic principles to draw conclusions about the extensibility of parallel processor architectures. 7 refs., 5 figs., 2 tabs.
NASA Astrophysics Data System (ADS)
Ford, Eric B.; Dindar, Saleh; Peters, Jorg
2015-08-01
The realism of astrophysical simulations and statistical analyses of astronomical data are set by the available computational resources. Thus, astronomers and astrophysicists are constantly pushing the limits of computational capabilities. For decades, astronomers benefited from massive improvements in computational power that were driven primarily by increasing clock speeds and required relatively little attention to details of the computational hardware. For nearly a decade, increases in computational capabilities have come primarily from increasing the degree of parallelism, rather than increasing clock speeds. Further increases in computational capabilities will likely be led by many-core architectures such as Graphical Processing Units (GPUs) and Intel Xeon Phi. Successfully harnessing these new architectures, requires significantly more understanding of the hardware architecture, cache hierarchy, compiler capabilities and network network characteristics.I will provide an astronomer's overview of the opportunities and challenges provided by modern many-core architectures and elastic cloud computing. The primary goal is to help an astronomical audience understand what types of problems are likely to yield more than order of magnitude speed-ups and which problems are unlikely to parallelize sufficiently efficiently to be worth the development time and/or costs.I will draw on my experience leading a team in developing the Swarm-NG library for parallel integration of large ensembles of small n-body systems on GPUs, as well as several smaller software projects. I will share lessons learned from collaborating with computer scientists, including both technical and soft skills. Finally, I will discuss the challenges of training the next generation of astronomers to be proficient in this new era of high-performance computing, drawing on experience teaching a graduate class on High-Performance Scientific Computing for Astrophysics and organizing a 2014 advanced summer
Data communications in a parallel active messaging interface of a parallel computer
Archer, Charles J; Blocksome, Michael A; Ratterman, Joseph D; Smith, Brian E
2014-02-11
Data communications in a parallel active messaging interface ('PAMI') or a parallel computer, the parallel computer including a plurality of compute nodes that execute a parallel application, the PAMI composed of data communications endpoints, each endpoint including a specification of data communications parameters for a thread of execution of a compute node, including specification of a client, a context, and a task, the compute nodes and the endpoints coupled for data communications instruction, the instruction characterized by instruction type, the instruction specifying a transmission of transfer data from the origin endpoint to a target endpoint and transmitting, in accordance witht the instruction type, the transfer data from the origin endpoin to the target endpoint.
Data communications in a parallel active messaging interface of a parallel computer
Archer, Charles J; Blocksome, Michael A; Ratterman, Joseph D; Smith, Brian E
2013-10-29
Data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the parallel computer including a plurality of compute nodes that execute a parallel application, the PAMI composed of data communications endpoints, each endpoint including a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, the compute nodes and the endpoints coupled for data communications through the PAMI and through data communications resources, including receiving in an origin endpoint of the PAMI a data communications instruction, the instruction characterized by an instruction type, the instruction specifying a transmission of transfer data from the origin endpoint to a target endpoint and transmitting, in accordance with the instruction type, the transfer data from the origin endpoint to the target endpoint.
Review of An Introduction to Parallel and Vector Scientific Computing
Bailey, David H.; Lefton, Lew
2006-06-30
On one hand, the field of high-performance scientific computing is thriving beyond measure. Performance of leading-edge systems on scientific calculations, as measured say by the Top500 list, has increased by an astounding factor of 8000 during the 15-year period from 1993 to 2008, which is slightly faster even than Moore's Law. Even more importantly, remarkable advances in numerical algorithms, numerical libraries and parallel programming environments have led to improvements in the scope of what can be computed that are entirely on a par with the advances in computing hardware. And these successes have spread far beyond the confines of large government-operated laboratories, many universities, modest-sized research institutes and private firms now operate clusters that differ only in scale from the behemoth systems at the large-scale facilities. In the wake of these recent successes, researchers from fields that heretofore have not been part of the scientific computing world have been drawn into the arena. For example, at the recent SC07 conference, the exhibit hall, which long has hosted displays from leading computer systems vendors and government laboratories, featured some 70 exhibitors who had not previously participated. In spite of all these exciting developments, and in spite of the clear need to present these concepts to a much broader technical audience, there is a perplexing dearth of training material and textbooks in the field, particularly at the introductory level. Only a handful of universities offer coursework in the specific area of highly parallel scientific computing, and instructors of such courses typically rely on custom-assembled material. For example, the present reviewer and Robert F. Lucas relied on materials assembled in a somewhat ad-hoc fashion from colleagues and personal resources when presenting a course on parallel scientific computing at the University of California, Berkeley, a few years ago. Thus it is indeed refreshing to see
Flexibility and Performance of Parallel File Systems
NASA Technical Reports Server (NTRS)
Kotz, David; Nieuwejaar, Nils
1996-01-01
As we gain experience with parallel file systems, it becomes increasingly clear that a single solution does not suit all applications. For example, it appears to be impossible to find a single appropriate interface, caching policy, file structure, or disk-management strategy. Furthermore, the proliferation of file-system interfaces and abstractions make applications difficult to port. We propose that the traditional functionality of parallel file systems be separated into two components: a fixed core that is standard on all platforms, encapsulating only primitive abstractions and interfaces, and a set of high-level libraries to provide a variety of abstractions and application-programmer interfaces (API's). We present our current and next-generation file systems as examples of this structure. Their features, such as a three-dimensional file structure, strided read and write interfaces, and I/O-node programs, are specifically designed with the flexibility and performance necessary to support a wide range of applications.
Parallel computation of multigroup reactivity coefficient using iterative method
NASA Astrophysics Data System (ADS)
Susmikanti, Mike; Dewayatna, Winter
2013-09-01
One of the research activities to support the commercial radioisotope production program is a safety research target irradiation FPM (Fission Product Molybdenum). FPM targets form a tube made of stainless steel in which the nuclear degrees of superimposed high-enriched uranium. FPM irradiation tube is intended to obtain fission. The fission material widely used in the form of kits in the world of nuclear medicine. Irradiation FPM tube reactor core would interfere with performance. One of the disorders comes from changes in flux or reactivity. It is necessary to study a method for calculating safety terrace ongoing configuration changes during the life of the reactor, making the code faster became an absolute necessity. Neutron safety margin for the research reactor can be reused without modification to the calculation of the reactivity of the reactor, so that is an advantage of using perturbation method. The criticality and flux in multigroup diffusion model was calculate at various irradiation positions in some uranium content. This model has a complex computation. Several parallel algorithms with iterative method have been developed for the sparse and big matrix solution. The Black-Red Gauss Seidel Iteration and the power iteration parallel method can be used to solve multigroup diffusion equation system and calculated the criticality and reactivity coeficient. This research was developed code for reactivity calculation which used one of safety analysis with parallel processing. It can be done more quickly and efficiently by utilizing the parallel processing in the multicore computer. This code was applied for the safety limits calculation of irradiated targets FPM with increment Uranium.
A hybrid method for the parallel computation of Green's functions
Petersen, Dan Erik; Li Song; Stokbro, Kurt; Sorensen, Hans Henrik B.; Hansen, Per Christian; Skelboe, Stig; Darve, Eric
2009-08-01
Quantum transport models for nanodevices using the non-equilibrium Green's function method require the repeated calculation of the block tridiagonal part of the Green's and lesser Green's function matrices. This problem is related to the calculation of the inverse of a sparse matrix. Because of the large number of times this calculation needs to be performed, this is computationally very expensive even on supercomputers. The classical approach is based on recurrence formulas which cannot be efficiently parallelized. This practically prevents the solution of large problems with hundreds of thousands of atoms. We propose new recurrences for a general class of sparse matrices to calculate Green's and lesser Green's function matrices which extend formulas derived by Takahashi and others. We show that these recurrences may lead to a dramatically reduced computational cost because they only require computing a small number of entries of the inverse matrix. Then, we propose a parallelization strategy for block tridiagonal matrices which involves a combination of Schur complement calculations and cyclic reduction. It achieves good scalability even on problems of modest size.
Prediction of parallel NIKE3D performance on the KSR1 system
Su, P.S.; Zacharia, T.; Fulton, R.E.
1995-05-01
Finite element method is one of the bases for numerical solutions to engineering problems. Complex engineering problems using finite element analysis typically imply excessively large computational time. Parallel supercomputers have the potential for significantly increasing calculation speeds in order to meet these computational requirements. This paper predicts parallel NIKE3D performance on the Kendall Square Research (KSR1) system. The first part of the prediction is based on the implementation of parallel Cholesky (U{sup T}DU) matrix decomposition algorithm through actual computations on the KSRI multiprocessor system, with 64 processors, at Oak Ridge National Laboratory. The other predictions are based on actual computations for parallel element matrix generation, parallel global stiffness matrix assembly, and parallel forward/backward substitution on the BBN TC2000 multiprocessor system at Lawrence Livermore National Laboratory. The preliminary results indicate that parallel NIKE3D performance can be attractive under local/shared-memory multiprocessor system environments.
Performance Evaluation Methodologies and Tools for Massively Parallel Programs
NASA Technical Reports Server (NTRS)
Yan, Jerry C.; Sarukkai, Sekhar; Tucker, Deanne (Technical Monitor)
1994-01-01
The need for computing power has forced a migration from serial computation on a single processor to parallel processing on multiprocessors. However, without effective means to monitor (and analyze) program execution, tuning the performance of parallel programs becomes exponentially difficult as program complexity and machine size increase. The recent introduction of performance tuning tools from various supercomputer vendors (Intel's ParAide, TMC's PRISM, CSI'S Apprentice, and Convex's CXtrace) seems to indicate the maturity of performance tool technologies and vendors'/customers' recognition of their importance. However, a few important questions remain: What kind of performance bottlenecks can these tools detect (or correct)? How time consuming is the performance tuning process? What are some important technical issues that remain to be tackled in this area? This workshop reviews the fundamental concepts involved in analyzing and improving the performance of parallel and heterogeneous message-passing programs. Several alternative strategies will be contrasted, and for each we will describe how currently available tuning tools (e.g., AIMS, ParAide, PRISM, Apprentice, CXtrace, ATExpert, Pablo, IPS-2)) can be used to facilitate the process. We will characterize the effectiveness of the tools and methodologies based on actual user experiences at NASA Ames Research Center. Finally, we will discuss their limitations and outline recent approaches taken by vendors and the research community to address them.
Performance variability of highly parallel architectures
Kramer, William T.C.; Ryan, Clint
2003-05-01
The design and evaluation of high performance computers has concentrated on increasing computational speed for applications. This performance is often measured on a well configured dedicated system to show the best case. In the real environment, resources are not always dedicated to a single task, and systems run tasks that may influence each other, so run times vary, sometimes to an unreasonably large extent. This paper explores the amount of variation seen across four large distributed memory systems in a systematic manner. It then analyzes the causes for the variations seen and discusses what can be done to decrease the variation without impacting performance.
[Series: Medical Applications of the PHITS Code (2): Acceleration by Parallel Computing].
Furuta, Takuya; Sato, Tatsuhiko
2015-01-01
Time-consuming Monte Carlo dose calculation becomes feasible owing to the development of computer technology. However, the recent development is due to emergence of the multi-core high performance computers. Therefore, parallel computing becomes a key to achieve good performance of software programs. A Monte Carlo simulation code PHITS contains two parallel computing functions, the distributed-memory parallelization using protocols of message passing interface (MPI) and the shared-memory parallelization using open multi-processing (OpenMP) directives. Users can choose the two functions according to their needs. This paper gives the explanation of the two functions with their advantages and disadvantages. Some test applications are also provided to show their performance using a typical multi-core high performance workstation.
Performance and Scalability of the NAS Parallel Benchmarks in Java
NASA Technical Reports Server (NTRS)
Frumkin, Michael A.; Schultz, Matthew; Jin, Haoqiang; Yan, Jerry; Biegel, Bryan A. (Technical Monitor)
2002-01-01
Several features make Java an attractive choice for scientific applications. In order to gauge the applicability of Java to Computational Fluid Dynamics (CFD), we have implemented the NAS (NASA Advanced Supercomputing) Parallel Benchmarks in Java. The performance and scalability of the benchmarks point out the areas where improvement in Java compiler technology and in Java thread implementation would position Java closer to Fortran in the competition for scientific applications.
Analytic reconstruction approach for parallel translational computed tomography.
Kong, Huihua; Yu, Hengyong
2015-01-01
To develop low-cost and low-dose computed tomography (CT) scanners for developing countries, recently a parallel translational computed tomography (PTCT) is proposed, and the source and detector are translated oppositely with respect to the imaging object without a slip-ring. In this paper, we develop an analytic filtered-backprojection (FBP)-type reconstruction algorithm for two dimensional (2D) fan-beam PTCT and extend it to three dimensional (3D) cone-beam geometry in a Feldkamp-type framework. Particularly, a weighting function is constructed to deal with data redundancy for multiple translations PTCT to eliminate image artifacts. Extensive numerical simulations are performed to validate and evaluate the proposed analytic reconstruction algorithms, and the results confirm their correctness and merits. PMID:25882732
An efficient parallel algorithm for accelerating computational protein design
Zhou, Yichao; Xu, Wei; Donald, Bruce R.; Zeng, Jianyang
2014-01-01
Motivation: Structure-based computational protein design (SCPR) is an important topic in protein engineering. Under the assumption of a rigid backbone and a finite set of discrete conformations of side-chains, various methods have been proposed to address this problem. A popular method is to combine the dead-end elimination (DEE) and A* tree search algorithms, which provably finds the global minimum energy conformation (GMEC) solution. Results: In this article, we improve the efficiency of computing A* heuristic functions for protein design and propose a variant of A* algorithm in which the search process can be performed on a single GPU in a massively parallel fashion. In addition, we make some efforts to address the memory exceeding problem in A* search. As a result, our enhancements can achieve a significant speedup of the A*-based protein design algorithm by four orders of magnitude on large-scale test data through pre-computation and parallelization, while still maintaining an acceptable memory overhead. We also show that our parallel A* search algorithm could be successfully combined with iMinDEE, a state-of-the-art DEE criterion, for rotamer pruning to further improve SCPR with the consideration of continuous side-chain flexibility. Availability: Our software is available and distributed open-source under the GNU Lesser General License Version 2.1 (GNU, February 1999). The source code can be downloaded from http://www.cs.duke.edu/donaldlab/osprey.php or http://iiis.tsinghua.edu.cn/∼compbio/software.html. Contact: zengjy321@tsinghua.edu.cn Supplementary information: Supplementary data are available at Bioinformatics online. PMID:24931991
High Performance Computing Today
Dongarra, Jack; Meuer,Hans; Simon,Horst D.; Strohmaier,Erich
2000-04-01
In last 50 years, the field of scientific computing has seen a rapid change of vendors, architectures, technologies and the usage of systems. Despite all these changes the evolution of performance on a large scale however seems to be a very steady and continuous process. Moore's Law is often cited in this context. If the authors plot the peak performance of various computers of the last 5 decades in Figure 1 that could have been called the supercomputers of their time they indeed see how well this law holds for almost the complete lifespan of modern computing. On average they see an increase in performance of two magnitudes of order every decade.
Integrating Cache Performance Modeling and Tuning Support in Parallelization Tools
NASA Technical Reports Server (NTRS)
Waheed, Abdul; Yan, Jerry; Saini, Subhash (Technical Monitor)
1998-01-01
With the resurgence of distributed shared memory (DSM) systems based on cache-coherent Non Uniform Memory Access (ccNUMA) architectures and increasing disparity between memory and processors speeds, data locality overheads are becoming the greatest bottlenecks in the way of realizing potential high performance of these systems. While parallelization tools and compilers facilitate the users in porting their sequential applications to a DSM system, a lot of time and effort is needed to tune the memory performance of these applications to achieve reasonable speedup. In this paper, we show that integrating cache performance modeling and tuning support within a parallelization environment can alleviate this problem. The Cache Performance Modeling and Prediction Tool (CPMP), employs trace-driven simulation techniques without the overhead of generating and managing detailed address traces. CPMP predicts the cache performance impact of source code level "what-if" modifications in a program to assist a user in the tuning process. CPMP is built on top of a customized version of the Computer Aided Parallelization Tools (CAPTools) environment. Finally, we demonstrate how CPMP can be applied to tune a real Computational Fluid Dynamics (CFD) application.
Cooperative storage of shared files in a parallel computing system with dynamic block size
Bent, John M.; Faibish, Sorin; Grider, Gary
2015-11-10
Improved techniques are provided for parallel writing of data to a shared object in a parallel computing system. A method is provided for storing data generated by a plurality of parallel processes to a shared object in a parallel computing system. The method is performed by at least one of the processes and comprises: dynamically determining a block size for storing the data; exchanging a determined amount of the data with at least one additional process to achieve a block of the data having the dynamically determined block size; and writing the block of the data having the dynamically determined block size to a file system. The determined block size comprises, e.g., a total amount of the data to be stored divided by the number of parallel processes. The file system comprises, for example, a log structured virtual parallel file system, such as a Parallel Log-Structured File System (PLFS).
Research in Parallel Algorithms and Software for Computational Aerosciences
NASA Technical Reports Server (NTRS)
Domel, Neal D.
1996-01-01
Phase I is complete for the development of a Computational Fluid Dynamics parallel code with automatic grid generation and adaptation for the Euler analysis of flow over complex geometries. SPLITFLOW, an unstructured Cartesian grid code developed at Lockheed Martin Tactical Aircraft Systems, has been modified for a distributed memory/massively parallel computing environment. The parallel code is operational on an SGI network, Cray J90 and C90 vector machines, SGI Power Challenge, and Cray T3D and IBM SP2 massively parallel machines. Parallel Virtual Machine (PVM) is the message passing protocol for portability to various architectures. A domain decomposition technique was developed which enforces dynamic load balancing to improve solution speed and memory requirements. A host/node algorithm distributes the tasks. The solver parallelizes very well, and scales with the number of processors. Partially parallelized and non-parallelized tasks consume most of the wall clock time in a very fine grain environment. Timing comparisons on a Cray C90 demonstrate that Parallel SPLITFLOW runs 2.4 times faster on 8 processors than its non-parallel counterpart autotasked over 8 processors.
Research in Parallel Algorithms and Software for Computational Aerosciences
NASA Technical Reports Server (NTRS)
Domel, Neal D.
1996-01-01
Phase 1 is complete for the development of a computational fluid dynamics CFD) parallel code with automatic grid generation and adaptation for the Euler analysis of flow over complex geometries. SPLITFLOW, an unstructured Cartesian grid code developed at Lockheed Martin Tactical Aircraft Systems, has been modified for a distributed memory/massively parallel computing environment. The parallel code is operational on an SGI network, Cray J90 and C90 vector machines, SGI Power Challenge, and Cray T3D and IBM SP2 massively parallel machines. Parallel Virtual Machine (PVM) is the message passing protocol for portability to various architectures. A domain decomposition technique was developed which enforces dynamic load balancing to improve solution speed and memory requirements. A host/node algorithm distributes the tasks. The solver parallelizes very well, and scales with the number of processors. Partially parallelized and non-parallelized tasks consume most of the wall clock time in a very fine grain environment. Timing comparisons on a Cray C90 demonstrate that Parallel SPLITFLOW runs 2.4 times faster on 8 processors than its non-parallel counterpart autotasked over 8 processors.
Parallel processing using an optical delay-based reservoir computer
NASA Astrophysics Data System (ADS)
Van der Sande, Guy; Nguimdo, Romain Modeste; Verschaffelt, Guy
2016-04-01
Delay systems subject to delayed optical feedback have recently shown great potential in solving computationally hard tasks. By implementing a neuro-inspired computational scheme relying on the transient response to optical data injection, high processing speeds have been demonstrated. However, reservoir computing systems based on delay dynamics discussed in the literature are designed by coupling many different stand-alone components which lead to bulky, lack of long-term stability, non-monolithic systems. Here we numerically investigate the possibility of implementing reservoir computing schemes based on semiconductor ring lasers. Semiconductor ring lasers are semiconductor lasers where the laser cavity consists of a ring-shaped waveguide. SRLs are highly integrable and scalable, making them ideal candidates for key components in photonic integrated circuits. SRLs can generate light in two counterpropagating directions between which bistability has been demonstrated. We demonstrate that two independent machine learning tasks , even with different nature of inputs with different input data signals can be simultaneously computed using a single photonic nonlinear node relying on the parallelism offered by photonics. We illustrate the performance on simultaneous chaotic time series prediction and a classification of the Nonlinear Channel Equalization. We take advantage of different directional modes to process individual tasks. Each directional mode processes one individual task to mitigate possible crosstalk between the tasks. Our results indicate that prediction/classification with errors comparable to the state-of-the-art performance can be obtained even with noise despite the two tasks being computed simultaneously. We also find that a good performance is obtained for both tasks for a broad range of the parameters. The results are discussed in detail in [Nguimdo et al., IEEE Trans. Neural Netw. Learn. Syst. 26, pp. 3301-3307, 2015
QCD on the Massively Parallel Computer AP1000
NASA Astrophysics Data System (ADS)
Akemi, K.; Fujisaki, M.; Okuda, M.; Tago, Y.; Hashimoto, T.; Hioki, S.; Miyamura, O.; Takaishi, T.; Nakamura, A.; de Forcrand, Ph.; Hege, C.; Stamatescu, I. O.
We present the QCD-TARO program of calculations which uses the parallel computer AP1000 of Fujitsu. We discuss the results on scaling, correlation times and hadronic spectrum, some aspects of the implementation and the future prospects.
Collignon, Barbara; Schulz, Roland; Smith, Jeremy C; Baudry, Jerome
2011-04-30
A message passing interface (MPI)-based implementation (Autodock4.lga.MPI) of the grid-based docking program Autodock4 has been developed to allow simultaneous and independent docking of multiple compounds on up to thousands of central processing units (CPUs) using the Lamarkian genetic algorithm. The MPI version reads a single binary file containing precalculated grids that represent the protein-ligand interactions, i.e., van der Waals, electrostatic, and desolvation potentials, and needs only two input parameter files for the entire docking run. In comparison, the serial version of Autodock4 reads ASCII grid files and requires one parameter file per compound. The modifications performed result in significantly reduced input/output activity compared with the serial version. Autodock4.lga.MPI scales up to 8192 CPUs with a maximal overhead of 16.3%, of which two thirds is due to input/output operations and one third originates from MPI operations. The optimal docking strategy, which minimizes docking CPU time without lowering the quality of the database enrichments, comprises the docking of ligands preordered from the most to the least flexible and the assignment of the number of energy evaluations as a function of the number of rotatable bounds. In 24 h, on 8192 high-performance computing CPUs, the present MPI version would allow docking to a rigid protein of about 300K small flexible compounds or 11 million rigid compounds.
Archer, Charles J.; Blocksome, Michael A.; Peters, Amanda A.; Ratterman, Joseph D.; Smith, Brian E.
2012-01-10
Methods, apparatus, and products are disclosed for reducing power consumption while synchronizing a plurality of compute nodes during execution of a parallel application that include: beginning, by each compute node, performance of a blocking operation specified by the parallel application, each compute node beginning the blocking operation asynchronously with respect to the other compute nodes; reducing, for each compute node, power to one or more hardware components of that compute node in response to that compute node beginning the performance of the blocking operation; and restoring, for each compute node, the power to the hardware components having power reduced in response to all of the compute nodes beginning the performance of the blocking operation.
Archer, Charles J.; Blocksome, Michael A.; Peters, Amanda E.; Ratterman, Joseph D.; Smith, Brian E.
2012-04-17
Methods, apparatus, and products are disclosed for reducing power consumption while synchronizing a plurality of compute nodes during execution of a parallel application that include: beginning, by each compute node, performance of a blocking operation specified by the parallel application, each compute node beginning the blocking operation asynchronously with respect to the other compute nodes; reducing, for each compute node, power to one or more hardware components of that compute node in response to that compute node beginning the performance of the blocking operation; and restoring, for each compute node, the power to the hardware components having power reduced in response to all of the compute nodes beginning the performance of the blocking operation.
Parallel Computing for Probabilistic Response Analysis of High Temperature Composites
NASA Technical Reports Server (NTRS)
Sues, R. H.; Lua, Y. J.; Smith, M. D.
1994-01-01
The objective of this Phase I research was to establish the required software and hardware strategies to achieve large scale parallelism in solving PCM problems. To meet this objective, several investigations were conducted. First, we identified the multiple levels of parallelism in PCM and the computational strategies to exploit these parallelisms. Next, several software and hardware efficiency investigations were conducted. These involved the use of three different parallel programming paradigms and solution of two example problems on both a shared-memory multiprocessor and a distributed-memory network of workstations.
Programming Probabilistic Structural Analysis for Parallel Processing Computer
NASA Technical Reports Server (NTRS)
Sues, Robert H.; Chen, Heh-Chyun; Twisdale, Lawrence A.; Chamis, Christos C.; Murthy, Pappu L. N.
1991-01-01
The ultimate goal of this research program is to make Probabilistic Structural Analysis (PSA) computationally efficient and hence practical for the design environment by achieving large scale parallelism. The paper identifies the multiple levels of parallelism in PSA, identifies methodologies for exploiting this parallelism, describes the development of a parallel stochastic finite element code, and presents results of two example applications. It is demonstrated that speeds within five percent of those theoretically possible can be achieved. A special-purpose numerical technique, the stochastic preconditioned conjugate gradient method, is also presented and demonstrated to be extremely efficient for certain classes of PSA problems.
Multitasking TORT Under UNICOS: Parallel Performance Models and Measurements
Azmy, Y.Y.; Barnett, D.A.
1999-09-27
The existing parallel algorithms in the TORT discrete ordinates were updated to function in a UNI-COS environment. A performance model for the parallel overhead was derived for the existing algorithms. The largest contributors to the parallel overhead were identified and a new algorithm was developed. A parallel overhead model was also derived for the new algorithm. The results of the comparison of parallel performance models were compared to applications of the code to two TORT standard test problems and a large production problem. The parallel performance models agree well with the measured parallel overhead.
Multitasking TORT under UNICOS: Parallel performance models and measurements
Barnett, A.; Azmy, Y.Y.
1999-09-27
The existing parallel algorithms in the TORT discrete ordinates code were updated to function in a UNICOS environment. A performance model for the parallel overhead was derived for the existing algorithms. The largest contributors to the parallel overhead were identified and a new algorithm was developed. A parallel overhead model was also derived for the new algorithm. The results of the comparison of parallel performance models were compared to applications of the code to two TORT standard test problems and a large production problem. The parallel performance models agree well with the measured parallel overhead.
History Matching in Parallel Computational Environments
Steven Bryant; Sanjay Srinivasan; Alvaro Barrera; Sharad Yadav
2004-08-31
In the probabilistic approach for history matching, the information from the dynamic data is merged with the prior geologic information in order to generate permeability models consistent with the observed dynamic data as well as the prior geology. The relationship between dynamic response data and reservoir attributes may vary in different regions of the reservoir due to spatial variations in reservoir attributes, fluid properties, well configuration, flow constrains on wells etc. This implies probabilistic approach should then update different regions of the reservoir in different ways. This necessitates delineation of multiple reservoir domains in order to increase the accuracy of the approach. The research focuses on a probabilistic approach to integrate dynamic data that ensures consistency between reservoir models developed from one stage to the next. The algorithm relies on efficient parameterization of the dynamic data integration problem and permits rapid assessment of the updated reservoir model at each stage. The report also outlines various domain decomposition schemes from the perspective of increasing the accuracy of probabilistic approach of history matching. Research progress in three important areas of the project are discussed: {lg_bullet}Validation and testing the probabilistic approach to incorporating production data in reservoir models. {lg_bullet}Development of a robust scheme for identifying reservoir regions that will result in a more robust parameterization of the history matching process. {lg_bullet}Testing commercial simulators for parallel capability and development of a parallel algorithm for history matching.
Access and visualization using clusters and other parallel computers
NASA Technical Reports Server (NTRS)
Katz, Daniel S.; Bergou, Attila; Berriman, Bruce; Block, Gary; Collier, Jim; Curkendall, Dave; Good, John; Husman, Laura; Jacob, Joe; Laity, Anastasia; Li, Peggy; Miller, Craig; Plesea, Lucian; Prince, Tom; Siegel, Herb; Williams, Roy
2003-01-01
JPL's Parallel Applications Technologies Group has been exploring the issues of data access and visualization of very large data sets over the past 10 or so years. this work has used a number of types of parallel computers, and today includes the use of commodity clusters. This talk will highlight some of the applications and tools we have developed, including how they use parallel computing resources, and specifically how we are using modern clusters. Our applications focus on NASA's needs; thus our data sets are usually related to Earth and Space Science, including data delivered from instruments in space, and data produced by telescopes on the ground.
Partitioning problems in parallel, pipelined and distributed computing
NASA Technical Reports Server (NTRS)
Bokhari, S.
1985-01-01
The problem of optimally assigning the modules of a parallel program over the processors of a multiple computer system is addressed. A Sum-Bottleneck path algorithm is developed that permits the efficient solution of many variants of this problem under some constraints on the structure of the partitions. In particular, the following problems are solved optimally for a single-host, multiple satellite system: partitioning multiple chain structured parallel programs, multiple arbitrarily structured serial programs and single tree structured parallel programs. In addition, the problems of partitioning chain structured parallel programs across chain connected systems and across shared memory (or shared bus) systems are also solved under certain constraints. All solutions for parallel programs are equally applicable to pipelined programs. These results extend prior research in this area by explicitly taking concurrency into account and permit the efficient utilization of multiple computer architectures for a wide range of problems of practical interest.
Cloud identification using genetic algorithms and massively parallel computation
NASA Technical Reports Server (NTRS)
Buckles, Bill P.; Petry, Frederick E.
1996-01-01
As a Guest Computational Investigator under the NASA administered component of the High Performance Computing and Communication Program, we implemented a massively parallel genetic algorithm on the MasPar SIMD computer. Experiments were conducted using Earth Science data in the domains of meteorology and oceanography. Results obtained in these domains are competitive with, and in most cases better than, similar problems solved using other methods. In the meteorological domain, we chose to identify clouds using AVHRR spectral data. Four cloud speciations were used although most researchers settle for three. Results were remarkedly consistent across all tests (91% accuracy). Refinements of this method may lead to more timely and complete information for Global Circulation Models (GCMS) that are prevalent in weather forecasting and global environment studies. In the oceanographic domain, we chose to identify ocean currents from a spectrometer having similar characteristics to AVHRR. Here the results were mixed (60% to 80% accuracy). Given that one is willing to run the experiment several times (say 10), then it is acceptable to claim the higher accuracy rating. This problem has never been successfully automated. Therefore, these results are encouraging even though less impressive than the cloud experiment. Successful conclusion of an automated ocean current detection system would impact coastal fishing, naval tactics, and the study of micro-climates. Finally we contributed to the basic knowledge of GA (genetic algorithm) behavior in parallel environments. We developed better knowledge of the use of subpopulations in the context of shared breeding pools and the migration of individuals. Rigorous experiments were conducted based on quantifiable performance criteria. While much of the work confirmed current wisdom, for the first time we were able to submit conclusive evidence. The software developed under this grant was placed in the public domain. An extensive user
Parallel algorithms and archtectures for computational structural mechanics
NASA Technical Reports Server (NTRS)
Patrick, Merrell; Ma, Shing; Mahajan, Umesh
1989-01-01
The determination of the fundamental (lowest) natural vibration frequencies and associated mode shapes is a key step used to uncover and correct potential failures or problem areas in most complex structures. However, the computation time taken by finite element codes to evaluate these natural frequencies is significant, often the most computationally intensive part of structural analysis calculations. There is continuing need to reduce this computation time. This study addresses this need by developing methods for parallel computation.
Parallel structures in human and computer memory
NASA Technical Reports Server (NTRS)
Kanerva, P.
1986-01-01
If one thinks of our experiences as being recorded continuously on film, then human memory can be compared to a film library that is indexed by the contents of the film strips stored in it. Moreover, approximate retrieval cues suffice to retrieve information stored in this library. One recognizes a familiar person in a fuzzy photograph or a familiar tune played on a strange instrument. A computer memory that would allow a computer to recognize patterns and to recall sequences the way humans do is constructed. Such a memory is remarkably similiar in structure to a conventional computer memory and also to the neural circuits in the cortex of the cerebellum of the human brain. It is concluded that the frame problem of artificial intelligence could be solved by the use of such a memory if one were able to encode information about the world properly.
Parallel structures in human and computer memory
NASA Astrophysics Data System (ADS)
Kanerva, Pentti
1986-08-01
If we think of our experiences as being recorded continuously on film, then human memory can be compared to a film library that is indexed by the contents of the film strips stored in it. Moreover, approximate retrieval cues suffice to retrieve information stored in this library: We recognize a familiar person in a fuzzy photograph or a familiar tune played on a strange instrument. This paper is about how to construct a computer memory that would allow a computer to recognize patterns and to recall sequences the way humans do. Such a memory is remarkably similar in structure to a conventional computer memory and also to the neural circuits in the cortex of the cerebellum of the human brain. The paper concludes that the frame problem of artificial intelligence could be solved by the use of such a memory if we were able to encode information about the world properly.
A parallel performance study of the Cartesian method for partial differential equations on a sphere
Drake, J.B.; Coddington, M.P.
1997-04-01
A 3-D Cartesian method for integration of partial differential equations on a spherical surface is developed for parallel computation. The target computer architectures are distributed memory, message passing computers such as the Intel Paragon. The parallel algorithms are described along with mesh partitioning strategies. Performance of the algorithms is considered for a standard test case of the shallow water equations on the sphere. The authors find the computation time scale well with increasing numbers of processors.
NASA Technical Reports Server (NTRS)
Fijany, Amir (Inventor); Bejczy, Antal K. (Inventor)
1993-01-01
This is a real-time robotic controller and simulator which is a MIMD-SIMD parallel architecture for interfacing with an external host computer and providing a high degree of parallelism in computations for robotic control and simulation. It includes a host processor for receiving instructions from the external host computer and for transmitting answers to the external host computer. There are a plurality of SIMD microprocessors, each SIMD processor being a SIMD parallel processor capable of exploiting fine grain parallelism and further being able to operate asynchronously to form a MIMD architecture. Each SIMD processor comprises a SIMD architecture capable of performing two matrix-vector operations in parallel while fully exploiting parallelism in each operation. There is a system bus connecting the host processor to the plurality of SIMD microprocessors and a common clock providing a continuous sequence of clock pulses. There is also a ring structure interconnecting the plurality of SIMD microprocessors and connected to the clock for providing the clock pulses to the SIMD microprocessors and for providing a path for the flow of data and instructions between the SIMD microprocessors. The host processor includes logic for controlling the RRCS by interpreting instructions sent by the external host computer, decomposing the instructions into a series of computations to be performed by the SIMD microprocessors, using the system bus to distribute associated data among the SIMD microprocessors, and initiating activity of the SIMD microprocessors to perform the computations on the data by procedure call.
Parallel computing for simultaneous iterative tomographic imaging by graphics processing units
NASA Astrophysics Data System (ADS)
Bello-Maldonado, Pedro D.; López, Ricardo; Rogers, Colleen; Jin, Yuanwei; Lu, Enyue
2016-05-01
In this paper, we address the problem of accelerating inversion algorithms for nonlinear acoustic tomographic imaging by parallel computing on graphics processing units (GPUs). Nonlinear inversion algorithms for tomographic imaging often rely on iterative algorithms for solving an inverse problem, thus computationally intensive. We study the simultaneous iterative reconstruction technique (SIRT) for the multiple-input-multiple-output (MIMO) tomography algorithm which enables parallel computations of the grid points as well as the parallel execution of multiple source excitation. Using graphics processing units (GPUs) and the Compute Unified Device Architecture (CUDA) programming model an overall improvement of 26.33x was achieved when combining both approaches compared with sequential algorithms. Furthermore we propose an adaptive iterative relaxation factor and the use of non-uniform weights to improve the overall convergence of the algorithm. Using these techniques, fast computations can be performed in parallel without the loss of image quality during the reconstruction process.
A Lanczos eigenvalue method on a parallel computer
NASA Technical Reports Server (NTRS)
Bostic, Susan W.; Fulton, Robert E.
1987-01-01
Eigenvalue analyses of complex structures is a computationally intensive task which can benefit significantly from new and impending parallel computers. This study reports on a parallel computer implementation of the Lanczos method for free vibration analysis. The approach used here subdivides the major Lanczos calculation tasks into subtasks and introduces parallelism down to the subtask levels such as matrix decomposition and forward/backward substitution. The method was implemented on a commercial parallel computer and results were obtained for a long flexible space structure. While parallel computing efficiency for the Lanczos method was good for a moderate number of processors for the test problem, the greatest reduction in time was realized for the decomposition of the stiffness matrix, a calculation which took 70 percent of the time in the sequential program and which took 25 percent of the time on eight processors. For a sample calculation of the twenty lowest frequencies of a 486 degree of freedom problem, the total sequential computing time was reduced by almost a factor of ten using 16 processors.
History Matching in Parallel Computational Environments
Steven Bryant; Sanjay Srinivasan; Alvaro Barrera; Sharad Yadav
2005-10-01
A novel methodology for delineating multiple reservoir domains for the purpose of history matching in a distributed computing environment has been proposed. A fully probabilistic approach to perturb permeability within the delineated zones is implemented. The combination of robust schemes for identifying reservoir zones and distributed computing significantly increase the accuracy and efficiency of the probabilistic approach. The information pertaining to the permeability variations in the reservoir that is contained in dynamic data is calibrated in terms of a deformation parameter rD. This information is merged with the prior geologic information in order to generate permeability models consistent with the observed dynamic data as well as the prior geology. The relationship between dynamic response data and reservoir attributes may vary in different regions of the reservoir due to spatial variations in reservoir attributes, well configuration, flow constrains etc. The probabilistic approach then has to account for multiple r{sub D} values in different regions of the reservoir. In order to delineate reservoir domains that can be characterized with different rD parameters, principal component analysis (PCA) of the Hessian matrix has been done. The Hessian matrix summarizes the sensitivity of the objective function at a given step of the history matching to model parameters. It also measures the interaction of the parameters in affecting the objective function. The basic premise of PC analysis is to isolate the most sensitive and least correlated regions. The eigenvectors obtained during the PCA are suitably scaled and appropriate grid block volume cut-offs are defined such that the resultant domains are neither too large (which increases interactions between domains) nor too small (implying ineffective history matching). The delineation of domains requires calculation of Hessian, which could be computationally costly and as well as restricts the current approach to
New computing environments:Parallel, vector and systolic
Wouk, A.
1986-01-01
This book presents papers on supercomputers and array processors. Topics considered include nested dissection, the systolic level 2 BLAS, parallel processing a hydrodynamic shock wave problem, MACH-1, portable standard LISP on the Cray, distributed combinator evaluation, performance and library issues, scale problems, multiprocessor architecture, the MIDAS multiprocessor system, parallel algorithms for incompressible and compressible flows on a multiprocessor, and parallel algorithms for elliptic equations.
Seismic imaging using finite-differences and parallel computers
Ober, C.C.
1997-12-31
A key to reducing the risks and costs of associated with oil and gas exploration is the fast, accurate imaging of complex geologies, such as salt domes in the Gulf of Mexico and overthrust regions in US onshore regions. Prestack depth migration generally yields the most accurate images, and one approach to this is to solve the scalar wave equation using finite differences. As part of an ongoing ACTI project funded by the US Department of Energy, a finite difference, 3-D prestack, depth migration code has been developed. The goal of this work is to demonstrate that massively parallel computers can be used efficiently for seismic imaging, and that sufficient computing power exists (or soon will exist) to make finite difference, prestack, depth migration practical for oil and gas exploration. Several problems had to be addressed to get an efficient code for the Intel Paragon. These include efficient I/O, efficient parallel tridiagonal solves, and high single-node performance. Furthermore, to provide portable code the author has been restricted to the use of high-level programming languages (C and Fortran) and interprocessor communications using MPI. He has been using the SUNMOS operating system, which has affected many of his programming decisions. He will present images created from two verification datasets (the Marmousi Model and the SEG/EAEG 3D Salt Model). Also, he will show recent images from real datasets, and point out locations of improved imaging. Finally, he will discuss areas of current research which will hopefully improve the image quality and reduce computational costs.
Parallel computation of geometry control in adaptive truss structures
NASA Technical Reports Server (NTRS)
Ramesh, A. V.; Utku, S.; Wada, B. K.
1992-01-01
The fast computation of geometry control in adaptive truss structures involves two distinct parts: the efficient integration of the inverse kinematic differential equations that govern the geometry control and the fast computation of the Jacobian, which appears on the right-hand-side of the inverse kinematic equations. This paper present an efficient parallel implementation of the Jacobian computation on an MIMD machine. Large speedup from the parallel implementation is obtained, which reduces the Jacobian computation to an O(M-squared/n) procedure on an n-processor machine, where M is the number of members in the adaptive truss. The parallel algorithm given here is a good candidate for on-line geometry control of adaptive structures using attached processors.
Development of a massively parallel parachute performance prediction code
Peterson, C.W.; Strickland, J.H.; Wolfe, W.P.; Sundberg, W.D.; McBride, D.D.
1997-04-01
The Department of Energy has given Sandia full responsibility for the complete life cycle (cradle to grave) of all nuclear weapon parachutes. Sandia National Laboratories is initiating development of a complete numerical simulation of parachute performance, beginning with parachute deployment and continuing through inflation and steady state descent. The purpose of the parachute performance code is to predict the performance of stockpile weapon parachutes as these parachutes continue to age well beyond their intended service life. A new massively parallel computer will provide unprecedented speed and memory for solving this complex problem, and new software will be written to treat the coupled fluid, structure and trajectory calculations as part of a single code. Verification and validation experiments have been proposed to provide the necessary confidence in the computations.
Dynamic remapping decisions in multi-phase parallel computations
NASA Technical Reports Server (NTRS)
Nicol, D. M.; Reynolds, P. F., Jr.
1986-01-01
The effectiveness of any given mapping of workload to processors in a parallel system is dependent on the stochastic behavior of the workload. Program behavior is often characterized by a sequence of phases, with phase changes occurring unpredictably. During a phase, the behavior is fairly stable, but may become quite different during the next phase. Thus a workload assignment generated for one phase may hinder performance during the next phase. We consider the problem of deciding whether to remap a paralled computation in the face of uncertainty in remapping's utility. Fundamentally, it is necessary to balance the expected remapping performance gain against the delay cost of remapping. This paper treats this problem formally by constructing a probabilistic model of a computation with at most two phases. We use stochastic dynamic programming to show that the remapping decision policy which minimizes the expected running time of the computation has an extremely simple structure: the optimal decision at any step is followed by comparing the probability of remapping gain against a threshold. This theoretical result stresses the importance of detecting a phase change, and assessing the possibility of gain from remapping. We also empirically study the sensitivity of optimal performance to imprecise decision threshold. Under a wide range of model parameter values, we find nearly optimal performance if remapping is chosen simply when the gain probability is high. These results strongly suggest that except in extreme cases, the remapping decision problem is essentially that of dynamically determining whether gain can be achieved by remapping after a phase change; precise quantification of the decision model parameters is not necessary.
Modeling groundwater flow on massively parallel computers
Ashby, S.F.; Falgout, R.D.; Fogwell, T.W.; Tompson, A.F.B.
1994-12-31
The authors will explore the numerical simulation of groundwater flow in three-dimensional heterogeneous porous media. An interdisciplinary team of mathematicians, computer scientists, hydrologists, and environmental engineers is developing a sophisticated simulation code for use on workstation clusters and MPPs. To date, they have concentrated on modeling flow in the saturated zone (single phase), which requires the solution of a large linear system. they will discuss their implementation of preconditioned conjugate gradient solvers. The preconditioners under consideration include simple diagonal scaling, s-step Jacobi, adaptive Chebyshev polynomial preconditioning, and multigrid. They will present some preliminary numerical results, including simulations of groundwater flow at the LLNL site. They also will demonstrate the code`s scalability.
Performance of the Wavelet Decomposition on Massively Parallel Architectures
NASA Technical Reports Server (NTRS)
El-Ghazawi, Tarek A.; LeMoigne, Jacqueline; Zukor, Dorothy (Technical Monitor)
2001-01-01
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.
Computational mechanics analysis tools for parallel-vector supercomputers
NASA Technical Reports Server (NTRS)
Storaasli, O. O.; Nguyen, D. T.; Baddourah, M. A.; Qin, J.
1993-01-01
Computational algorithms for structural analysis on parallel-vector supercomputers are reviewed. These parallel algorithms, developed by the authors, are for the assembly of structural equations, 'out-of-core' strategies for linear equation solution, massively distributed-memory equation solution, unsymmetric equation solution, general eigen-solution, geometrically nonlinear finite element analysis, design sensitivity analysis for structural dynamics, optimization algorithm and domain decomposition. The source code for many of these algorithms is available from NASA Langley.
Computational mechanics analysis tools for parallel-vector supercomputers
NASA Technical Reports Server (NTRS)
Storaasli, Olaf O.; Nguyen, Duc T.; Baddourah, Majdi; Qin, Jiangning
1993-01-01
Computational algorithms for structural analysis on parallel-vector supercomputers are reviewed. These parallel algorithms, developed by the authors, are for the assembly of structural equations, 'out-of-core' strategies for linear equation solution, massively distributed-memory equation solution, unsymmetric equation solution, general eigensolution, geometrically nonlinear finite element analysis, design sensitivity analysis for structural dynamics, optimization search analysis and domain decomposition. The source code for many of these algorithms is available.
Parallel and pipeline computation of fast unitary transforms
NASA Technical Reports Server (NTRS)
Fino, B. J.; Algazi, V. R.
1975-01-01
The letter discusses the parallel and pipeline organization of fast-unitary-transform algorithms such as the fast Fourier transform, and points out the efficiency of a combined parallel-pipeline processor of a transform such as the Haar transform, in which (2 to the n-th power) -1 hardware 'butterflies' generate a transform of order 2 to the n-th power every computation cycle.
Dynamic grid refinement for partial differential equations on parallel computers
NASA Technical Reports Server (NTRS)
Mccormick, S.; Quinlan, D.
1989-01-01
The fast adaptive composite grid method (FAC) is an algorithm that uses various levels of uniform grids to provide adaptive resolution and fast solution of PDEs. An asynchronous version of FAC, called AFAC, that completely eliminates the bottleneck to parallelism is presented. This paper describes the advantage that this algorithm has in adaptive refinement for moving singularities on multiprocessor computers. This work is applicable to the parallel solution of two- and three-dimensional shock tracking problems.
Parallel algorithm for computing 3-D reachable workspaces
NASA Astrophysics Data System (ADS)
Alameldin, Tarek K.; Sobh, Tarek M.
1992-03-01
The problem of computing the 3-D workspace for redundant articulated chains has applications in a variety of fields such as robotics, computer aided design, and computer graphics. The computational complexity of the workspace problem is at least NP-hard. The recent advent of parallel computers has made practical solutions for the workspace problem possible. Parallel algorithms for computing the 3-D workspace for redundant articulated chains with joint limits are presented. The first phase of these algorithms computes workspace points in parallel. The second phase uses workspace points that are computed in the first phase and fits a 3-D surface around the volume that encompasses the workspace points. The second phase also maps the 3- D points into slices, uses region filling to detect the holes and voids in the workspace, extracts the workspace boundary points by testing the neighboring cells, and tiles the consecutive contours with triangles. The proposed algorithms are efficient for computing the 3-D reachable workspace for articulated linkages, not only those with redundant degrees of freedom but also those with joint limits.
RAMA: A file system for massively parallel computers
NASA Technical Reports Server (NTRS)
Miller, Ethan L.; Katz, Randy H.
1993-01-01
This paper describes a file system design for massively parallel computers which makes very efficient use of a few disks per processor. This overcomes the traditional I/O bottleneck of massively parallel machines by storing the data on disks within the high-speed interconnection network. In addition, the file system, called RAMA, requires little inter-node synchronization, removing another common bottleneck in parallel processor file systems. Support for a large tertiary storage system can easily be integrated in lo the file system; in fact, RAMA runs most efficiently when tertiary storage is used.
Parallel language constructs for tensor product computations on loosely coupled architectures
NASA Technical Reports Server (NTRS)
Mehrotra, Piyush; Vanrosendale, John
1989-01-01
Distributed memory architectures offer high levels of performance and flexibility, but have proven awkard to program. Current languages for nonshared memory architectures provide a relatively low level programming environment, and are poorly suited to modular programming, and to the construction of libraries. A set of language primitives designed to allow the specification of parallel numerical algorithms at a higher level is described. Tensor product array computations are focused on along with a simple but important class of numerical algorithms. The problem of programming 1-D kernal routines is focused on first, such as parallel tridiagonal solvers, and then how such parallel kernels can be combined to form parallel tensor product algorithms is examined.
Multithreaded Model for Dynamic Load Balancing Parallel Adaptive PDE Computations
NASA Technical Reports Server (NTRS)
Chrisochoides, Nikos
1995-01-01
We present a multithreaded model for the dynamic load-balancing of numerical, adaptive computations required for the solution of Partial Differential Equations (PDE's) on multiprocessors. Multithreading is used as a means of exploring concurrency in the processor level in order to tolerate synchronization costs inherent to traditional (non-threaded) parallel adaptive PDE solvers. Our preliminary analysis for parallel, adaptive PDE solvers indicates that multithreading can be used an a mechanism to mask overheads required for the dynamic balancing of processor workloads with computations required for the actual numerical solution of the PDE's. Also, multithreading can simplify the implementation of dynamic load-balancing algorithms, a task that is very difficult for traditional data parallel adaptive PDE computations. Unfortunately, multithreading does not always simplify program complexity, often makes code re-usability not an easy task, and increases software complexity.
Parallel grid generation algorithm for distributed memory computers
NASA Technical Reports Server (NTRS)
Moitra, Stuti; Moitra, Anutosh
1994-01-01
A parallel grid-generation algorithm and its implementation on the Intel iPSC/860 computer are described. The grid-generation scheme is based on an algebraic formulation of homotopic relations. Methods for utilizing the inherent parallelism of the grid-generation scheme are described, and implementation of multiple levELs of parallelism on multiple instruction multiple data machines are indicated. The algorithm is capable of providing near orthogonality and spacing control at solid boundaries while requiring minimal interprocessor communications. Results obtained on the Intel hypercube for a blended wing-body configuration are used to demonstrate the effectiveness of the algorithm. Fortran implementations bAsed on the native programming model of the iPSC/860 computer and the Express system of software tools are reported. Computational gains in execution time speed-up ratios are given.
Pratt, T.J.; Martinez, L.G.; Vahle, M.O.; Archuleta, T.V.; Williams, V.K.
1998-05-01
The advanced networking department at Sandia National Laboratories has used the annual Supercomputing conference sponsored by the IEEE and ACM for the past several years as a forum to demonstrate and focus communication and networking developments. At SC `97, Sandia National Laboratories (SNL), Los Alamos National Laboratory (LANL), and Lawrence Livermore National Laboratory (LLNL) combined their SC `97 activities within a single research booth under the Advance Strategic Computing Initiative (ASCI) banner. For the second year in a row, Sandia provided the network design and coordinated the networking activities within the booth. At SC `97, Sandia elected to demonstrate the capability of the Computation Plant, the visualization of scientific data, scalable ATM encryption, and ATM video and telephony capabilities. At SC `97, LLNL demonstrated an application, called RIPTIDE, that also required significant networking resources. The RIPTIDE application had computational visualization and steering capabilities. This paper documents those accomplishments, discusses the details of their implementation, and describes how these demonstrations support Sandia`s overall strategies in ATM networking.
The performance realities of massively parallel processors: A case study
Lubeck, O.M.; Simmons, M.L.; Wasserman, H.J.
1992-07-01
This paper presents the results of an architectural comparison of SIMD massive parallelism, as implemented in the Thinking Machines Corp. CM-2 computer, and vector or concurrent-vector processing, as implemented in the Cray Research Inc. Y-MP/8. The comparison is based primarily upon three application codes that represent Los Alamos production computing. Tests were run by porting optimized CM Fortran codes to the Y-MP, so that the same level of optimization was obtained on both machines. The results for fully-configured systems, using measured data rather than scaled data from smaller configurations, show that the Y-MP/8 is faster than the 64k CM-2 for all three codes. A simple model that accounts for the relative characteristic computational speeds of the two machines, and reduction in overall CM-2 performance due to communication or SIMD conditional execution, is included. The model predicts the performance of two codes well, but fails for the third code, because the proportion of communications in this code is very high. Other factors, such as memory bandwidth and compiler effects, are also discussed. Finally, the paper attempts to show the equivalence of the CM-2 and Y-MP programming models, and also comments on selected future massively parallel processor designs.
Implicit schemes and parallel computing in unstructured grid CFD
NASA Technical Reports Server (NTRS)
Venkatakrishnam, V.
1995-01-01
The development of implicit schemes for obtaining steady state solutions to the Euler and Navier-Stokes equations on unstructured grids is outlined. Applications are presented that compare the convergence characteristics of various implicit methods. Next, the development of explicit and implicit schemes to compute unsteady flows on unstructured grids is discussed. Next, the issues involved in parallelizing finite volume schemes on unstructured meshes in an MIMD (multiple instruction/multiple data stream) fashion are outlined. Techniques for partitioning unstructured grids among processors and for extracting parallelism in explicit and implicit solvers are discussed. Finally, some dynamic load balancing ideas, which are useful in adaptive transient computations, are presented.
Small file aggregation in a parallel computing system
Faibish, Sorin; Bent, John M.; Tzelnic, Percy; Grider, Gary; Zhang, Jingwang
2014-09-02
Techniques are provided for small file aggregation in a parallel computing system. An exemplary method for storing a plurality of files generated by a plurality of processes in a parallel computing system comprises aggregating the plurality of files into a single aggregated file; and generating metadata for the single aggregated file. The metadata comprises an offset and a length of each of the plurality of files in the single aggregated file. The metadata can be used to unpack one or more of the files from the single aggregated file.
Computational acceleration for MR image reconstruction in partially parallel imaging.
Ye, Xiaojing; Chen, Yunmei; Huang, Feng
2011-05-01
In this paper, we present a fast numerical algorithm for solving total variation and l(1) (TVL1) based image reconstruction with application in partially parallel magnetic resonance imaging. Our algorithm uses variable splitting method to reduce computational cost. Moreover, the Barzilai-Borwein step size selection method is adopted in our algorithm for much faster convergence. Experimental results on clinical partially parallel imaging data demonstrate that the proposed algorithm requires much fewer iterations and/or less computational cost than recently developed operator splitting and Bregman operator splitting methods, which can deal with a general sensing matrix in reconstruction framework, to get similar or even better quality of reconstructed images. PMID:20833599
Monte Carlo simulations of converging laser beam propagating in turbid media with parallel computing
NASA Astrophysics Data System (ADS)
Wu, Di; Lu, Jun Q.; Hu, Xin H.; Zhao, S. S.
1999-11-01
Due to its flexibility and simplicity, Monte Carlo method is often used to study light propagation in turbid medium where the photons are treated like classic particles being scattered and absorbed randomly based on a radiative transfer theory. However, due to the need of large number of photons to produce statistically significance results, this type of calculations requires large computing resources. To overcome such difficulty, we implemented parallel computing technique into our Monte Carlo simulations. The algorithm is based on the fact that the classic particles are uncorrelated, and the trajectories of multiple photons can be tracked simultaneously. When a beam of focused light incident to the medium, the incident photons are divided into groups according to the available processes on a parallel machine and the calculations are carried out in parallel. Utilizing PVM (Parallel Virtual Machine, a parallel computing software), the parallel programs in both C and FORTRAN are developed on the massive parallel computer Cray T3E at the North Carolina Supercomputer Center and a local PC-cluster network running UNIX/Sun Solaris. The parallel performances of our codes have been excellent on both Cray T3E and the PC clusters. In this paper, we present results on a focusing laser beam propagating through a highly scattering and diluted solution of intralipid. The dependence of the spatial distribution of light near the focal point on the concentration of intralipid solution is studied and its significance is discussed.
Archer, Charles J.; Blocksome, Michael A.; Ratterman, Joseph D.; Smith, Brian E.
2016-03-15
Processing data communications events in a parallel active messaging interface (`PAMI`) of a parallel computer that includes compute nodes that execute a parallel application, with the PAMI including data communications endpoints, and the endpoints are coupled for data communications through the PAMI and through other data communications resources, including determining by an advance function that there are no actionable data communications events pending for its context, placing by the advance function its thread of execution into a wait state, waiting for a subsequent data communications event for the context; responsive to occurrence of a subsequent data communications event for the context, awakening by the thread from the wait state; and processing by the advance function the subsequent data communications event now pending for the context.
Requirements for supercomputing in energy research: The transition to massively parallel computing
Not Available
1993-02-01
This report discusses: The emergence of a practical path to TeraFlop computing and beyond; requirements of energy research programs at DOE; implementation: supercomputer production computing environment on massively parallel computers; and implementation: user transition to massively parallel computing.
Routing Application for Parallel computatIon of Discharge
NASA Astrophysics Data System (ADS)
David, C. H.; Maidment, D. R.; Yang, Z.
2008-12-01
Today, meteorological models can predict storm events and climate patterns, but there are few models that connect atmospheric models to the hydraulics of rivers. Such a model is necessary to advance the prediction of events such as floods or droughts. Furthermore, the increasingly available Geographic Information System-based hydrographic datasets offer ways to use actual mapped river for routing. Such GIS-based datasets include NHDPlus at the continental-scale [USEPA and USGS, 2007]and HydroSHEDS at the global- scale [Lehner, et al., 2006]. The objective of ongoing work is to develop RAPID (Routing Application for Parallel computatIon of Discharge), a large-scale river routing model that: - has physical representation of river flow - allows for coupling with both land surface models and groundwater models. In particular enabling bi-directional exchanges between rivers and aquifers through a computation of water volume and flow on a reach-to-reach basis - has some specific treatment for man-made infrastructures and anthropogenic actions (dams, pumping) - benefits from the latest scientific computing developments (supercomputing facilities and high performance parallel computing libraries) - will benefit from the increasingly available Geographic Information System hydrographic databases. RAPID has already been tested over France through a ten-year coupling with SIM-France [Habets, et al., 2008, using a river network made of 24,264 river reaches. RAPID has been adapted to run on the Lonestar supercomputer (http://www.tacc.utexas.edu/resources/hpcsystems/) and to allow the use of the NHDPlus dataset. RAPID is now being tested with the 74,615 river reaches of the entire Texas Gulf. This type of innovative model has strong implications for the future of hydrology. Such a tool can help improve the understanding of the effect of climate change on water resources as well as provide information on how many gages are needed and where are they needed the most. More broadly
Parallel Computing for the Computed-Tomography Imaging Spectrometer
NASA Technical Reports Server (NTRS)
Lee, Seungwon
2008-01-01
This software computes the tomographic reconstruction of spatial-spectral data from raw detector images of the Computed-Tomography Imaging Spectrometer (CTIS), which enables transient-level, multi-spectral imaging by capturing spatial and spectral information in a single snapshot.
A Programming Model Performance Study Using the NAS Parallel Benchmarks
Shan, Hongzhang; Blagojević, Filip; Min, Seung-Jai; Hargrove, Paul; Jin, Haoqiang; Fuerlinger, Karl; Koniges, Alice; Wright, Nicholas J.
2010-01-01
Harnessing the power of multicore platforms is challenging due to the additional levels of parallelism present. In this paper we use the NAS Parallel Benchmarks to study three programming models, MPI, OpenMP and PGAS to understand their performance and memory usage characteristics on current multicore architectures. To understand these characteristics we use the Integrated Performance Monitoring tool and other ways to measure communication versus computation time, as well as the fraction of the run time spent in OpenMP. The benchmarks are run on two different Cray XT5 systems and an Infiniband cluster. Our results show that in general the threemore » programming models exhibit very similar performance characteristics. In a few cases, OpenMP is significantly faster because it explicitly avoids communication. For these particular cases, we were able to re-write the UPC versions and achieve equal performance to OpenMP. Using OpenMP was also the most advantageous in terms of memory usage. Also we compare performance differences between the two Cray systems, which have quad-core and hex-core processors. We show that at scale the performance is almost always slower on the hex-core system because of increased contention for network resources.« less
A Testbed of Parallel Kernels for Computer Science Research
Bailey, David; Demmel, James; Ibrahim, Khaled; Kaiser, Alex; Koniges, Alice; Madduri, Kamesh; Shalf, John; Strohmaier, Erich; Williams, Samuel
2010-04-30
For several decades, computer scientists have sought guidance on how to evolve architectures, languages, and programming models for optimal performance, efficiency, and productivity. Unfortunately, this guidance is most often taken from the existing software/hardware ecosystem. Architects attempt to provide micro-architectural solutions to improve performance on fixed binaries. Researchers tweak compilers to improve code generation for existing architectures and implementations, and they may invent new programming models for fixed processor and memory architectures and computational algorithms. In today's rapidly evolving world of on-chip parallelism, these isolated and iterative improvements to performance may miss superior solutions in the same way gradient descent optimization techniques may get stuck in local minima. In an initial study, we have developed an alternate approach that, rather than starting with an existing hardware/software solution laced with hidden assumptions, defines the computational problems of interest and invites architects, researchers and programmers to implement novel hardware/ software co-designed solutions. Our work builds on the previous ideas of computational dwarfs, motifs, and parallel patterns by selecting a representative set of essential problems for which we provide: An algorithmic description; scalable problem definition; illustrative reference implementations; verification schemes. For simplicity, we focus initially on the computational problems of interest to the scientific computing community but proclaim the methodology (and perhaps a subset of the problems) as applicable to other communities. We intend to broaden the coverage of this problem space through stronger community involvement. Previous work has established a broad categorization of numerical methods of interest to the scientific computing, in the spirit of the NAS Benchmarks, which pioneered the basic idea of a 'pencil and paper benchmark' in the 1990s. The
Joubert, W.; Carey, G.F.
1994-12-31
A great need exists for high performance numerical software libraries transportable across parallel machines. This talk concerns the PCG package, which solves systems of linear equations by iterative methods on parallel computers. The features of the package are discussed, as well as techniques used to obtain high performance as well as transportability across architectures. Representative numerical results are presented for several machines including the Connection Machine CM-5, Intel Paragon and Cray T3D parallel computers.
Solution of partial differential equations on vector and parallel computers
NASA Technical Reports Server (NTRS)
Ortega, J. M.; Voigt, R. G.
1985-01-01
The present status of numerical methods for partial differential equations on vector and parallel computers was reviewed. The relevant aspects of these computers are discussed and a brief review of their development is included, with particular attention paid to those characteristics that influence algorithm selection. Both direct and iterative methods are given for elliptic equations as well as explicit and implicit methods for initial boundary value problems. The intent is to point out attractive methods as well as areas where this class of computer architecture cannot be fully utilized because of either hardware restrictions or the lack of adequate algorithms. Application areas utilizing these computers are briefly discussed.
A new parallel-vector finite element analysis software on distributed-memory computers
NASA Technical Reports Server (NTRS)
Qin, Jiangning; Nguyen, Duc T.
1993-01-01
A new parallel-vector finite element analysis software package MPFEA (Massively Parallel-vector Finite Element Analysis) is developed for large-scale structural analysis on massively parallel computers with distributed-memory. MPFEA is designed for parallel generation and assembly of the global finite element stiffness matrices as well as parallel solution of the simultaneous linear equations, since these are often the major time-consuming parts of a finite element analysis. Block-skyline storage scheme along with vector-unrolling techniques are used to enhance the vector performance. Communications among processors are carried out concurrently with arithmetic operations to reduce the total execution time. Numerical results on the Intel iPSC/860 computers (such as the Intel Gamma with 128 processors and the Intel Touchstone Delta with 512 processors) are presented, including an aircraft structure and some very large truss structures, to demonstrate the efficiency and accuracy of MPFEA.
Solving very large, sparse linear systems on mesh-connected parallel computers
NASA Technical Reports Server (NTRS)
Opsahl, Torstein; Reif, John
1987-01-01
The implementation of Pan and Reif's Parallel Nested Dissection (PND) algorithm on mesh connected parallel computers is described. This is the first known algorithm that allows very large, sparse linear systems of equations to be solved efficiently in polylog time using a small number of processors. How the processor bound of PND can be matched to the number of processors available on a given parallel computer by slowing down the algorithm by constant factors is described. Also, for the important class of problems where G(A) is a grid graph, a unique memory mapping that reduces the inter-processor communication requirements of PND to those that can be executed on mesh connected parallel machines is detailed. A description of an implementation on the Goodyear Massively Parallel Processor (MPP), located at Goddard is given. Also, a detailed discussion of data mappings and performance issues is given.
Parallel algorithms for computation of the manipulator inertia matrix
NASA Technical Reports Server (NTRS)
Amin-Javaheri, Masoud; Orin, David E.
1989-01-01
The development of an O(log2N) parallel algorithm for the manipulator inertia matrix is presented. It is based on the most efficient serial algorithm which uses the composite rigid body method. Recursive doubling is used to reformulate the linear recurrence equations which are required to compute the diagonal elements of the matrix. It results in O(log2N) levels of computation. Computation of the off-diagonal elements involves N linear recurrences of varying-size and a new method, which avoids redundant computation of position and orientation transforms for the manipulator, is developed. The O(log2N) algorithm is presented in both equation and graphic forms which clearly show the parallelism inherent in the algorithm.
Analysis of composite ablators using massively parallel computation
NASA Technical Reports Server (NTRS)
Shia, David
1995-01-01
In this work, the feasibility of using massively parallel computation to study the response of ablative materials is investigated. Explicit and implicit finite difference methods are used on a massively parallel computer, the Thinking Machines CM-5. The governing equations are a set of nonlinear partial differential equations. The governing equations are developed for three sample problems: (1) transpiration cooling, (2) ablative composite plate, and (3) restrained thermal growth testing. The transpiration cooling problem is solved using a solution scheme based solely on the explicit finite difference method. The results are compared with available analytical steady-state through-thickness temperature and pressure distributions and good agreement between the numerical and analytical solutions is found. It is also found that a solution scheme based on the explicit finite difference method has the following advantages: incorporates complex physics easily, results in a simple algorithm, and is easily parallelizable. However, a solution scheme of this kind needs very small time steps to maintain stability. A solution scheme based on the implicit finite difference method has the advantage that it does not require very small times steps to maintain stability. However, this kind of solution scheme has the disadvantages that complex physics cannot be easily incorporated into the algorithm and that the solution scheme is difficult to parallelize. A hybrid solution scheme is then developed to combine the strengths of the explicit and implicit finite difference methods and minimize their weaknesses. This is achieved by identifying the critical time scale associated with the governing equations and applying the appropriate finite difference method according to this critical time scale. The hybrid solution scheme is then applied to the ablative composite plate and restrained thermal growth problems. The gas storage term is included in the explicit pressure calculation of both
Hardware packet pacing using a DMA in a parallel computer
Chen, Dong; Heidelberger, Phillip; Vranas, Pavlos
2013-08-13
Method and system for hardware packet pacing using a direct memory access controller in a parallel computer which, in one aspect, keeps track of a total number of bytes put on the network as a result of a remote get operation, using a hardware token counter.
PCCM2: A GCM adapted for scalable parallel computers
Drake, J.; Semeraro, B.D.; Worley, P.; Foster, I.; Michalakes, J.; Toonen, B.; Hack, J.J.; Williamson, D.L.
1994-01-01
The Computer Hardware, Advanced Mathematics and Model Physics (CHAMMP) program seeks to provide climate researchers with an advanced modeling capability for the study of global change issues. One of the more ambitious projects being undertaken in the CHAMMP program is the development of PCCM2, an adaptation of the Community Climate Model (CCM2) for scalable parallel computers. PCCM2 uses a message-passing, domain-decomposition approach, in which each processor is allocated responsibility for computation on one part of the computational grid, and messages are generated to communicate data between processors. Much of the research effort associated with development of a parallel code of this sort is concerned with identifying efficient decomposition and communication strategies. In PCCM2, this task is complicated by the need to support both semi-Lagrangian transport and spectral transport. Load balancing and parallel I/O techniques are also required. In this paper, the authors review the various parallel algorithms used in PCCM2 and the work done to arrive at a validated model.
Efficiently modeling neural networks on massively parallel computers
Farber, R.M.
1992-01-01
Neural networks are a very useful tool for analyzing and modeling complex real world systems. Applying neural network simulations to real world problems generally involves large amounts of data and massive amounts of computation. To efficiently handle the computational requirements of large problems, we have implemented at Los Alamos a highly efficient neural network compiler for serial computers, vector computers, vector parallel computers, and fine grain SIMD computers such as the CM-2 connection machine. This paper will describe the mapping used by the compiler to implement feed-forward backpropagation neural networks for a SIMD architecture parallel computer. Thinking Machines Corporation has benchmarked our code at 1.3 billion interconnects per second (approximately 3 gigaflops) on a 64,000 processor CM-2 connection machine (Singer 1990). This mapping is applicable to other SMM computers and can be implemented on computers such as the CM-5 connection machine. Our mapping has virtually no communications overhead with the exception of the communications required for a global summation across the processors. We can efficiently model very large neural networks which have many neurons and interconnects and our mapping can be extend to arbitrarily large networks by merging the memory space of separate processors with fast adjacent processor inter-processor communications. This paper will consider the simulation of only feed forward neural network although this method is extendible to recurrent networks.
Efficiently modeling neural networks on massively parallel computers
Farber, R.M.
1992-12-01
Neural networks are a very useful tool for analyzing and modeling complex real world systems. Applying neural network simulations to real world problems generally involves large amounts of data and massive amounts of computation. To efficiently handle the computational requirements of large problems, we have implemented at Los Alamos a highly efficient neural network compiler for serial computers, vector computers, vector parallel computers, and fine grain SIMD computers such as the CM-2 connection machine. This paper will describe the mapping used by the compiler to implement feed-forward backpropagation neural networks for a SIMD architecture parallel computer. Thinking Machines Corporation has benchmarked our code at 1.3 billion interconnects per second (approximately 3 gigaflops) on a 64,000 processor CM-2 connection machine (Singer 1990). This mapping is applicable to other SMM computers and can be implemented on computers such as the CM-5 connection machine. Our mapping has virtually no communications overhead with the exception of the communications required for a global summation across the processors. We can efficiently model very large neural networks which have many neurons and interconnects and our mapping can be extend to arbitrarily large networks by merging the memory space of separate processors with fast adjacent processor inter-processor communications. This paper will consider the simulation of only feed forward neural network although this method is extendible to recurrent networks.
Efficiently modeling neural networks on massively parallel computers
NASA Technical Reports Server (NTRS)
Farber, Robert M.
1993-01-01
Neural networks are a very useful tool for analyzing and modeling complex real world systems. Applying neural network simulations to real world problems generally involves large amounts of data and massive amounts of computation. To efficiently handle the computational requirements of large problems, we have implemented at Los Alamos a highly efficient neural network compiler for serial computers, vector computers, vector parallel computers, and fine grain SIMD computers such as the CM-2 connection machine. This paper describes the mapping used by the compiler to implement feed-forward backpropagation neural networks for a SIMD (Single Instruction Multiple Data) architecture parallel computer. Thinking Machines Corporation has benchmarked our code at 1.3 billion interconnects per second (approximately 3 gigaflops) on a 64,000 processor CM-2 connection machine (Singer 1990). This mapping is applicable to other SIMD computers and can be implemented on MIMD computers such as the CM-5 connection machine. Our mapping has virtually no communications overhead with the exception of the communications required for a global summation across the processors (which has a sub-linear runtime growth on the order of O(log(number of processors)). We can efficiently model very large neural networks which have many neurons and interconnects and our mapping can extend to arbitrarily large networks (within memory limitations) by merging the memory space of separate processors with fast adjacent processor interprocessor communications. This paper will consider the simulation of only feed forward neural network although this method is extendable to recurrent networks.
Climate Modeling using High-Performance Computing
Mirin, A A; Wickett, M E; Duffy, P B; Rotman, D A
2005-03-03
The Center for Applied Scientific Computing (CASC) and the LLNL Atmospheric Science Division (ASD) are working together to improve predictions of future climate by applying the best available computational methods and computer resources to this problem. Over the last decade, researchers at the Lawrence Livermore National Laboratory (LLNL) have developed a number of climate models that provide state-of-the-art simulations on a wide variety of massively parallel computers. We are now developing and applying a second generation of high-performance climate models. As part of LLNL's participation in DOE's Scientific Discovery through Advanced Computing (SciDAC) program, members of CASC and ASD are collaborating with other DOE labs and NCAR in the development of a comprehensive, next-generation global climate model. This model incorporates the most current physics and numerics and capably exploits the latest massively parallel computers. One of LLNL's roles in this collaboration is the scalable parallelization of NASA's finite-volume atmospheric dynamical core. We have implemented multiple two-dimensional domain decompositions, where the different decompositions are connected by high-speed transposes. Additional performance is obtained through shared memory parallelization constructs and one-sided interprocess communication. The finite-volume dynamical core is particularly important to atmospheric chemistry simulations, where LLNL has a leading role.
Fencing data transfers in a parallel active messaging interface of a parallel computer
Blocksome, Michael A.; Mamidala, Amith R.
2015-06-09
Fencing data transfers in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI including data communications endpoints, each endpoint including a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task; the compute nodes coupled for data communications through the PAMI and through data communications resources including at least one segment of shared random access memory; including initiating execution through the PAMI of an ordered sequence of active SEND instructions for SEND data transfers between two endpoints, effecting deterministic SEND data transfers through a segment of shared memory; and executing through the PAMI, with no FENCE accounting for SEND data transfers, an active FENCE instruction, the FENCE instruction completing execution only after completion of all SEND instructions initiated prior to execution of the FENCE instruction for SEND data transfers between the two endpoints.
Fencing data transfers in a parallel active messaging interface of a parallel computer
Blocksome, Michael A.; Mamidala, Amith R.
2015-06-30
Fencing data transfers in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI including data communications endpoints, each endpoint comprising a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, the compute nodes coupled for data communications through the PAMI and through data communications resources including a deterministic data communications network, including initiating execution through the PAMI of an ordered sequence of active SEND instructions for SEND data transfers between two endpoints, effecting deterministic SEND data transfers; and executing through the PAMI, with no FENCE accounting for SEND data transfers, an active FENCE instruction, the FENCE instruction completing execution only after completion of all SEND instructions initiated prior to execution of the FENCE instruction for SEND data transfers between the two endpoints.
Fencing data transfers in a parallel active messaging interface of a parallel computer
Blocksome, Michael A.; Mamidala, Amith R.
2015-08-11
Fencing data transfers in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI including data communications endpoints, each endpoint comprising a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, the compute nodes coupled for data communications through the PAMI and through data communications resources including a deterministic data communications network, including initiating execution through the PAMI of an ordered sequence of active SEND instructions for SEND data transfers between two endpoints, effecting deterministic SEND data transfers; and executing through the PAMI, with no FENCE accounting for SEND data transfers, an active FENCE instruction, the FENCE instruction completing execution only after completion of all SEND instructions initiated prior to execution of the FENCE instruction for SEND data transfers between the two endpoints.
Fencing data transfers in a parallel active messaging interface of a parallel computer
Blocksome, Michael A.; Mamidala, Amith R.
2015-06-02
Fencing data transfers in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI including data communications endpoints, each endpoint including a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task; the compute nodes coupled for data communications through the PAMI and through data communications resources including at least one segment of shared random access memory; including initiating execution through the PAMI of an ordered sequence of active SEND instructions for SEND data transfers between two endpoints, effecting deterministic SEND data transfers through a segment of shared memory; and executing through the PAMI, with no FENCE accounting for SEND data transfers, an active FENCE instruction, the FENCE instruction completing execution only after completion of all SEND instructions initiated prior to execution of the FENCE instruction for SEND data transfers between the two endpoints.
Parallel multiphysics algorithms and software for computational nuclear engineering
NASA Astrophysics Data System (ADS)
Gaston, D.; Hansen, G.; Kadioglu, S.; Knoll, D. A.; Newman, C.; Park, H.; Permann, C.; Taitano, W.
2009-07-01
There is a growing trend in nuclear reactor simulation to consider multiphysics problems. This can be seen in reactor analysis where analysts are interested in coupled flow, heat transfer and neutronics, and in fuel performance simulation where analysts are interested in thermomechanics with contact coupled to species transport and chemistry. These more ambitious simulations usually motivate some level of parallel computing. Many of the coupling efforts to date utilize simple code coupling or first-order operator splitting, often referred to as loose coupling. While these approaches can produce answers, they usually leave questions of accuracy and stability unanswered. Additionally, the different physics often reside on separate grids which are coupled via simple interpolation, again leaving open questions of stability and accuracy. Utilizing state of the art mathematics and software development techniques we are deploying next generation tools for nuclear engineering applications. The Jacobian-free Newton-Krylov (JFNK) method combined with physics-based preconditioning provide the underlying mathematical structure for our tools. JFNK is understood to be a modern multiphysics algorithm, but we are also utilizing its unique properties as a scale bridging algorithm. To facilitate rapid development of multiphysics applications we have developed the Multiphysics Object-Oriented Simulation Environment (MOOSE). Examples from two MOOSE-based applications: PRONGHORN, our multiphysics gas cooled reactor simulation tool and BISON, our multiphysics, multiscale fuel performance simulation tool will be presented.
High Performance Parallel Methods for Space Weather Simulations
NASA Technical Reports Server (NTRS)
Hunter, Paul (Technical Monitor); Gombosi, Tamas I.
2003-01-01
This is the final report of our NASA AISRP grant entitled 'High Performance Parallel Methods for Space Weather Simulations'. The main thrust of the proposal was to achieve significant progress towards new high-performance methods which would greatly accelerate global MHD simulations and eventually make it possible to develop first-principles based space weather simulations which run much faster than real time. We are pleased to report that with the help of this award we made major progress in this direction and developed the first parallel implicit global MHD code with adaptive mesh refinement. The main limitation of all earlier global space physics MHD codes was the explicit time stepping algorithm. Explicit time steps are limited by the Courant-Friedrichs-Lewy (CFL) condition, which essentially ensures that no information travels more than a cell size during a time step. This condition represents a non-linear penalty for highly resolved calculations, since finer grid resolution (and consequently smaller computational cells) not only results in more computational cells, but also in smaller time steps.
Parallel performance optimizations on unstructured mesh-based simulations
Sarje, Abhinav; Song, Sukhyun; Jacobsen, Douglas; Huck, Kevin; Hollingsworth, Jeffrey; Malony, Allen; Williams, Samuel; Oliker, Leonid
2015-06-01
This paper addresses two key parallelization challenges the unstructured mesh-based ocean modeling code, MPAS-Ocean, which uses a mesh based on Voronoi tessellations: (1) load imbalance across processes, and (2) unstructured data access patterns, that inhibit intra- and inter-node performance. Our work analyzes the load imbalance due to naive partitioning of the mesh, and develops methods to generate mesh partitioning with better load balance and reduced communication. Furthermore, we present methods that minimize both inter- and intranode data movement and maximize data reuse. Our techniques include predictive ordering of data elements for higher cache efficiency, as well as communication reduction approaches. We present detailed performance data when running on thousands of cores using the Cray XC30 supercomputer and show that our optimization strategies can exceed the original performance by over 2×. Additionally, many of these solutions can be broadly applied to a wide variety of unstructured grid-based computations.
A Framework to Develop Symbolic Performance Models of Parallel Applications
Alam, Sadaf R; Vetter, Jeffrey S
2006-01-01
Performance and workload modeling has numerous uses at every stage of the high-end computing lifecycle: design, integration, procurement, installation and tuning. Despite the tremendous usefulness of performance models, their construction remains largely a manual, complex, and time-consuming exercise. We propose a new approach to the model construction, called modeling assertions (MA), which borrows advantages from both the empirical and analytical modeling techniques. This strategy has many advantages over traditional methods: incremental construction of realistic performance models, straightforward model validation against empirical data, and intuitive error bounding on individual model terms. We demonstrate this new technique on the NAS parallel CG and SP benchmarks by constructing high fidelity models for the floating-point operation cost, memory requirements, and MPI message volume. These models are driven by a small number of key input parameters thereby allowing efficient design space exploration of future problem sizes and architectures.
Performance Evaluation of Evasion Maneuvers for Parallel Approach Collision Avoidance
NASA Technical Reports Server (NTRS)
Winder, Lee F.; Kuchar, James K.; Waller, Marvin (Technical Monitor)
2000-01-01
Current plans for independent instrument approaches to closely spaced parallel runways call for an automated pilot alerting system to ensure separation of aircraft in the case of a "blunder," or unexpected deviation from the a normal approach path. Resolution advisories by this system would require the pilot of an endangered aircraft to perform a trained evasion maneuver. The potential performance of two evasion maneuvers, referred to as the "turn-climb" and "climb-only," was estimated using an experimental NASA alerting logic (AILS) and a computer simulation of relative trajectory scenarios between two aircraft. One aircraft was equipped with the NASA alerting system, and maneuvered accordingly. Observation of the rates of different types of alerting failure allowed judgement of evasion maneuver performance. System Operating Characteristic (SOC) curves were used to assess the benefit of alerting with each maneuver.
Improving efficiency of a global barrier operation in a parallel computer
None
2016-10-04
Performing a global barrier operation in a parallel computer that includes compute nodes coupled for data communications, where each compute node executes tasks, with one task on each compute node designated as a master task, including: for each task on each compute node until all master tasks have joined a global barrier: determining whether the task is a master task; if the task is not a master task, joining a single local barrier; if the task is a master task, joining the global barrier and the single local barrier only after all other tasks on the compute node have joined the single local barrier.
Locating and computing in parallel all the simple roots of special functions using PVM
NASA Astrophysics Data System (ADS)
Plagianakos, V. P.; Nousis, N. K.; Vrahatis, M. N.
2001-08-01
An algorithm is proposed for locating and computing in parallel and with certainty all the simple roots of any twice continuously differentiable function in any specific interval. To compute with certainty all the roots, the proposed method is heavily based on the knowledge of the total number of roots within the given interval. To obtain this information we use results from topological degree theory and, in particular, the Kronecker-Picard approach. This theory gives a formula for the computation of the total number of roots of a system of equations within a given region, which can be computed in parallel. With this tool in hand, we construct a parallel procedure for the localization and isolation of all the roots by dividing the given region successively and applying the above formula to these subregions until the final domains contain at the most one root. The subregions with no roots are discarded, while for the rest a modification of the well-known bisection method is employed for the computation of the contained root. The new aspect of the present contribution is that the computation of the total number of zeros using the Kronecker-Picard integral as well as the localization and computation of all the roots is performed in parallel using the parallel virtual machine (PVM). PVM is an integrated set of software tools and libraries that emulates a general-purpose, flexible, heterogeneous concurrent computing framework on interconnected computers of varied architectures. The proposed algorithm has large granularity and low synchronization, and is robust. It has been implemented and tested and our experience is that it can massively compute with certainty all the roots in a certain interval. Performance information from massive computations related to a recently proposed conjecture due to Elbert (this issue, J. Comput. Appl. Math. 133 (2001) 65-83) is reported.
Aggregating job exit statuses of a plurality of compute nodes executing a parallel application
Aho, Michael E.; Attinella, John E.; Gooding, Thomas M.; Mundy, Michael B.
2015-07-21
Aggregating job exit statuses of a plurality of compute nodes executing a parallel application, including: identifying a subset of compute nodes in the parallel computer to execute the parallel application; selecting one compute node in the subset of compute nodes in the parallel computer as a job leader compute node; initiating execution of the parallel application on the subset of compute nodes; receiving an exit status from each compute node in the subset of compute nodes, where the exit status for each compute node includes information describing execution of some portion of the parallel application by the compute node; aggregating each exit status from each compute node in the subset of compute nodes; and sending an aggregated exit status for the subset of compute nodes in the parallel computer.
Improved CDMA Performance Using Parallel Interference Cancellation
NASA Technical Reports Server (NTRS)
Simon, Marvin; Divsalar, Dariush
1995-01-01
This report considers a general parallel interference cancellation scheme that significantly reduces the degradation effect of user interference but with a lesser implementation complexity than the maximum-likelihood technique. The scheme operates on the fact that parallel processing simultaneously removes from each user the interference produced by the remaining users accessing the channel in an amount proportional to their reliability. The parallel processing can be done in multiple stages. The proposed scheme uses tentative decision devices with different optimum thresholds at the multiple stages to produce the most reliably received data for generation and cancellation of user interference. The 1-stage interference cancellation is analyzed for three types of tentative decision devices, namely, hard, null zone, and soft decision, and two types of user power distribution, namely, equal and unequal powers. Simulation results are given for a multitude of different situations, in particular, those cases for which the analysis is too complex.
Lober, R.R.; Tautges, T.J.; Vaughan, C.T.
1997-03-01
Paving is an automated mesh generation algorithm which produces all-quadrilateral elements. It can additionally generate these elements in varying sizes such that the resulting mesh adapts to a function distribution, such as an error function. While powerful, conventional paving is a very serial algorithm in its operation. Parallel paving is the extension of serial paving into parallel environments to perform the same meshing functions as conventional paving only on distributed, discretized models. This extension allows large, adaptive, parallel finite element simulations to take advantage of paving`s meshing capabilities for h-remap remeshing. A significantly modified version of the CUBIT mesh generation code has been developed to host the parallel paving algorithm and demonstrate its capabilities on both two dimensional and three dimensional surface geometries and compare the resulting parallel produced meshes to conventionally paved meshes for mesh quality and algorithm performance. Sandia`s {open_quotes}tiling{close_quotes} dynamic load balancing code has also been extended to work with the paving algorithm to retain parallel efficiency as subdomains undergo iterative mesh refinement.
A Simple Physical Optics Algorithm Perfect for Parallel Computing
NASA Technical Reports Server (NTRS)
Imbriale, W. A.; Cwik, T.
1993-01-01
One of the simplest reflector antenna computer programs is based upon a discrete approximation of the radiation integral. This calculation replaces the actual reflector surface with a triangular facet representation so that the reflector resembles a geodesic dome. The Physical Optics (PO) current is assumed to be constant in magnitude and phase over each facet so the radiation integral is reduced to a simple summation. This program has proven to be surprisingly robust and useful for the analysis of arbitrary reflectors, particularly when the near-field is desired and surface derivatives are not known. Because of its simplicity, the algorithm has proven to be extremely easy to adapt to the parallel computing architecture of a modest number of large-grain computing elements such as are used in the Intel iPSC and Touchstone Delta parallel machines.
A domain decomposition study of massively parallel computing in compressible gas dynamics
Wong, C.C.; Blottner, F.G.; Payne, J.L.; Soetrisno, M.
1995-01-01
The appropriate utilization of massively parallel computers for solving the Navier-Stokes equations is investigated and determined from an engineering perspective. The issues investigated are: (1) Should strip or patch domain decomposition of the spatial mesh be used to reduce computer time? (2) How many computer nodes should be used for a problem with a given sized mesh to reduce computer time? (3) Is the convergence of the Navier-Stokes solution procedure (LU-SGS) adversely influenced by the domain decomposition approach? The results of the paper show that the present Navier-Stokes solution technique has good performance on a massively parallel computer for transient flow problems. For steady-state problems with a large number of mesh cells, the solution procedure will require significant computer time due to an increased number of iterations to achieve a converged solution. There is an optimum number of computer nodes to use for a problem with a given global mesh size.
Computing NLTE Opacities -- Node Level Parallel
Holladay, Daniel
2015-09-11
Presentation. The goal: to produce a robust library capable of computing reasonably accurate opacities inline with the assumption of LTE relaxed (non-LTE). Near term: demonstrate acceleration of non-LTE opacity computation. Far term (if funded): connect to application codes with in-line capability and compute opacities. Study science problems. Use efficient algorithms that expose many levels of parallelism and utilize good memory access patterns for use on advanced architectures. Portability to multiple types of hardware including multicore processors, manycore processors such as KNL, GPUs, etc. Easily coupled to radiation hydrodynamics and thermal radiative transfer codes.
New Parallel computing framework for radiation transport codes
Kostin, M.A.; Mokhov, N.V.; Niita, K.; /JAERI, Tokai
2010-09-01
A new parallel computing framework has been developed to use with general-purpose radiation transport codes. The framework was implemented as a C++ module that uses MPI for message passing. The module is significantly independent of radiation transport codes it can be used with, and is connected to the codes by means of a number of interface functions. The framework was integrated with the MARS15 code, and an effort is under way to deploy it in PHITS. Besides the parallel computing functionality, the framework offers a checkpoint facility that allows restarting calculations with a saved checkpoint file. The checkpoint facility can be used in single process calculations as well as in the parallel regime. Several checkpoint files can be merged into one thus combining results of several calculations. The framework also corrects some of the known problems with the scheduling and load balancing found in the original implementations of the parallel computing functionality in MARS15 and PHITS. The framework can be used efficiently on homogeneous systems and networks of workstations, where the interference from the other users is possible.
Design and performance of a scalable, parallel statistics toolkit.
Thompson, David C.; Bennett, Janine Camille; Pebay, Philippe Pierre
2010-11-01
Most statistical software packages implement a broad range of techniques but do so in an ad hoc fashion, leaving users who do not have a broad knowledge of statistics at a disadvantage since they may not understand all the implications of a given analysis or how to test the validity of results. These packages are also largely serial in nature, or target multicore architectures instead of distributed-memory systems, or provide only a small number of statistics in parallel. This paper surveys a collection of parallel implementations of statistics algorithm developed as part of a common framework over the last 3 years. The framework strategically groups modeling techniques with associated verification and validation techniques to make the underlying assumptions of the statistics more clear. Furthermore it employs a design pattern specifically targeted for distributed-memory parallelism, where architectural advances in large-scale high-performance computing have been focused. Moment-based statistics (which include descriptive, correlative, and multicorrelative statistics, principal component analysis (PCA), and k-means statistics) scale nearly linearly with the data set size and number of processes. Entropy-based statistics (which include order and contingency statistics) do not scale well when the data in question is continuous or quasi-diffuse but do scale well when the data is discrete and compact. We confirm and extend our earlier results by now establishing near-optimal scalability with up to 10,000 processes.
Experiences with the Lanczos method on a parallel computer
NASA Technical Reports Server (NTRS)
Bostic, Susan W.; Fulton, Robert E.
1987-01-01
A parallel computer implementation of the Lanczos method for the free-vibration analysis of structures is considered, and results for two example problems show substantial time-reduction over the sequential solutions. The major Lanczos calculation tasks are subdivided into subtasks, and parallelism is introduced at the subtask level. A speedup of 7.8 on eight processors was obtained for the decomposition step of the problem involving a 60-m three-longeron space mast, and a speedup of 14.6 on 16 processors was obtained for the decomposition step of the problem involving a blade-stiffened graphite-epoxy panel.
The Design and Evaluation of "CAPTools"--A Computer Aided Parallelization Toolkit
NASA Technical Reports Server (NTRS)
Yan, Jerry; Frumkin, Michael; Hribar, Michelle; Jin, Haoqiang; Waheed, Abdul; Johnson, Steve; Cross, Jark; Evans, Emyr; Ierotheou, Constantinos; Leggett, Pete; Saini, Subhash (Technical Monitor)
1998-01-01
Writing applications for high performance computers is a challenging task. Although writing code by hand still offers the best performance, it is extremely costly and often not very portable. The Computer Aided Parallelization Tools (CAPTools) are a toolkit designed to help automate the mapping of sequential FORTRAN scientific applications onto multiprocessors. CAPTools consists of the following major components: an inter-procedural dependence analysis module that incorporates user knowledge; a 'self-propagating' data partitioning module driven via user guidance; an execution control mask generation and optimization module for the user to fine tune parallel processing of individual partitions; a program transformation/restructuring facility for source code clean up and optimization; a set of browsers through which the user interacts with CAPTools at each stage of the parallelization process; and a code generator supporting multiple programming paradigms on various multiprocessors. Besides describing the rationale behind the architecture of CAPTools, the parallelization process is illustrated via case studies involving structured and unstructured meshes. The programming process and the performance of the generated parallel programs are compared against other programming alternatives based on the NAS Parallel Benchmarks, ARC3D and other scientific applications. Based on these results, a discussion on the feasibility of constructing architectural independent parallel applications is presented.
Climate Modeling using High-Performance Computing
Mirin, A A
2007-02-05
The Center for Applied Scientific Computing (CASC) and the LLNL Climate and Carbon Science Group of Energy and Environment (E and E) are working together to improve predictions of future climate by applying the best available computational methods and computer resources to this problem. Over the last decade, researchers at the Lawrence Livermore National Laboratory (LLNL) have developed a number of climate models that provide state-of-the-art simulations on a wide variety of massively parallel computers. We are now developing and applying a second generation of high-performance climate models. Through the addition of relevant physical processes, we are developing an earth systems modeling capability as well.
Parallel In Situ Indexing for Data-intensive Computing
Kim, Jinoh; Abbasi, Hasan; Chacon, Luis; Docan, Ciprian; Klasky, Scott; Liu, Qing; Podhorszki, Norbert; Shoshani, Arie; Wu, Kesheng
2011-09-09
As computing power increases exponentially, vast amount of data is created by many scientific re- search activities. However, the bandwidth for storing the data to disks and reading the data from disks has been improving at a much slower pace. These two trends produce an ever-widening data access gap. Our work brings together two distinct technologies to address this data access issue: indexing and in situ processing. From decades of database research literature, we know that indexing is an effective way to address the data access issue, particularly for accessing relatively small fraction of data records. As data sets increase in sizes, more and more analysts need to use selective data access, which makes indexing an even more important for improving data access. The challenge is that most implementations of in- dexing technology are embedded in large database management systems (DBMS), but most scientific datasets are not managed by any DBMS. In this work, we choose to include indexes with the scientific data instead of requiring the data to be loaded into a DBMS. We use compressed bitmap indexes from the FastBit software which are known to be highly effective for query-intensive workloads common to scientific data analysis. To use the indexes, we need to build them first. The index building procedure needs to access the whole data set and may also require a significant amount of compute time. In this work, we adapt the in situ processing technology to generate the indexes, thus removing the need of read- ing data from disks and to build indexes in parallel. The in situ data processing system used is ADIOS, a middleware for high-performance I/O. Our experimental results show that the indexes can improve the data access time up to 200 times depending on the fraction of data selected, and using in situ data processing system can effectively reduce the time needed to create the indexes, up to 10 times with our in situ technique when using identical parallel settings.
Data communications in a parallel active messaging interface of a parallel computer
Archer, Charles J; Blocksome, Michael A; Ratterman, Joseph D; Smith, Brian E
2014-11-18
Data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints, each endpoint including a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, endpoints coupled for data communications through the PAMI and through data communications resources, including receiving in an origin endpoint of the PAMI a SEND instruction, the SEND instruction specifying a transmission of transfer data from the origin endpoint to a first target endpoint; transmitting from the origin endpoint to the first target endpoint a Request-To-Send (`RTS`) message advising the first target endpoint of the location and size of the transfer data; assigning by the first target endpoint to each of a plurality of target endpoints separate portions of the transfer data; and receiving by the plurality of target endpoints the transfer data.
Data communications in a parallel active messaging interface of a parallel computer
Archer, Charles J; Blocksome, Michael A; Ratterman, Joseph D; Smith, Brian E
2015-02-03
Data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints, each endpoint including a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, endpoints coupled for data communications through the PAMI and through data communications resources, including receiving in an origin endpoint of the PAMI a SEND instruction, the SEND instruction specifying a transmission of transfer data from the origin endpoint to a first target endpoint; transmitting from the origin endpoint to the first target endpoint a Request-To-Send (`RTS`) message advising the first target endpoint of the location and size of the transfer data; assigning by the first target endpoint to each of a plurality of target endpoints separate portions of the transfer data; and receiving by the plurality of target endpoints the transfer data.
The 2nd Symposium on the Frontiers of Massively Parallel Computations
NASA Technical Reports Server (NTRS)
Mills, Ronnie (Editor)
1988-01-01
Programming languages, computer graphics, neural networks, massively parallel computers, SIMD architecture, algorithms, digital terrain models, sort computation, simulation of charged particle transport on the massively parallel processor and image processing are among the topics discussed.
Parallel block schemes for large scale least squares computations
Golub, G.H.; Plemmons, R.J.; Sameh, A.
1986-04-01
Large scale least squares computations arise in a variety of scientific and engineering problems, including geodetic adjustments and surveys, medical image analysis, molecular structures, partial differential equations and substructuring methods in structural engineering. In each of these problems, matrices often arise which possess a block structure which reflects the local connection nature of the underlying physical problem. For example, such super-large nonlinear least squares computations arise in geodesy. Here the coordinates of positions are calculated by iteratively solving overdetermined systems of nonlinear equations by the Gauss-Newton method. The US National Geodetic Survey will complete this year (1986) the readjustment of the North American Datum, a problem which involves over 540 thousand unknowns and over 6.5 million observations (equations). The observation matrix for these least squares computations has a block angular form with 161 diagnonal blocks, each containing 3 to 4 thousand unknowns. In this paper parallel schemes are suggested for the orthogonal factorization of matrices in block angular form and for the associated backsubstitution phase of the least squares computations. In addition, a parallel scheme for the calculation of certain elements of the covariance matrix for such problems is described. It is shown that these algorithms are ideally suited for multiprocessors with three levels of parallelism such as the Cedar system at the University of Illinois. 20 refs., 7 figs.
Distributed parallel computing in stochastic modeling of groundwater systems.
Dong, Yanhui; Li, Guomin; Xu, Haizhen
2013-03-01
Stochastic modeling is a rapidly evolving, popular approach to the study of the uncertainty and heterogeneity of groundwater systems. However, the use of Monte Carlo-type simulations to solve practical groundwater problems often encounters computational bottlenecks that hinder the acquisition of meaningful results. To improve the computational efficiency, a system that combines stochastic model generation with MODFLOW-related programs and distributed parallel processing is investigated. The distributed computing framework, called the Java Parallel Processing Framework, is integrated into the system to allow the batch processing of stochastic models in distributed and parallel systems. As an example, the system is applied to the stochastic delineation of well capture zones in the Pinggu Basin in Beijing. Through the use of 50 processing threads on a cluster with 10 multicore nodes, the execution times of 500 realizations are reduced to 3% compared with those of a serial execution. Through this application, the system demonstrates its potential in solving difficult computational problems in practical stochastic modeling.
3D seismic imaging on massively parallel computers
Womble, D.E.; Ober, C.C.; Oldfield, R.
1997-02-01
The ability to image complex geologies such as salt domes in the Gulf of Mexico and thrusts in mountainous regions is a key to reducing the risk and cost associated with oil and gas exploration. Imaging these structures, however, is computationally expensive. Datasets can be terabytes in size, and the processing time required for the multiple iterations needed to produce a velocity model can take months, even with the massively parallel computers available today. Some algorithms, such as 3D, finite-difference, prestack, depth migration remain beyond the capacity of production seismic processing. Massively parallel processors (MPPs) and algorithms research are the tools that will enable this project to provide new seismic processing capabilities to the oil and gas industry. The goals of this work are to (1) develop finite-difference algorithms for 3D, prestack, depth migration; (2) develop efficient computational approaches for seismic imaging and for processing terabyte datasets on massively parallel computers; and (3) develop a modular, portable, seismic imaging code.
Parallel computation of meshless methods for explicit dynamic analysis.
Danielson, K. T.; Hao, S.; Liu, W. K.; Uras, R. A.; Li, S.; Reactor Engineering; Northwestern Univ.; Waterways Experiment Station
2000-03-10
A parallel computational implementation of modern meshless methods is presented for explicit dynamic analysis. The procedures are demonstrated by application of the Reproducing Kernel Particle Method (RKPM). Aspects of a coarse grain parallel paradigm are detailed for a Lagrangian formulation using model partitioning. Integration points are uniquely defined on separate processors and particle definitions are duplicated, as necessary, so that all support particles for each point are defined locally on the corresponding processor. Several partitioning schemes are considered and a reduced graph-based procedure is presented. Partitioning issues are discussed and procedures to accommodate essential boundary conditions in parallel are presented. Explicit MPI message passing statements are used for all communications among partitions on different processors. The effectiveness of the procedure is demonstrated by highly deformable inelastic example problems.
Cawkwell, M J; Wood, M A; Niklasson, Anders M N; Mniszewski, S M
2014-12-01
The algorithm developed in Cawkwell, M. J. et al. J. Chem. Theory Comput. 2012 , 8 , 4094 for the computation of the density matrix in electronic structure theory on a graphics processing unit (GPU) using the second-order spectral projection (SP2) method [ Niklasson, A. M. N. Phys. Rev. B 2002 , 66 , 155115 ] has been efficiently parallelized over multiple GPUs on a single compute node. The parallel implementation provides significant speed-ups with respect to the single GPU version with no loss of accuracy. The performance and accuracy of the parallel GPU-based algorithm is compared with the performance of the SP2 algorithm and traditional matrix diagonalization methods on a multicore central processing unit (CPU).
Scalable simulations for directed self-assembly patterning with the use of GPU parallel computing
NASA Astrophysics Data System (ADS)
Yoshimoto, Kenji; Peters, Brandon L.; Khaira, Gurdaman S.; de Pablo, Juan J.
2012-03-01
Directed self-assembly (DSA) patterning has been increasingly investigated as an alternative lithographic process for future technology nodes. One of the critical specs for DSA patterning is defects generated through annealing process or by roughness of pre-patterned structure. Due to their high sensitivity to the process and wafer conditions, however, characterization of those defects still remain challenging. DSA simulations can be a powerful tool to predict the formation of the DSA defects. In this work, we propose a new method to perform parallel computing of DSA Monte Carlo (MC) simulations. A consumer graphics card was used to access its hundreds of processing units for parallel computing. By partitioning the simulation system into non-interacting domains, we were able to run MC trial moves in parallel on multiple graphics-processing units (GPUs). Our results show a significant improvement in computational performance.
Three-dimensional radiative transfer on a massively parallel computer
NASA Technical Reports Server (NTRS)
Vath, H. M.
1994-01-01
We perform 3D radiative transfer calculations in non-local thermodynamic equilibrium (NLTE) in the simple two-level atom approximation on the Mas-Par MP-1, which contains 8192 processors and is a single instruction multiple data (SIMD) machine, an example of the new generation of massively parallel computers. On such a machine, all processors execute the same command at a given time, but on different data. To make radiative transfer calculations efficient, we must re-consider the numerical methods and storage of data. To solve the transfer equation, we adopt the short characteristic method and examine different acceleration methods to obtain the source function. We use the ALI method and test local and non-local operators. Furthermore, we compare the Ng and the orthomin methods of acceleration. We also investigate the use of multi-grid methods to get fast solutions for the NLTE case. In order to test these numerical methods, we apply them to two problems with and without periodic boundary conditions.
Parallel, distributed and GPU computing technologies in single-particle electron microscopy
Schmeisser, Martin; Heisen, Burkhard C.; Luettich, Mario; Busche, Boris; Hauer, Florian; Koske, Tobias; Knauber, Karl-Heinz; Stark, Holger
2009-01-01
Most known methods for the determination of the structure of macromolecular complexes are limited or at least restricted at some point by their computational demands. Recent developments in information technology such as multicore, parallel and GPU processing can be used to overcome these limitations. In particular, graphics processing units (GPUs), which were originally developed for rendering real-time effects in computer games, are now ubiquitous and provide unprecedented computational power for scientific applications. Each parallel-processing paradigm alone can improve overall performance; the increased computational performance obtained by combining all paradigms, unleashing the full power of today’s technology, makes certain applications feasible that were previously virtually impossible. In this article, state-of-the-art paradigms are introduced, the tools and infrastructure needed to apply these paradigms are presented and a state-of-the-art infrastructure and solution strategy for moving scientific applications to the next generation of computer hardware is outlined. PMID:19564686
Executing a gather operation on a parallel computer
Archer, Charles J.; Ratterman, Joseph D.
2012-03-20
Methods, apparatus, and computer program products are disclosed for executing a gather operation on a parallel computer according to embodiments of the present invention. Embodiments include configuring, by the logical root, a result buffer or the logical root, the result buffer having positions, each position corresponding to a ranked node in the operational group and for storing contribution data gathered from that ranked node. Embodiments also include repeatedly for each position in the result buffer: determining, by each compute node of an operational group, whether the current position in the result buffer corresponds with the rank of the compute node, if the current position in the result buffer corresponds with the rank of the compute node, contributing, by that compute node, the compute node's contribution data, if the current position in the result buffer does not correspond with the rank of the compute node, contributing, by that compute node, a value of zero for the contribution data, and storing, by the logical root in the current position in the result buffer, results of a bitwise OR operation of all the contribution data by all compute nodes of the operational group for the current position, the results received through the global combining network.
NASA Astrophysics Data System (ADS)
Fazanaro, Filipe I.; Soriano, Diogo C.; Suyama, Ricardo; Madrid, Marconi K.; Oliveira, José Raimundo de; Muñoz, Ignacio Bravo; Attux, Romis
2016-08-01
The characterization of nonlinear dynamical systems and their attractors in terms of invariant measures, basins of attractions and the structure of their vector fields usually outlines a task strongly related to the underlying computational cost. In this work, the practical aspects related to the use of parallel computing - specially the use of Graphics Processing Units (GPUS) and of the Compute Unified Device Architecture (CUDA) - are reviewed and discussed in the context of nonlinear dynamical systems characterization. In this work such characterization is performed by obtaining both local and global Lyapunov exponents for the classical forced Duffing oscillator. The local divergence measure was employed by the computation of the Lagrangian Coherent Structures (LCSS), revealing the general organization of the flow according to the obtained separatrices, while the global Lyapunov exponents were used to characterize the attractors obtained under one or more bifurcation parameters. These simulation sets also illustrate the required computation time and speedup gains provided by different parallel computing strategies, justifying the employment and the relevance of GPUS and CUDA in such extensive numerical approach. Finally, more than simply providing an overview supported by a representative set of simulations, this work also aims to be a unified introduction to the use of the mentioned parallel computing tools in the context of nonlinear dynamical systems, providing codes and examples to be executed in MATLAB and using the CUDA environment, something that is usually fragmented in different scientific communities and restricted to specialists on parallel computing strategies.
Parallel Information Processing.
ERIC Educational Resources Information Center
Rasmussen, Edie M.
1992-01-01
Examines parallel computer architecture and the use of parallel processors for text. Topics discussed include parallel algorithms; performance evaluation; parallel information processing; parallel access methods for text; parallel and distributed information retrieval systems; parallel hardware for text; and network models for information…
Probabilistic structural mechanics research for parallel processing computers
NASA Technical Reports Server (NTRS)
Sues, Robert H.; Chen, Heh-Chyun; Twisdale, Lawrence A.; Martin, William R.
1991-01-01
Aerospace structures and spacecraft are a complex assemblage of structural components that are subjected to a variety of complex, cyclic, and transient loading conditions. Significant modeling uncertainties are present in these structures, in addition to the inherent randomness of material properties and loads. To properly account for these uncertainties in evaluating and assessing the reliability of these components and structures, probabilistic structural mechanics (PSM) procedures must be used. Much research has focused on basic theory development and the development of approximate analytic solution methods in random vibrations and structural reliability. Practical application of PSM methods was hampered by their computationally intense nature. Solution of PSM problems requires repeated analyses of structures that are often large, and exhibit nonlinear and/or dynamic response behavior. These methods are all inherently parallel and ideally suited to implementation on parallel processing computers. New hardware architectures and innovative control software and solution methodologies are needed to make solution of large scale PSM problems practical.
Establishing a group of endpoints in a parallel computer
Archer, Charles J.; Blocksome, Michael A.; Ratterman, Joseph D.; Smith, Brian E.; Xue, Hanhong
2016-02-02
A parallel computer executes a number of tasks, each task includes a number of endpoints and the endpoints are configured to support collective operations. In such a parallel computer, establishing a group of endpoints receiving a user specification of a set of endpoints included in a global collection of endpoints, where the user specification defines the set in accordance with a predefined virtual representation of the endpoints, the predefined virtual representation is a data structure setting forth an organization of tasks and endpoints included in the global collection of endpoints and the user specification defines the set of endpoints without a user specification of a particular endpoint; and defining a group of endpoints in dependence upon the predefined virtual representation of the endpoints and the user specification.
Final Report: Center for Programming Models for Scalable Parallel Computing
Mellor-Crummey, John
2011-09-13
As part of the Center for Programming Models for Scalable Parallel Computing, Rice University collaborated with project partners in the design, development and deployment of language, compiler, and runtime support for parallel programming models to support application development for the “leadership-class” computer systems at DOE national laboratories. Work over the course of this project has focused on the design, implementation, and evaluation of a second-generation version of Coarray Fortran. Research and development efforts of the project have focused on the CAF 2.0 language, compiler, runtime system, and supporting infrastructure. This has involved working with the teams that provide infrastructure for CAF that we rely on, implementing new language and runtime features, producing an open source compiler that enabled us to evaluate our ideas, and evaluating our design and implementation through the use of benchmarks. The report details the research, development, findings, and conclusions from this work.
Parallel I/O and computation for 3D post-stack depth migration
NASA Astrophysics Data System (ADS)
Mosher, Charles C.; Joyner, Calvin L.
1994-09-01
Scalable parallel algorithms for seismic imaging remain a significant challenge for the oil and gas industry. Scalability must address both the computational and the input/output portions of the algorithm in question. These issues are addressed by the ARCO Seismic Benchmark Suite, a public domain software system that provides an environment for development and performance analysis of parallel seismic processing algorithm. We illustrate some of the issues in the design of scalable parallel imaging algorithms with an example process, 3D post-stack depth migration. The algorithm used is based on an implicit finite difference formulation described by Zhiming Li. Scalability is obtained by designing computation, communication between processors, and input/output as parallel operations. The resulting application runs efficiently on both distributed memory and shared memory hardware platforms with processor counts from 1 - 128 nodes.
About parallel computing on spatial rotations in spin mesomorphic structures
NASA Astrophysics Data System (ADS)
Nesterov, Michal M.; Tarkhanov, Victor I.
2004-05-01
An approach to coherent parallel computing on finite spatial rotations is considered. It is shown that such rotations are elements of 4-dimensional (4D) space and to design any algorithms with them we need to change Boole logic to more complicated rotational one. The latter is provided by formalism of Cayley-Klein parameters, which is compatible with geometric Clifford algebra and all its elements of pure and mixed grades: vectors, scalars, paravectors, quaternions, spinors, etc.
NASA Technical Reports Server (NTRS)
Farhat, Charbel; Crivelli, Luis
1993-01-01
Explicit codes are often used to simulate the nonlinear dynamics of large-scale structural systems, even for low frequency response, because the storage and CPU requirements entailed by the repeated factorizations traditionally found in implicit codes rapidly overwhelm the available computing resources. With the advent of parallel processing, this trend is accelerating because explicit schemes are also easier to parallellize than implicit ones. However, the time step restriction imposed by the Courant stability condition on all explicit schemes cannot yet and perhaps will never be offset by the speed of parallel hardware. Therefore, it is essential to develop efficient and robust alternatives to direct methods that are also amenable to massively parallel processing because implicit codes using unconditionally stable time-integration algorithms are computationally more efficient than explicit codes when simulating low-frequency dynamics. Here we present a domain decomposition method for implicit schemes that requires significantly less storage than factorization algorithms, that is several times faster than other popular direct and iterative methods, that can be easily implemented on both shared and local memory parallel processors, and that is both computationally and communication-wise efficient. The proposed transient domain decomposition method is an extension of the method of Finite Element Tearing and Interconnecting (FETI) developed by Farhat and Roux for the solution of static problems. Serial and parallel performance results on the CRAY Y-MP/8 and the iPSC-860/128 systems are reported and analyzed for realistic structural dynamics problems. These results establish the superiority of the FETI method over both the serial/parallel conjugate gradient algorithm with diagonal scaling and the serial/parallel direct method, and contrast the computational power of the iPSC-860/128 parallel processor with that of the CRAY Y-MP/8 system.
A transient FETI methodology for large-scale parallel implicit computations in structural mechanics
NASA Technical Reports Server (NTRS)
Farhat, Charbel; Crivelli, Luis; Roux, Francois-Xavier
1992-01-01
Explicit codes are often used to simulate the nonlinear dynamics of large-scale structural systems, even for low frequency response, because the storage and CPU requirements entailed by the repeated factorizations traditionally found in implicit codes rapidly overwhelm the available computing resources. With the advent of parallel processing, this trend is accelerating because explicit schemes are also easier to parallelize than implicit ones. However, the time step restriction imposed by the Courant stability condition on all explicit schemes cannot yet -- and perhaps will never -- be offset by the speed of parallel hardware. Therefore, it is essential to develop efficient and robust alternatives to direct methods that are also amenable to massively parallel processing because implicit codes using unconditionally stable time-integration algorithms are computationally more efficient when simulating low-frequency dynamics. Here we present a domain decomposition method for implicit schemes that requires significantly less storage than factorization algorithms, that is several times faster than other popular direct and iterative methods, that can be easily implemented on both shared and local memory parallel processors, and that is both computationally and communication-wise efficient. The proposed transient domain decomposition method is an extension of the method of Finite Element Tearing and Interconnecting (FETI) developed by Farhat and Roux for the solution of static problems. Serial and parallel performance results on the CRAY Y-MP/8 and the iPSC-860/128 systems are reported and analyzed for realistic structural dynamics problems. These results establish the superiority of the FETI method over both the serial/parallel conjugate gradient algorithm with diagonal scaling and the serial/parallel direct method, and contrast the computational power of the iPSC-860/128 parallel processor with that of the CRAY Y-MP/8 system.
A computational fluid dynamics algorithm on a massively parallel computer
NASA Technical Reports Server (NTRS)
Jespersen, Dennis C.; Levit, Creon
1989-01-01
The implementation and performance of a finite-difference algorithm for the compressible Navier-Stokes equations in two or three dimensions on the Connection Machine are described. This machine is a single-instruction multiple-data machine with up to 65536 physical processors. The implicit portion of the algorithm is of particular interest. Running times and megadrop rates are given for two- and three-dimensional problems. Included are comparisons with the standard codes on a Cray X-MP/48.
A Parallel Iterative Method for Computing Molecular Absorption Spectra.
Koval, Peter; Foerster, Dietrich; Coulaud, Olivier
2010-09-14
We describe a fast parallel iterative method for computing molecular absorption spectra within TDDFT linear response and using the LCAO method. We use a local basis of "dominant products" to parametrize the space of orbital products that occur in the LCAO approach. In this basis, the dynamic polarizability is computed iteratively within an appropriate Krylov subspace. The iterative procedure uses a matrix-free GMRES method to determine the (interacting) density response. The resulting code is about 1 order of magnitude faster than our previous full-matrix method. This acceleration makes the speed of our TDDFT code comparable with codes based on Casida's equation. The implementation of our method uses hybrid MPI and OpenMP parallelization in which load balancing and memory access are optimized. To validate our approach and to establish benchmarks, we compute spectra of large molecules on various types of parallel machines. The methods developed here are fairly general, and we believe they will find useful applications in molecular physics/chemistry, even for problems that are beyond TDDFT, such as organic semiconductors, particularly in photovoltaics.
Domain decomposition methods for the parallel computation of reacting flows
NASA Technical Reports Server (NTRS)
Keyes, David E.
1988-01-01
Domain decomposition is a natural route to parallel computing for partial differential equation solvers. Subdomains of which the original domain of definition is comprised are assigned to independent processors at the price of periodic coordination between processors to compute global parameters and maintain the requisite degree of continuity of the solution at the subdomain interfaces. In the domain-decomposed solution of steady multidimensional systems of PDEs by finite difference methods using a pseudo-transient version of Newton iteration, the only portion of the computation which generally stands in the way of efficient parallelization is the solution of the large, sparse linear systems arising at each Newton step. For some Jacobian matrices drawn from an actual two-dimensional reacting flow problem, comparisons are made between relaxation-based linear solvers and also preconditioned iterative methods of Conjugate Gradient and Chebyshev type, focusing attention on both iteration count and global inner product count. The generalized minimum residual method with block-ILU preconditioning is judged the best serial method among those considered, and parallel numerical experiments on the Encore Multimax demonstrate for it approximately 10-fold speedup on 16 processors.
RISC Processors and High Performance Computing
NASA Technical Reports Server (NTRS)
Bailey, David H.; Saini, Subhash; Craw, James M. (Technical Monitor)
1995-01-01
This tutorial will discuss the top five RISC microprocessors and the parallel systems in which they are used. It will provide a unique cross-machine comparison not available elsewhere. The effective performance of these processors will be compared by citing standard benchmarks in the context of real applications. The latest NAS Parallel Benchmarks, both absolute performance and performance per dollar, will be listed. The next generation of the NPB will be described. The tutorial will conclude with a discussion of future directions in the field. Technology Transfer Considerations: All of these computer systems are commercially available internationally. Information about these processors is available in the public domain, mostly from the vendors themselves. The NAS Parallel Benchmarks and their results have been previously approved numerous times for public release, beginning back in 1991.
Parallel and vector computation for stochastic optimal control applications
NASA Technical Reports Server (NTRS)
Hanson, F. B.
1989-01-01
A general method for parallel and vector numerical solutions of stochastic dynamic programming problems is described for optimal control of general nonlinear, continuous time, multibody dynamical systems, perturbed by Poisson as well as Gaussian random white noise. Possible applications include lumped flight dynamics models for uncertain environments, such as large scale and background random atmospheric fluctuations. The numerical formulation is highly suitable for a vector multiprocessor or vectorizing supercomputer, and results exhibit high processor efficiency and numerical stability. Advanced computing techniques, data structures, and hardware help alleviate Bellman's curse of dimensionality in dynamic programming computations.
Idle waves in high-performance computing.
Markidis, Stefano; Vencels, Juris; Peng, Ivy Bo; Akhmetova, Dana; Laure, Erwin; Henri, Pierre
2015-01-01
The vast majority of parallel scientific applications distributes computation among processes that are in a busy state when computing and in an idle state when waiting for information from other processes. We identify the propagation of idle waves through processes in scientific applications with a local information exchange between the two processes. Idle waves are nondispersive and have a phase velocity inversely proportional to the average busy time. The physical mechanism enabling the propagation of idle waves is the local synchronization between two processes due to remote data dependency. This study provides a description of the large number of processes in parallel scientific applications as a continuous medium. This work also is a step towards an understanding of how localized idle periods can affect remote processes, leading to the degradation of global performance in parallel scientific applications.
Spatiotemporal Domain Decomposition for Massive Parallel Computation of Space-Time Kernel Density
NASA Astrophysics Data System (ADS)
Hohl, A.; Delmelle, E. M.; Tang, W.
2015-07-01
Accelerated processing capabilities are deemed critical when conducting analysis on spatiotemporal datasets of increasing size, diversity and availability. High-performance parallel computing offers the capacity to solve computationally demanding problems in a limited timeframe, but likewise poses the challenge of preventing processing inefficiency due to workload imbalance between computing resources. Therefore, when designing new algorithms capable of implementing parallel strategies, careful spatiotemporal domain decomposition is necessary to account for heterogeneity in the data. In this study, we perform octtree-based adaptive decomposition of the spatiotemporal domain for parallel computation of space-time kernel density. In order to avoid edge effects near subdomain boundaries, we establish spatiotemporal buffers to include adjacent data-points that are within the spatial and temporal kernel bandwidths. Then, we quantify computational intensity of each subdomain to balance workloads among processors. We illustrate the benefits of our methodology using a space-time epidemiological dataset of Dengue fever, an infectious vector-borne disease that poses a severe threat to communities in tropical climates. Our parallel implementation of kernel density reaches substantial speedup compared to sequential processing, and achieves high levels of workload balance among processors due to great accuracy in quantifying computational intensity. Our approach is portable of other space-time analytical tests.
Runtime optimization of an application executing on a parallel computer
Faraj, Daniel A; Smith, Brian E
2014-11-25
Identifying a collective operation within an application executing on a parallel computer; identifying a call site of the collective operation; determining whether the collective operation is root-based; if the collective operation is not root-based: establishing a tuning session and executing the collective operation in the tuning session; if the collective operation is root-based, determining whether all compute nodes executing the application identified the collective operation at the same call site; if all compute nodes identified the collective operation at the same call site, establishing a tuning session and executing the collective operation in the tuning session; and if all compute nodes executing the application did not identify the collective operation at the same call site, executing the collective operation without establishing a tuning session.
Runtime optimization of an application executing on a parallel computer
Faraj, Daniel A.; Smith, Brian E.
2013-01-29
Identifying a collective operation within an application executing on a parallel computer; identifying a call site of the collective operation; determining whether the collective operation is root-based; if the collective operation is not root-based: establishing a tuning session and executing the collective operation in the tuning session; if the collective operation is root-based, determining whether all compute nodes executing the application identified the collective operation at the same call site; if all compute nodes identified the collective operation at the same call site, establishing a tuning session and executing the collective operation in the tuning session; and if all compute nodes executing the application did not identify the collective operation at the same call site, executing the collective operation without establishing a tuning session.
Runtime optimization of an application executing on a parallel computer
Faraj, Daniel A; Smith, Brian E
2014-11-18
Identifying a collective operation within an application executing on a parallel computer; identifying a call site of the collective operation; determining whether the collective operation is root-based; if the collective operation is not root-based: establishing a tuning session and executing the collective operation in the tuning session; if the collective operation is root-based, determining whether all compute nodes executing the application identified the collective operation at the same call site; if all compute nodes identified the collective operation at the same call site, establishing a tuning session and executing the collective operation in the tuning session; and if all compute nodes executing the application did not identify the collective operation at the same call site, executing the collective operation without establishing a tuning session.
Dust Dynamics in Protoplanetary Disks: Parallel Computing with PVM
NASA Astrophysics Data System (ADS)
de La Fuente Marcos, Carlos; Barge, Pierre; de La Fuente Marcos, Raúl
2002-03-01
We describe a parallel version of our high-order-accuracy particle-mesh code for the simulation of collisionless protoplanetary disks. We use this code to carry out a massively parallel, two-dimensional, time-dependent, numerical simulation, which includes dust particles, to study the potential role of large-scale, gaseous vortices in protoplanetary disks. This noncollisional problem is easy to parallelize on message-passing multicomputer architectures. We performed the simulations on a cache-coherent nonuniform memory access Origin 2000 machine, using both the parallel virtual machine (PVM) and message-passing interface (MPI) message-passing libraries. Our performance analysis suggests that, for our problem, PVM is about 25% faster than MPI. Using PVM and MPI made it possible to reduce CPU time and increase code performance. This allows for simulations with a large number of particles (N ~ 105-106) in reasonable CPU times. The performances of our implementation of the pa! rallel code on an Origin 2000 supercomputer are presented and discussed. They exhibit very good speedup behavior and low load unbalancing. Our results confirm that giant gaseous vortices can play a dominant role in giant planet formation.
Performance Measurement, Visualization and Modeling of Parallel and Distributed Programs
NASA Technical Reports Server (NTRS)
Yan, Jerry C.; Sarukkai, Sekhar R.; Mehra, Pankaj; Lum, Henry, Jr. (Technical Monitor)
1994-01-01
This paper presents a methodology for debugging the performance of message-passing programs on both tightly coupled and loosely coupled distributed-memory machines. The AIMS (Automated Instrumentation and Monitoring System) toolkit, a suite of software tools for measurement and analysis of performance, is introduced and its application illustrated using several benchmark programs drawn from the field of computational fluid dynamics. AIMS includes (i) Xinstrument, a powerful source-code instrumentor, which supports both Fortran77 and C as well as a number of different message-passing libraries including Intel's NX Thinking Machines' CMMD, and PVM; (ii) Monitor, a library of timestamping and trace -collection routines that run on supercomputers (such as Intel's iPSC/860, Delta, and Paragon and Thinking Machines' CM5) as well as on networks of workstations (including Convex Cluster and SparcStations connected by a LAN); (iii) Visualization Kernel, a trace-animation facility that supports source-code clickback, simultaneous visualization of computation and communication patterns, as well as analysis of data movements; (iv) Statistics Kernel, an advanced profiling facility, that associates a variety of performance data with various syntactic components of a parallel program; (v) Index Kernel, a diagnostic tool that helps pinpoint performance bottlenecks through the use of abstract indices; (vi) Modeling Kernel, a facility for automated modeling of message-passing programs that supports both simulation -based and analytical approaches to performance prediction and scalability analysis; (vii) Intrusion Compensator, a utility for recovering true performance from observed performance by removing the overheads of monitoring and their effects on the communication pattern of the program; and (viii) Compatibility Tools, that convert AIMS-generated traces into formats used by other performance-visualization tools, such as ParaGraph, Pablo, and certain AVS/Explorer modules.
User-microprogrammable, local host computer with low-level parallelism
Tomita, S.; Shibayama, K.; Kitamura, T.; Nakata, T.; Hagiwara, H.
1983-01-01
This paper describes the architecture of a dynamically microprogrammable computer with low-level parallelism, called QA-2, which is designed as a high-performance, local host computer for laboratory use. The architectural principle of the QA-2 is the marriage of high-speed, parallel processing capability offered by four powerful arithmetic and logic units (ALUS) with architectural flexibility provided by large scale, dynamic user-microprogramming. By changing its writable control storage dynamically, the QA-2 can be tailored to a wide spectrum of research-oriented applications covering high-level language processing and real-time processing. 11 references.
ERIC Educational Resources Information Center
Chen, Hsinchun; Martinez, Joanne; Kirchhoff, Amy; Ng, Tobun D.; Schatz, Bruce R.
1998-01-01
Grounded on object filtering, automatic indexing, and co-occurrence analysis, an experiment was performed using a parallel supercomputer to analyze over 400,000 abstracts in an INSPEC computer engineering collection. A user evaluation revealed that system-generated thesauri were better than the human-generated INSPEC subject thesaurus in concept…
Data communications in a parallel active messaging interface of a parallel computer
Davis, Kristan D; Faraj, Daniel A
2013-07-09
Algorithm selection for data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints, each endpoint including specifications of a client, a context, and a task, endpoints coupled for data communications through the PAMI, including associating in the PAMI data communications algorithms and ranges of message sizes so that each algorithm is associated with a separate range of message sizes; receiving in an origin endpoint of the PAMI a data communications instruction, the instruction specifying transmission of a data communications message from the origin endpoint to a target endpoint, the data communications message characterized by a message size; selecting, from among the associated algorithms and ranges, a data communications algorithm in dependence upon the message size; and transmitting, according to the selected data communications algorithm from the origin endpoint to the target endpoint, the data communications message.
Blocksome, Michael A; Mamidala, Amith R
2014-02-11
Fencing direct memory access (`DMA`) data transfers in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI including data communications endpoints, each endpoint including specifications of a client, a context, and a task, the endpoints coupled for data communications through the PAMI and through DMA controllers operatively coupled to segments of shared random access memory through which the DMA controllers deliver data communications deterministically, including initiating execution through the PAMI of an ordered sequence of active DMA instructions for DMA data transfers between two endpoints, effecting deterministic DMA data transfers through a DMA controller and a segment of shared memory; and executing through the PAMI, with no FENCE accounting for DMA data transfers, an active FENCE instruction, the FENCE instruction completing execution only after completion of all DMA instructions initiated prior to execution of the FENCE instruction for DMA data transfers between the two endpoints.
Blocksome, Michael A.; Mamidala, Amith R.
2013-09-03
Fencing direct memory access (`DMA`) data transfers in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI including data communications endpoints, each endpoint including specifications of a client, a context, and a task, the endpoints coupled for data communications through the PAMI and through DMA controllers operatively coupled to segments of shared random access memory through which the DMA controllers deliver data communications deterministically, including initiating execution through the PAMI of an ordered sequence of active DMA instructions for DMA data transfers between two endpoints, effecting deterministic DMA data transfers through a DMA controller and a segment of shared memory; and executing through the PAMI, with no FENCE accounting for DMA data transfers, an active FENCE instruction, the FENCE instruction completing execution only after completion of all DMA instructions initiated prior to execution of the FENCE instruction for DMA data transfers between the two endpoints.
Data communications in a parallel active messaging interface of a parallel computer
Blocksome, Michael A.; Ratterman, Joseph D.; Smith, Brian E.
2014-09-02
Eager send data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints that specify a client, a context, and a task, including receiving an eager send data communications instruction with transfer data disposed in a send buffer characterized by a read/write send buffer memory address in a read/write virtual address space of the origin endpoint; determining for the send buffer a read-only send buffer memory address in a read-only virtual address space, the read-only virtual address space shared by both the origin endpoint and the target endpoint, with all frames of physical memory mapped to pages of virtual memory in the read-only virtual address space; and communicating by the origin endpoint to the target endpoint an eager send message header that includes the read-only send buffer memory address.
Data communications in a parallel active messaging interface of a parallel computer
Blocksome, Michael A.; Ratterman, Joseph D.; Smith, Brian E.
2014-09-16
Eager send data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints that specify a client, a context, and a task, including receiving an eager send data communications instruction with transfer data disposed in a send buffer characterized by a read/write send buffer memory address in a read/write virtual address space of the origin endpoint; determining for the send buffer a read-only send buffer memory address in a read-only virtual address space, the read-only virtual address space shared by both the origin endpoint and the target endpoint, with all frames of physical memory mapped to pages of virtual memory in the read-only virtual address space; and communicating by the origin endpoint to the target endpoint an eager send message header that includes the read-only send buffer memory address.
Faraj, Daniel A.
2015-11-19
Algorithm selection for data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints, each endpoint including specifications of a client, a context, and a task, endpoints coupled for data communications through the PAMI, including associating in the PAMI data communications algorithms and bit masks; receiving in an origin endpoint of the PAMI a collective instruction, the instruction specifying transmission of a data communications message from the origin endpoint to a target endpoint; constructing a bit mask for the received collective instruction; selecting, from among the associated algorithms and bit masks, a data communications algorithm in dependence upon the constructed bit mask; and executing the collective instruction, transmitting, according to the selected data communications algorithm from the origin endpoint to the target endpoint, the data communications message.
Data communications in a parallel active messaging interface of a parallel computer
Davis, Kristan D.; Faraj, Daniel A.
2014-07-22
Algorithm selection for data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints, each endpoint including specifications of a client, a context, and a task, endpoints coupled for data communications through the PAMI, including associating in the PAMI data communications algorithms and ranges of message sizes so that each algorithm is associated with a separate range of message sizes; receiving in an origin endpoint of the PAMI a data communications instruction, the instruction specifying transmission of a data communications message from the origin endpoint to a target endpoint, the data communications message characterized by a message size; selecting, from among the associated algorithms and ranges, a data communications algorithm in dependence upon the message size; and transmitting, according to the selected data communications algorithm from the origin endpoint to the target endpoint, the data communications message.
Faraj, Daniel A
2013-07-16
Algorithm selection for data communications in a parallel active messaging interface (`PAMI`) of a parallel computer, the PAMI composed of data communications endpoints, each endpoint including specifications of a client, a context, and a task, endpoints coupled for data communications through the PAMI, including associating in the PAMI data communications algorithms and bit masks; receiving in an origin endpoint of the PAMI a collective instruction, the instruction specifying transmission of a data communications message from the origin endpoint to a target endpoint; constructing a bit mask for the received collective instruction; selecting, from among the associated algorithms and bit masks, a data communications algorithm in dependence upon the constructed bit mask; and executing the collective instruction, transmitting, according to the selected data communications algorithm from the origin endpoint to the target endpoint, the data communications message.
Reverse Computation for Rollback-based Fault Tolerance in Large Parallel Systems
Perumalla, Kalyan S; Park, Alfred J
2013-01-01
Reverse computation is presented here as an important future direction in addressing the challenge of fault tolerant execution on very large cluster platforms for parallel computing. As the scale of parallel jobs increases, traditional checkpointing approaches suffer scalability problems ranging from computational slowdowns to high congestion at the persistent stores for checkpoints. Reverse computation can overcome such problems and is also better suited for parallel computing on newer architectures with smaller, cheaper or energy-efficient memories and file systems. Initial evidence for the feasibility of reverse computation in large systems is presented with detailed performance data from a particle simulation scaling to 65,536 processor cores and 950 accelerators (GPUs). Reverse computation is observed to deliver very large gains relative to checkpointing schemes when nodes rely on their host processors/memory to tolerate faults at their accelerators. A comparison between reverse computation and checkpointing with measurements such as cache miss ratios, TLB misses and memory usage indicates that reverse computation is hard to ignore as a future alternative to be pursued in emerging architectures.
Parallel matrix transpose algorithms on distributed memory concurrent computers
Choi, J.; Walker, D.W.; Dongarra, J.J. |
1993-10-01
This paper describes parallel matrix transpose algorithms on distributed memory concurrent processors. It is assumed that the matrix is distributed over a P x Q processor template with a block scattered data distribution. P, Q, and the block size can be arbitrary, so the algorithms have wide applicability. The communication schemes of the algorithms are determined by the greatest common divisor (GCD) of P and Q. If P and Q are relatively prime, the matrix transpose algorithm involves complete exchange communication. If P and Q are not relatively prime, processors are divided into GCD groups and the communication operations are overlapped for different groups of processors. Processors transpose GCD wrapped diagonal blocks simultaneously, and the matrix can be transposed with LCM/GCD steps, where LCM is the least common multiple of P and Q. The algorithms make use of non-blocking, point-to-point communication between processors. The use of nonblocking communication allows a processor to overlap the messages that it sends to different processors, thereby avoiding unnecessary synchronization. Combined with the matrix multiplication routine, C = A{center_dot}B, the algorithms are used to compute parallel multiplications of transposed matrices, C = A{sup T}{center_dot}B{sup T}, in the PUMMA package. Details of the parallel implementation of the algorithms are given, and results are presented for runs on the Intel Touchstone Delta computer.
RISC Processors and High Performance Computing
NASA Technical Reports Server (NTRS)
Saini, Subhash; Bailey, David H.; Lasinski, T. A. (Technical Monitor)
1995-01-01
In this tutorial, we will discuss top five current RISC microprocessors: The IBM Power2, which is used in the IBM RS6000/590 workstation and in the IBM SP2 parallel supercomputer, the DEC Alpha, which is in the DEC Alpha workstation and in the Cray T3D; the MIPS R8000, which is used in the SGI Power Challenge; the HP PA-RISC 7100, which is used in the HP 700 series workstations and in the Convex Exemplar; and the Cray proprietary processor, which is used in the new Cray J916. The architecture of these microprocessors will first be presented. The effective performance of these processors will then be compared, both by citing standard benchmarks and also in the context of implementing a real applications. In the process, different programming models such as data parallel (CM Fortran and HPF) and message passing (PVM and MPI) will be introduced and compared. The latest NAS Parallel Benchmark (NPB) absolute performance and performance per dollar figures will be presented. The next generation of the NP13 will also be described. The tutorial will conclude with a discussion of general trends in the field of high performance computing, including likely future developments in hardware and software technology, and the relative roles of vector supercomputers tightly coupled parallel computers, and clusters of workstations. This tutorial will provide a unique cross-machine comparison not available elsewhere.
Algorithmic support for commodity-based parallel computing systems.
Leung, Vitus Joseph; Bender, Michael A.; Bunde, David P.; Phillips, Cynthia Ann
2003-10-01
The Computational Plant or Cplant is a commodity-based distributed-memory supercomputer under development at Sandia National Laboratories. Distributed-memory supercomputers run many parallel programs simultaneously. Users submit their programs to a job queue. When a job is scheduled to run, it is assigned to a set of available processors. Job runtime depends not only on the number of processors but also on the particular set of processors assigned to it. Jobs should be allocated to localized clusters of processors to minimize communication costs and to avoid bandwidth contention caused by overlapping jobs. This report introduces new allocation strategies and performance metrics based on space-filling curves and one dimensional allocation strategies. These algorithms are general and simple. Preliminary simulations and Cplant experiments indicate that both space-filling curves and one-dimensional packing improve processor locality compared to the sorted free list strategy previously used on Cplant. These new allocation strategies are implemented in Release 2.0 of the Cplant System Software that was phased into the Cplant systems at Sandia by May 2002. Experimental results then demonstrated that the average number of communication hops between the processors allocated to a job strongly correlates with the job's completion time. This report also gives processor-allocation algorithms for minimizing the average number of communication hops between the assigned processors for grid architectures. The associated clustering problem is as follows: Given n points in {Re}d, find k points that minimize their average pairwise L{sub 1} distance. Exact and approximate algorithms are given for these optimization problems. One of these algorithms has been implemented on Cplant and will be included in Cplant System Software, Version 2.1, to be released. In more preliminary work, we suggest improvements to the scheduler separate from the allocator.
Determining collective barrier operation skew in a parallel computer
Faraj, Daniel A.
2015-11-24
Determining collective barrier operation skew in a parallel computer that includes a number of compute nodes organized into an operational group includes: for each of the nodes until each node has been selected as a delayed node: selecting one of the nodes as a delayed node; entering, by each node other than the delayed node, a collective barrier operation; entering, after a delay by the delayed node, the collective barrier operation; receiving an exit signal from a root of the collective barrier operation; and measuring, for the delayed node, a barrier completion time. The barrier operation skew is calculated by: identifying, from the compute nodes' barrier completion times, a maximum barrier completion time and a minimum barrier completion time and calculating the barrier operation skew as the difference of the maximum and the minimum barrier completion time.
DMA shared byte counters in a parallel computer
Chen, Dong; Gara, Alan G.; Heidelberger, Philip; Vranas, Pavlos
2010-04-06
A parallel computer system is constructed as a network of interconnected compute nodes. Each of the compute nodes includes at least one processor, a memory and a DMA engine. The DMA engine includes a processor interface for interfacing with the at least one processor, DMA logic, a memory interface for interfacing with the memory, a DMA network interface for interfacing with the network, injection and reception byte counters, injection and reception FIFO metadata, and status registers and control registers. The injection FIFOs maintain memory locations of the injection FIFO metadata memory locations including its current head and tail, and the reception FIFOs maintain the reception FIFO metadata memory locations including its current head and tail. The injection byte counters and reception byte counters may be shared between messages.
An Evaluation of Architectural Platforms for Parallel Navier-Stokes Computations
NASA Technical Reports Server (NTRS)
Jayasimha, D. N.; Hayder, M. E.; Pillay, S. K.
1996-01-01
We study the computational, communication, and scalability characteristics of a computational fluid dynamics application, which solves the time accurate flow field of a jet using the compressible Navier-Stokes equations, on a variety of parallel architecture platforms. The platforms chosen for this study are a cluster of workstations (the LACE experimental testbed at NASA Lewis), a shared memory multiprocessor (the Cray YMP), and distributed memory multiprocessors with different topologies - the IBM SP and the Cray T3D. We investigate the impact of various networks connecting the cluster of workstations on the performance of the application and the overheads induced by popular message passing libraries used for parallelization. The work also highlights the importance of matching the memory bandwidth to the processor speed for good single processor performance. By studying the performance of an application on a variety of architectures, we are able to point out the strengths and weaknesses of each of the example computing platforms.
Parallelizing Navier-Stokes Computations on a Variety of Architectural Platforms
NASA Technical Reports Server (NTRS)
Jayasimha, D. N.; Hayder, M. E.; Pillay, S. K.
1997-01-01
We study the computational, communication, and scalability characteristics of a Computational Fluid Dynamics application, which solves the time accurate flow field of a jet using the compressible Navier-Stokes equations, on a variety of parallel architectural platforms. The platforms chosen for this study are a cluster of workstations (the LACE experimental testbed at NASA Lewis), a shared memory multiprocessor (the Cray YMP), distributed memory multiprocessors with different topologies-the IBM SP and the Cray T3D. We investigate the impact of various networks, connecting the cluster of workstations, on the performance of the application and the overheads induced by popular message passing libraries used for parallelization. The work also highlights the importance of matching the memory bandwidth to the processor speed for good single processor performance. By studying the performance of an application on a variety of architectures, we are able to point out the strengths and weaknesses of each of the example computing platforms.
Solving sparse triangular linear systems on parallel computers
NASA Technical Reports Server (NTRS)
Anderson, Edward; Saad, Youcef
1989-01-01
This paper describes and compares three parallel algorithms for solving sparse triangular systems of equations. These methods involve some preprocessing overhead and are primarily of interest in solving many systems with the same coefficient matrix. The first approach is to use a fixed blocksize and form the inverse of the diagonal blocks. The second approach is to use a variable blocksize and reorder the unknowns so that the diagonal blocks are diagonal matrices. The latter technique is called level scheduling because of how it is represented in the adjacency graph, and both row-wise and jagged diagonal storage for the off-diagonal blocks are considered. These techniques are analyzed for general parallel computers and experiments are presented for the eight-processor Alliant FX/8.
Eighth SIAM conference on parallel processing for scientific computing: Final program and abstracts
1997-12-31
This SIAM conference is the premier forum for developments in parallel numerical algorithms, a field that has seen very lively and fruitful developments over the past decade, and whose health is still robust. Themes for this conference were: combinatorial optimization; data-parallel languages; large-scale parallel applications; message-passing; molecular modeling; parallel I/O; parallel libraries; parallel software tools; parallel compilers; particle simulations; problem-solving environments; and sparse matrix computations.
Analysis and selection of optimal function implementations in massively parallel computer
Archer, Charles Jens; Peters, Amanda; Ratterman, Joseph D.
2011-05-31
An apparatus, program product and method optimize the operation of a parallel computer system by, in part, collecting performance data for a set of implementations of a function capable of being executed on the parallel computer system based upon the execution of the set of implementations under varying input parameters in a plurality of input dimensions. The collected performance data may be used to generate selection program code that is configured to call selected implementations of the function in response to a call to the function under varying input parameters. The collected performance data may be used to perform more detailed analysis to ascertain the comparative performance of the set of implementations of the function under the varying input parameters.
NASA Astrophysics Data System (ADS)
Bogdanov, P. B.; Gorobets, A. V.; Sukov, S. A.
2013-08-01
The design of efficient algorithms for large-scale gas dynamics computations with hybrid (heterogeneous) computing systems whose high performance relies on massively parallel accelerators is addressed. A high-order accurate finite volume algorithm with polynomial reconstruction on unstructured hybrid meshes is used to compute compressible gas flows in domains of complex geometry. The basic operations of the algorithm are implemented in detail for massively parallel accelerators, including AMD and NVIDIA graphics processing units (GPUs). Major optimization approaches and a computation transfer technique are covered. The underlying programming tool is the Open Computing Language (OpenCL) standard, which performs on accelerators of various architectures, both existing and emerging.
Parallel Midbrain Microcircuits Perform Independent Temporal Transformations
Huguenard, John; Knudsen, Eric
2014-01-01
The capacity to select the most important information and suppress distracting information is crucial for survival. The midbrain contains a network critical for the selection of the strongest stimulus for gaze and attention. In avians, the optic tectum (OT; called the superior colliculus in mammals) and the GABAergic nucleus isthmi pars magnocellularis (Imc) cooperate in the selection process. In the chicken, OT layer 10, located in intermediate layers, responds to afferent input with gamma periodicity (25–75 Hz), measured at the level of individual neurons and the local field potential. In contrast, Imc neurons, which receive excitatory input from layer 10 neurons, respond with tonic, unusually high discharge rates (>150 spikes/s). In this study, we reveal the source of this high-rate inhibitory activity: layer 10 neurons that project to the Imc possess specialized biophysical properties that enable them to transform afferent drive into high firing rates (∼130 spikes/s), whereas neighboring layer 10 neurons, which project elsewhere, transform afferent drive into lower-frequency, periodic discharge patterns. Thus, the intermediate layers of the OT contain parallel, intercalated microcircuits that generate different temporal patterns of activity linked to the functions of their respective downstream targets. PMID:24920618
An experiment in hurricane track prediction using parallel computing methods
NASA Technical Reports Server (NTRS)
Song, Chang G.; Jwo, Jung-Sing; Lakshmivarahan, S.; Dhall, S. K.; Lewis, John M.; Velden, Christopher S.
1994-01-01
The barotropic model is used to explore the advantages of parallel processing in deterministic forecasting. We apply this model to the track forecasting of hurricane Elena (1985). In this particular application, solutions to systems of elliptic equations are the essence of the computational mechanics. One set of equations is associated with the decomposition of the wind into irrotational and nondivergent components - this determines the initial nondivergent state. Another set is associated with recovery of the streamfunction from the forecasted vorticity. We demonstrate that direct parallel methods based on accelerated block cyclic reduction (BCR) significantly reduce the computational time required to solve the elliptic equations germane to this decomposition and forecast problem. A 72-h track prediction was made using incremental time steps of 16 min on a network of 3000 grid points nominally separated by 100 km. The prediction took 30 sec on the 8-processor Alliant FX/8 computer. This was a speed-up of 3.7 when compared to the one-processor version. The 72-h prediction of Elena's track was made as the storm moved toward Florida's west coast. Approximately 200 km west of Tampa Bay, Elena executed a dramatic recurvature that ultimately changed its course toward the northwest. Although the barotropic track forecast was unable to capture the hurricane's tight cycloidal looping maneuver, the subsequent northwesterly movement was accurately forecasted as was the location and timing of landfall near Mobile Bay.
Hyper-parallel photonic quantum computation with coupled quantum dots.
Ren, Bao-Cang; Deng, Fu-Guo
2014-04-11
It is well known that a parallel quantum computer is more powerful than a classical one. So far, there are some important works about the construction of universal quantum logic gates, the key elements in quantum computation. However, they are focused on operating on one degree of freedom (DOF) of quantum systems. Here, we investigate the possibility of achieving scalable hyper-parallel quantum computation based on two DOFs of photon systems. We construct a deterministic hyper-controlled-not (hyper-CNOT) gate operating on both the spatial-mode and the polarization DOFs of a two-photon system simultaneously, by exploiting the giant optical circular birefringence induced by quantum-dot spins in double-sided optical microcavities as a result of cavity quantum electrodynamics (QED). This hyper-CNOT gate is implemented by manipulating the four qubits in the two DOFs of a two-photon system without auxiliary spatial modes or polarization modes. It reduces the operation time and the resources consumed in quantum information processing, and it is more robust against the photonic dissipation noise, compared with the integration of several cascaded CNOT gates in one DOF.
Parallel implementation, validation, and performance of MM5
Michalakes, J.; Canfield, T.; Nanjundiah, R.; Hammond, S.; Grell, G.
1994-12-31
We describe a parallel implementation of the nonhydrostatic version of the Penn State/NCAR Mesoscale Model, MM5, that includes nesting capabilities. This version of the model can run on many different massively Parallel computers (including a cluster of workstations). The model has been implemented and run on the IBM SP and Intel multiprocessors using a columnwise decomposition that supports irregularly shaped allocations of the problem to processors. This stategy will facilitate dynamic load balancing for improved parallel efficiency and promotes a modular design that simplifies the nesting problem AU data communication for finite differencing, inter-domain exchange of data, and I/O is encapsulated within a parallel library, RSL. Hence, there are no sends or receives in the parallel model itself. The library is Generalizable to other, similar finite difference approximation codes. The code is validated by comparing the rate of growth in error between the sequential and parallel models with the error growth rate when the sequential model input is perturbed to simulate floating point rounding error. Series of runs on increasing numbers of parallel processors demonstrate that the parallel implementation is efficient and scalable to large numbers of processors.
ERIC Educational Resources Information Center
Abuzaghleh, Omar; Goldschmidt, Kathleen; Elleithy, Yasser; Lee, Jeongkyu
2013-01-01
With the advances in computing power, high-performance computing (HPC) platforms have had an impact on not only scientific research in advanced organizations but also computer science curriculum in the educational community. For example, multicore programming and parallel systems are highly desired courses in the computer science major. However,…
NASA Astrophysics Data System (ADS)
Yim, Keun Soo
This dissertation summarizes experimental validation and co-design studies conducted to optimize the fault detection capabilities and overheads in hybrid computer systems (e.g., using CPUs and Graphics Processing Units, or GPUs), and consequently to improve the scalability of parallel computer systems using computational accelerators. The experimental validation studies were conducted to help us understand the failure characteristics of CPU-GPU hybrid computer systems under various types of hardware faults. The main characterization targets were faults that are difficult to detect and/or recover from, e.g., faults that cause long latency failures (Ch. 3), faults in dynamically allocated resources (Ch. 4), faults in GPUs (Ch. 5), faults in MPI programs (Ch. 6), and microarchitecture-level faults with specific timing features (Ch. 7). The co-design studies were based on the characterization results. One of the co-designed systems has a set of source-to-source translators that customize and strategically place error detectors in the source code of target GPU programs (Ch. 5). Another co-designed system uses an extension card to learn the normal behavioral and semantic execution patterns of message-passing processes executing on CPUs, and to detect abnormal behaviors of those parallel processes (Ch. 6). The third co-designed system is a co-processor that has a set of new instructions in order to support software-implemented fault detection techniques (Ch. 7). The work described in this dissertation gains more importance because heterogeneous processors have become an essential component of state-of-the-art supercomputers. GPUs were used in three of the five fastest supercomputers that were operating in 2011. Our work included comprehensive fault characterization studies in CPU-GPU hybrid computers. In CPUs, we monitored the target systems for a long period of time after injecting faults (a temporally comprehensive experiment), and injected faults into various types of
JPARSS: A Java Parallel Network Package for Grid Computing
Chen, Jie; Akers, Walter; Chen, Ying; Watson, William
2002-03-01
The emergence of high speed wide area networks makes grid computinga reality. However grid applications that need reliable data transfer still have difficulties to achieve optimal TCP performance due to network tuning of TCP window size to improve bandwidth and to reduce latency on a high speed wide area network. This paper presents a Java package called JPARSS (Java Parallel Secure Stream (Socket)) that divides data into partitions that are sent over several parallel Java streams simultaneously and allows Java or Web applications to achieve optimal TCP performance in a grid environment without the necessity of tuning TCP window size. This package enables single sign-on, certificate delegation and secure or plain-text data transfer using several security components based on X.509 certificate and SSL. Several experiments will be presented to show that using Java parallelstreams is more effective than tuning TCP window size. In addition a simple architecture using Web services
Parallel computation of Euler and Navier-Stokes flows
Swisshelm, J.M.; Johnson, G.M.; Kumar, S.P.
1986-07-01
A multigrid technique useful for accelerating the convergence of Euler and Navier-Stokes flow computations has been restructured to improve its performance on both SIMD and MIMD computers. The new algorithm allows both the construction of longer coarse-grid vectors and the multitasking of entire grids. Computational results are presented for the CDC Cyber 205, Cray X-MP, and Denelcor HEP I. 15 references.
Optimization on a Network-based Parallel Computer System for Supersonic Laminar Wing Design
NASA Technical Reports Server (NTRS)
Garcia, Joseph A.; Cheung, Samson; Holst, Terry L. (Technical Monitor)
1995-01-01
A set of Computational Fluid Dynamics (CFD) routines and flow transition prediction tools are integrated into a network based parallel numerical optimization routine. Through this optimization routine, the design of a 2-D airfoil and an infinitely swept wing will be studied in order to advance the design cycle capability of supersonic laminar flow wings. The goal of advancing supersonic laminar flow wing design is achieved by wisely choosing the design variables used in the optimization routine. The design variables are represented by the theory of Fourier series and potential theory. These theories, combined with the parallel CFD flow routines and flow transition prediction tools, provide a design space for a global optimal point to be searched. Finally, the parallel optimization routine enables gradient evaluations to be performed in a fast and parallel fashion.
Effecting a broadcast with an allreduce operation on a parallel computer
Almasi, Gheorghe; Archer, Charles J.; Ratterman, Joseph D.; Smith, Brian E.
2010-11-02
A parallel computer comprises a plurality of compute nodes organized into at least one operational group for collective parallel operations. Each compute node is assigned a unique rank and is coupled for data communications through a global combining network. One compute node is assigned to be a logical root. A send buffer and a receive buffer is configured. Each element of a contribution of the logical root in the send buffer is contributed. One or more zeros corresponding to a size of the element are injected. An allreduce operation with a bitwise OR using the element and the injected zeros is performed. And the result for the allreduce operation is determined and stored in each receive buffer.
Boyle, Peter A.; Christ, Norman H.; Gara, Alan; Mawhinney, Robert D.; Ohmacht, Martin; Sugavanam, Krishnan
2012-12-11
A prefetch system improves a performance of a parallel computing system. The parallel computing system includes a plurality of computing nodes. A computing node includes at least one processor and at least one memory device. The prefetch system includes at least one stream prefetch engine and at least one list prefetch engine. The prefetch system operates those engines simultaneously. After the at least one processor issues a command, the prefetch system passes the command to a stream prefetch engine and a list prefetch engine. The prefetch system operates the stream prefetch engine and the list prefetch engine to prefetch data to be needed in subsequent clock cycles in the processor in response to the passed command.
A distributed, dynamic, parallel computational model: the role of noise in velocity storage
Merfeld, Daniel M.
2012-01-01
Networks of neurons perform complex calculations using distributed, parallel computation, including dynamic “real-time” calculations required for motion control. The brain must combine sensory signals to estimate the motion of body parts using imperfect information from noisy neurons. Models and experiments suggest that the brain sometimes optimally minimizes the influence of noise, although it remains unclear when and precisely how neurons perform such optimal computations. To investigate, we created a model of velocity storage based on a relatively new technique–“particle filtering”–that is both distributed and parallel. It extends existing observer and Kalman filter models of vestibular processing by simulating the observer model many times in parallel with noise added. During simulation, the variance of the particles defining the estimator state is used to compute the particle filter gain. We applied our model to estimate one-dimensional angular velocity during yaw rotation, which yielded estimates for the velocity storage time constant, afferent noise, and perceptual noise that matched experimental data. We also found that the velocity storage time constant was Bayesian optimal by comparing the estimate of our particle filter with the estimate of the Kalman filter, which is optimal. The particle filter demonstrated a reduced velocity storage time constant when afferent noise increased, which mimics what is known about aminoglycoside ablation of semicircular canal hair cells. This model helps bridge the gap between parallel distributed neural computation and systems-level behavioral responses like the vestibuloocular response and perception. PMID:22514288
NASA Technical Reports Server (NTRS)
Quealy, Angela; Cole, Gary L.; Blech, Richard A.
1993-01-01
The Application Portable Parallel Library (APPL) is a subroutine-based library of communication primitives that is callable from applications written in FORTRAN or C. APPL provides a consistent programmer interface to a variety of distributed and shared-memory multiprocessor MIMD machines. The objective of APPL is to minimize the effort required to move parallel applications from one machine to another, or to a network of homogeneous machines. APPL encompasses many of the message-passing primitives that are currently available on commercial multiprocessor systems. This paper describes APPL (version 2.3.1) and its usage, reports the status of the APPL project, and indicates possible directions for the future. Several applications using APPL are discussed, as well as performance and overhead results.
NASA Astrophysics Data System (ADS)
Herrera, I.; Herrera, G. S.
2015-12-01
Most geophysical systems are macroscopic physical systems. The behavior prediction of such systems is carried out by means of computational models whose basic models are partial differential equations (PDEs) [1]. Due to the enormous size of the discretized version of such PDEs it is necessary to apply highly parallelized super-computers. For them, at present, the most efficient software is based on non-overlapping domain decomposition methods (DDM). However, a limiting feature of the present state-of-the-art techniques is due to the kind of discretizations used in them. Recently, I. Herrera and co-workers using 'non-overlapping discretizations' have produced the DVS-Software which overcomes this limitation [2]. The DVS-software can be applied to a great variety of geophysical problems and achieves very high parallel efficiencies (90%, or so [3]). It is therefore very suitable for effectively applying the most advanced parallel supercomputers available at present. In a parallel talk, in this AGU Fall Meeting, Graciela Herrera Z. will present how this software is being applied to advance MOD-FLOW. Key Words: Parallel Software for Geophysics, High Performance Computing, HPC, Parallel Computing, Domain Decomposition Methods (DDM)REFERENCES [1]. Herrera Ismael and George F. Pinder, Mathematical Modelling in Science and Engineering: An axiomatic approach", John Wiley, 243p., 2012. [2]. Herrera, I., de la Cruz L.M. and Rosas-Medina A. "Non Overlapping Discretization Methods for Partial, Differential Equations". NUMER METH PART D E, 30: 1427-1454, 2014, DOI 10.1002/num 21852. (Open source) [3]. Herrera, I., & Contreras Iván "An Innovative Tool for Effectively Applying Highly Parallelized Software To Problems of Elasticity". Geofísica Internacional, 2015 (In press)
A hybrid-algorithm-based parallel computing framework for optimal reservoir operation
NASA Astrophysics Data System (ADS)
Li, X.; Wei, J.; Li, T.; Wang, G.
2012-12-01
Up to date, various optimization models have been developed to offer optimal operating policies for reservoirs. Each optimization model has its own merits and limitations, and no general algorithm exists even today. At times, some optimization models have to be combined to obtain desired results. In this paper, we present a parallel computing framework to combine various optimization models in a different way compared to traditional serial computing. This framework consists of three functional processor types, that is, master processor, slave processor and transfer processor. The master processor has a full computation scheme that allocates optimization models to slave processors; slave processors perform allocated optimization models; the transfer processor is in charge of the solution communication among all slave processors. Based on these, the proposed framework can perform various optimization models in parallel. Because of the solution communication, the framework can also integrate the merits of involved optimization models while in iteration and the performance of each optimization model can therefore be improved. And more, it can be concluded the framework can effectively improve the solution quality and increase the solution speed by making full use of computing power of parallel computers.
Fast parallel molecular algorithms for DNA-based computation: factoring integers.
Chang, Weng-Long; Guo, Minyi; Ho, Michael Shan-Hui
2005-06-01
The RSA public-key cryptosystem is an algorithm that converts input data to an unrecognizable encryption and converts the unrecognizable data back into its original decryption form. The security of the RSA public-key cryptosystem is based on the difficulty of factoring the product of two large prime numbers. This paper demonstrates to factor the product of two large prime numbers, and is a breakthrough in basic biological operations using a molecular computer. In order to achieve this, we propose three DNA-based algorithms for parallel subtractor, parallel comparator, and parallel modular arithmetic that formally verify our designed molecular solutions for factoring the product of two large prime numbers. Furthermore, this work indicates that the cryptosystems using public-key are perhaps insecure and also presents clear evidence of the ability of molecular computing to perform complicated mathematical operations.
Factorization of large integers on a massively parallel computer
Davis, J.A.; Holdridge, D.B.
1988-01-01
Our interest in integer factorization at Sandia National Laboratories is motivated by cryptographic applications and in particular the security of the RSA encryption-decryption algorithm. We have implemented our version of the quadratic sieve procedure on the NCUBE computer with 1024 processors (nodes). The new code is significantly different in all important aspects from the program used to factor number of order 10/sup 70/ on a single processor CRAY computer. Capabilities of parallel processing and limitation of small local memory necessitated this entirely new implementation. This effort involved several restarts as realizations of program structures that seemed appealing bogged down due to inter-processor communications. We are presently working with integers of magnitude about 10/sup 70/ in tuning this code to the novel hardware. 6 refs., 3 figs.
Parallel computing-based sclera recognition for human identification
NASA Astrophysics Data System (ADS)
Lin, Yong; Du, Eliza Y.; Zhou, Zhi
2012-06-01
Compared to iris recognition, sclera recognition which uses line descriptor can achieve comparable recognition accuracy in visible wavelengths. However, this method is too time-consuming to be implemented in a real-time system. In this paper, we propose a GPU-based parallel computing approach to reduce the sclera recognition time. We define a new descriptor in which the information of KD tree structure and sclera edge are added. Registration and matching task is divided into subtasks in various sizes according to their computation complexities. Every affine transform parameters are generated by searching on KD tree. Texture memory, constant memory, and shared memory are used to store templates and transform matrixes. The experiment results show that the proposed method executed on GPU can dramatically improve the sclera matching speed in hundreds of times without accuracy decreasing.
Progress in Parallel Schur Complement Preconditioning for Computational Fluid Dynamics
NASA Technical Reports Server (NTRS)
Barth, Timothy J.; Chan, Tony F.; Tang, Wei-Pai; Chancellor, Marisa K. (Technical Monitor)
1997-01-01
We consider preconditioning methods for nonself-adjoint advective-diffusive systems based on a non-overlapping Schur complement procedure for arbitrary triangulated domains. The ultimate goal of this research is to develop scalable preconditioning algorithms for fluid flow discretizations on parallel computing architectures. In our implementation of the Schur complement preconditioning technique, the triangulation is first partitioned into a number of subdomains using the METIS multi-level k-way partitioning code. This partitioning induces a natural 2X2 partitioning of the p.d.e. discretization matrix. By considering various inverse approximations of the 2X2 system, we have developed a family of robust preconditioning techniques. A computer code based on these ideas has been developed and tested on the IBM SP2 and the SGI Power Challenge array using MPI message passing protocol. A number of example CFD calculations will be presented to illustrate and assess various Schur complement approximations.
Parallel performance optimizations on unstructured mesh-based simulations
Sarje, Abhinav; Song, Sukhyun; Jacobsen, Douglas; Huck, Kevin; Hollingsworth, Jeffrey; Malony, Allen; Williams, Samuel; Oliker, Leonid
2015-06-01
This paper addresses two key parallelization challenges the unstructured mesh-based ocean modeling code, MPAS-Ocean, which uses a mesh based on Voronoi tessellations: (1) load imbalance across processes, and (2) unstructured data access patterns, that inhibit intra- and inter-node performance. Our work analyzes the load imbalance due to naive partitioning of the mesh, and develops methods to generate mesh partitioning with better load balance and reduced communication. Furthermore, we present methods that minimize both inter- and intranode data movement and maximize data reuse. Our techniques include predictive ordering of data elements for higher cache efficiency, as well as communication reduction approaches.more » We present detailed performance data when running on thousands of cores using the Cray XC30 supercomputer and show that our optimization strategies can exceed the original performance by over 2×. Additionally, many of these solutions can be broadly applied to a wide variety of unstructured grid-based computations.« less
OpenMP-style parallelism in data-centered multicore computing with R
Jiang, Lei; Patel, Pragneshkumar B; Ostrouchov, George; Jamitzky, Ferdinand
2012-01-01
R is a domain specific language widely used for data analysis by the statistics community as well as by researchers in finance, biology, social sciences, and many other disciplines. As R programs are linked to input data, the exponential growth of available data makes high-performance computing with R imperative. To ease the process of writing parallel programs in R, code transformation from a sequential program to a parallel version would bring much convenience to R users. In this paper, we present our work in semiautomatic parallelization of R codes with user-added OpenMPstyle pragmas. While such pragmas are used at the frontend, we take advantage of multiple parallel backends with different R packages. We provide flexibility for importing parallelism with plug-in components, impose built-in MapReduce for data processing, and also maintain code reusability. We illustrate the advantage of the on-the-fly mechanisms which can lead to significant applications in data-centered parallel computing.
On the relationship between parallel computation and graph embedding
Gupta, A.K.
1989-01-01
The problem of efficiently simulating an algorithm designed for an n-processor parallel machine G on an m-processor parallel machine H with n > m arises when parallel algorithms designed for an ideal size machine are simulated on existing machines which are of a fixed size. The author studies this problem when every processor of H takes over the function of a number of processors in G, and he phrases the simulation problem as a graph embedding problem. New embeddings presented address relevant issues arising from the parallel computation environment. The main focus centers around embedding complete binary trees into smaller-sized binary trees, butterflies, and hypercubes. He also considers simultaneous embeddings of r source machines into a single hypercube. Constant factors play a crucial role in his embeddings since they are not only important in practice but also lead to interesting theoretical problems. All of his embeddings minimize dilation and load, which are the conventional cost measures in graph embeddings and determine the maximum amount of time required to simulate one step of G on H. His embeddings also optimize a new cost measure called ({alpha},{beta})-utilization which characterizes how evenly the processors of H are used by the processors of G. Ideally, the utilization should be balanced (i.e., every processor of H simulates at most (n/m) processors of G) and the ({alpha},{beta})-utilization measures how far off from a balanced utilization the embedding is. He presents embeddings for the situation when some processors of G have different capabilities (e.g. memory or I/O) than others and the processors with different capabilities are to be distributed uniformly among the processors of H. Placing such conditions on an embedding results in an increase in some of the cost measures.
Execution models for mapping programs onto distributed memory parallel computers
NASA Technical Reports Server (NTRS)
Sussman, Alan
1992-01-01
The problem of exploiting the parallelism available in a program to efficiently employ the resources of the target machine is addressed. The problem is discussed in the context of building a mapping compiler for a distributed memory parallel machine. The paper describes using execution models to drive the process of mapping a program in the most efficient way onto a particular machine. Through analysis of the execution models for several mapping techniques for one class of programs, we show that the selection of the best technique for a particular program instance can make a significant difference in performance. On the other hand, the results of benchmarks from an implementation of a mapping compiler show that our execution models are accurate enough to select the best mapping technique for a given program.
A parallel finite-difference method for computational aerodynamics
NASA Technical Reports Server (NTRS)
Swisshelm, Julie M.
1989-01-01
A finite-difference scheme for solving complex three-dimensional aerodynamic flow on parallel-processing supercomputers is presented. The method consists of a basic flow solver with multigrid convergence acceleration, embedded grid refinements, and a zonal equation scheme. Multitasking and vectorization have been incorporated into the algorithm. Results obtained include multiprocessed flow simulations from the Cray X-MP and Cray-2. Speedups as high as 3.3 for the two-dimensional case and 3.5 for segments of the three-dimensional case have been achieved on the Cray-2. The entire solver attained a factor of 2.7 improvement over its unitasked version on the Cray-2. The performance of the parallel algorithm on each machine is analyzed.
Keçeli, Murat; Zhang, Hong; Zapol, Peter; Dixon, David A; Wagner, Albert F
2016-02-01
The Shift-and-invert parallel spectral transformations (SIPs), a computational approach to solve sparse eigenvalue problems, is developed for massively parallel architectures with exceptional parallel scalability and robustness. The capabilities of SIPs are demonstrated by diagonalization of density-functional based tight-binding (DFTB) Hamiltonian and overlap matrices for single-wall metallic carbon nanotubes, diamond nanowires, and bulk diamond crystals. The largest (smallest) example studied is a 128,000 (2000) atom nanotube for which ∼330,000 (∼5600) eigenvalues and eigenfunctions are obtained in ∼190 (∼5) seconds when parallelized over 266,144 (16,384) Blue Gene/Q cores. Weak scaling and strong scaling of SIPs are analyzed and the performance of SIPs is compared with other novel methods. Different matrix ordering methods are investigated to reduce the cost of the factorization step, which dominates the time-to-solution at the strong scaling limit. A parallel implementation of assembling the density matrix from the distributed eigenvectors is demonstrated.
Parallel Computation of Unsteady Flows on a Network of Workstations
NASA Technical Reports Server (NTRS)
1997-01-01
Parallel computation of unsteady flows requires significant computational resources. The utilization of a network of workstations seems an efficient solution to the problem where large problems can be treated at a reasonable cost. This approach requires the solution of several problems: 1) the partitioning and distribution of the problem over a network of workstation, 2) efficient communication tools, 3) managing the system efficiently for a given problem. Of course, there is the question of the efficiency of any given numerical algorithm to such a computing system. NPARC code was chosen as a sample for the application. For the explicit version of the NPARC code both two- and three-dimensional problems were studied. Again both steady and unsteady problems were investigated. The issues studied as a part of the research program were: 1) how to distribute the data between the workstations, 2) how to compute and how to communicate at each node efficiently, 3) how to balance the load distribution. In the following, a summary of these activities is presented. Details of the work have been presented and published as referenced.
Low latency, high bandwidth data communications between compute nodes in a parallel computer
Archer, Charles J.; Blocksome, Michael A.; Ratterman, Joseph D.; Smith, Brian E.
2010-11-02
Methods, parallel computers, and computer program products are disclosed for low latency, high bandwidth data communications between compute nodes in a parallel computer. Embodiments include receiving, by an origin direct memory access (`DMA`) engine of an origin compute node, data for transfer to a target compute node; sending, by the origin DMA engine of the origin compute node to a target DMA engine on the target compute node, a request to send (`RTS`) message; transferring, by the origin DMA engine, a predetermined portion of the data to the target compute node using memory FIFO operation; determining, by the origin DMA engine whether an acknowledgement of the RTS message has been received from the target DMA engine; if the an acknowledgement of the RTS message has not been received, transferring, by the origin DMA engine, another predetermined portion of the data to the target compute node using a memory FIFO operation; and if the acknowledgement of the RTS message has been received by the origin DMA engine, transferring, by the origin DMA engine, any remaining portion of the data to the target compute node using a direct put operation.
High Performance Astrophysics Computing
NASA Astrophysics Data System (ADS)
Capuzzo-Dolcetta, R.; Arca-Sedda, M.; Mastrobuono-Battisti, A.; Punzo, D.; Spera, M.
2012-07-01
The application of high end computing to astrophysical problems, mainly in the galactic environment, is developing for many years at the Dep. of Physics of Sapienza Univ. of Roma. The main scientific topic is the physics of self gravitating systems, whose specific subtopics are: i) celestial mechanics and interplanetary probe transfers in the solar system; ii) dynamics of globular clusters and of globular cluster systems in their parent galaxies; iii) nuclear clusters formation and evolution; iv) massive black hole formation and evolution; v) young star cluster early evolution. In this poster we describe the software and hardware computational resources available in our group and how we are developing both software and hardware to reach the scientific aims above itemized.
NASA Technical Reports Server (NTRS)
Biedron, Robert T.; Samareh, Jamshid A.; Green, Lawrence T.
1999-01-01
This paper focuses on the parallel computation of aerodynamic derivatives via automatic differentiation of the Euler/Navier-Stokes solver CFL3D. The comparison with derivatives obtained by finite differences is presented and the scaling of the time required to obtain the derivatives relative to the number of processors employed for the computation is shown. Finally, the derivative computations are coupled with an optimizer and surface/volume grid deformation tools to perform an optimization to reduce the drag of a three-dimensional wing.
NASA Astrophysics Data System (ADS)
Scott, N. S.; Scott, M. P.; Jézéquel, F.
2012-09-01
We describe recent progress of an ongoing research programme aimed at producing computational science software that can exploit high performance architectures in the atomic physics application domain. We examine the computational bottleneck of matrix construction in a suite of two-dimensional R-matrix propagation programs, 2DRMP, that are aimed at creating virtual electron collision experiments on HPC architectures. We build on Ixaru's extended frequency dependent quadrature rules (EFDQR) for Slater integrals and examine the challenge of constructing Hamiltonian matrices in parallel across an m-processor compute node in a block cyclic distribution for subsequent diagonalization by ScaLAPACK.
Computational Biology and High Performance Computing 2000
Simon, Horst D.; Zorn, Manfred D.; Spengler, Sylvia J.; Shoichet, Brian K.; Stewart, Craig; Dubchak, Inna L.; Arkin, Adam P.
2000-10-19
The pace of extraordinary advances in molecular biology has accelerated in the past decade due in large part to discoveries coming from genome projects on human and model organisms. The advances in the genome project so far, happening well ahead of schedule and under budget, have exceeded any dreams by its protagonists, let alone formal expectations. Biologists expect the next phase of the genome project to be even more startling in terms of dramatic breakthroughs in our understanding of human biology, the biology of health and of disease. Only today can biologists begin to envision the necessary experimental, computational and theoretical steps necessary to exploit genome sequence information for its medical impact, its contribution to biotechnology and economic competitiveness, and its ultimate contribution to environmental quality. High performance computing has become one of the critical enabling technologies, which will help to translate this vision of future advances in biology into reality. Biologists are increasingly becoming aware of the potential of high performance computing. The goal of this tutorial is to introduce the exciting new developments in computational biology and genomics to the high performance computing community.
Performance analysis of parallel branch and bound search with the hypercube architecture
NASA Technical Reports Server (NTRS)
Mraz, Richard T.
1987-01-01
With the availability of commercial parallel computers, researchers are examining new classes of problems which might benefit from parallel computing. This paper presents results of an investigation of the class of search intensive problems. The specific problem discussed is the Least-Cost Branch and Bound search method of deadline job scheduling. The object-oriented design methodology was used to map the problem into a parallel solution. While the initial design was good for a prototype, the best performance resulted from fine-tuning the algorithm for a specific computer. The experiments analyze the computation time, the speed up over a VAX 11/785, and the load balance of the problem when using loosely coupled multiprocessor system based on the hypercube architecture.
Development of magnetron sputtering simulator with GPU parallel computing
NASA Astrophysics Data System (ADS)
Sohn, Ilyoup; Kim, Jihun; Bae, Junkyeong; Lee, Jinpil
2014-12-01
Sputtering devices are widely used in the semiconductor and display panel manufacturing process. Currently, a number of surface treatment applications using magnetron sputtering techniques are being used to improve the efficiency of the sputtering process, through the installation of magnets outside the vacuum chamber. Within the internal space of the low pressure chamber, plasma generated from the combination of a rarefied gas and an electric field is influenced interactively. Since the quality of the sputtering and deposition rate on the substrate is strongly dependent on the multi-physical phenomena of the plasma regime, numerical simulations using PIC-MCC (Particle In Cell, Monte Carlo Collision) should be employed to develop an efficient sputtering device. In this paper, the development of a magnetron sputtering simulator based on the PIC-MCC method and the associated numerical techniques are discussed. To solve the electric field equations in the 2-D Cartesian domain, a Poisson equation solver based on the FDM (Finite Differencing Method) is developed and coupled with the Monte Carlo Collision method to simulate the motion of gas particles influenced by an electric field. The magnetic field created from the permanent magnet installed outside the vacuum chamber is also numerically calculated using Biot-Savart's Law. All numerical methods employed in the present PIC code are validated by comparison with analytical and well-known commercial engineering software results, with all of the results showing good agreement. Finally, the developed PIC-MCC code is parallelized to be suitable for general purpose computing on graphics processing unit (GPGPU) acceleration, so as to reduce the large computation time which is generally required for particle simulations. The efficiency and accuracy of the GPGPU parallelized magnetron sputtering simulator are examined by comparison with the calculated results and computation times from the original serial code. It is found that
Pacing a data transfer operation between compute nodes on a parallel computer
Blocksome, Michael A.
2011-09-13
Methods, systems, and products are disclosed for pacing a data transfer between compute nodes on a parallel computer that include: transferring, by an origin compute node, a chunk of an application message to a target compute node; sending, by the origin compute node, a pacing request to a target direct memory access (`DMA`) engine on the target compute node using a remote get DMA operation; determining, by the origin compute node, whether a pacing response to the pacing request has been received from the target DMA engine; and transferring, by the origin compute node, a next chunk of the application message if the pacing response to the pacing request has been received from the target DMA engine.
Parallel computation of GA search for the artery shape determinants with CFD
NASA Astrophysics Data System (ADS)
Himeno, M.; Noda, S.; Fukasaku, K.; Himeno, R.
2010-06-01
We studied which factors play important role to determine the shape of arteries at the carotid artery bifurcation by performing multi-objective optimization with computation fluid dynamics (CFD) and the genetic algorithm (GA). To perform it, the most difficult problem is how to reduce turn-around time of the GA optimization with 3D unsteady computation of blood flow. We devised two levels of parallel computation method with the following features: level 1: parallel CFD computation with appropriate number of cores; level 2: parallel jobs generated by "master", which finds quickly available job cue and dispatches jobs, to reduce turn-around time. As a result, the turn-around time of one GA trial, which would have taken 462 days with one core, was reduced to less than two days on RIKEN supercomputer system, RICC, with 8192 cores. We performed a multi-objective optimization to minimize the maximum mean WSS and to minimize the sum of circumference for four different shapes and obtained a set of trade-off solutions for each shape. In addition, we found that the carotid bulb has the feature of the minimum local mean WSS and minimum local radius. We confirmed that our method is effective for examining determinants of artery shapes.
Dynamic remapping of parallel computations with varying resource demands
NASA Technical Reports Server (NTRS)
Nicol, D. M.; Saltz, J. H.
1986-01-01
A large class of computational problems is characterized by frequent synchronization, and computational requirements which change as a function of time. When such a problem must be solved on a message passing multiprocessor machine, the combination of these characteristics lead to system performance which decreases in time. Performance can be improved with periodic redistribution of computational load; however, redistribution can exact a sometimes large delay cost. We study the issue of deciding when to invoke a global load remapping mechanism. Such a decision policy must effectively weigh the costs of remapping against the performance benefits. We treat this problem by constructing two analytic models which exhibit stochastically decreasing performance. One model is quite tractable; we are able to describe the optimal remapping algorithm, and the optimal decision policy governing when to invoke that algorithm. However, computational complexity prohibits the use of the optimal remapping decision policy. We then study the performance of a general remapping policy on both analytic models. This policy attempts to minimize a statistic W(n) which measures the system degradation (including the cost of remapping) per computation step over a period of n steps. We show that as a function of time, the expected value of W(n) has at most one minimum, and that when this minimum exists it defines the optimal fixed-interval remapping policy. Our decision policy appeals to this result by remapping when it estimates that W(n) is minimized. Our performance data suggests that this policy effectively finds the natural frequency of remapping. We also use the analytic models to express the relationship between performance and remapping cost, number of processors, and the computation's stochastic activity.
Towards an Accurate Performance Modeling of Parallel SparseFactorization
Grigori, Laura; Li, Xiaoye S.
2006-05-26
We present a performance model to analyze a parallel sparseLU factorization algorithm on modern cached-based, high-end parallelarchitectures. Our model characterizes the algorithmic behavior bytakingaccount the underlying processor speed, memory system performance, aswell as the interconnect speed. The model is validated using theSuperLU_DIST linear system solver, the sparse matrices from realapplications, and an IBM POWER3 parallel machine. Our modelingmethodology can be easily adapted to study performance of other types ofsparse factorizations, such as Cholesky or QR.
Tracking image features using a parallel computational model
NASA Astrophysics Data System (ADS)
Ellis, Timothy J.; Mirmehdi, Majid; Dowling, Geoff R.
1992-03-01
This paper describes a parallel implementation of an image feature tracking system. The system is designed to operate as the front-end of a vision system for controlling autonomous guided vehicles (AGV). Image features or tokens (edge-based line segments in the example given here) are extracted from the image and allocated to individual tracking processes. Both the extraction and the tracking stages are performed by concurrent processes. Arbitrary tracking algorithms may be associated with each process. In the current implementation, a Kalman filter is used to track and predict tokens in subsequent image frames.
Implementation of Parallel Computing Technology to Vortex Flow
NASA Technical Reports Server (NTRS)
Dacles-Mariani, Jennifer
1999-01-01
Mainframe supercomputers such as the Cray C90 was invaluable in obtaining large scale computations using several millions of grid points to resolve salient features of a tip vortex flow over a lifting wing. However, real flight configurations require tracking not only of the flow over several lifting wings but its growth and decay in the near- and intermediate- wake regions, not to mention the interaction of these vortices with each other. Resolving and tracking the evolution and interaction of these vortices shed from complex bodies is computationally intensive. Parallel computing technology is an attractive option in solving these flows. In planetary science vortical flows are also important in studying how planets and protoplanets form when cosmic dust and gases become gravitationally unstable and eventually form planets or protoplanets. The current paradigm for the formation of planetary systems maintains that the planets accreted from the nebula of gas and dust left over from the formation of the Sun. Traditional theory also indicate that such a preplanetary nebula took the form of flattened disk. The coagulation of dust led to the settling of aggregates toward the midplane of the disk, where they grew further into asteroid-like planetesimals. Some of the issues still remaining in this process are the onset of gravitational instability, the role of turbulence in the damping of particles and radial effects. In this study the focus will be with the role of turbulence and the radial effects.
Integration experiments and performance studies of a COTS parallel archive system
Chen, Hsing-bung; Scott, Cody; Grider, Gary; Torres, Aaron; Turley, Milton; Sanchez, Kathy; Bremer, John
2010-06-16
Current and future Archive Storage Systems have been asked to (a) scale to very high bandwidths, (b) scale in metadata performance, (c) support policy-based hierarchical storage management capability, (d) scale in supporting changing needs of very large data sets, (e) support standard interface, and (f) utilize commercial-off-the-shelf (COTS) hardware. Parallel file systems have been asked to do the same thing but at one or more orders of magnitude faster in performance. Archive systems continue to move closer to file systems in their design due to the need for speed and bandwidth, especially metadata searching speeds such as more caching and less robust semantics. Currently the number of extreme highly scalable parallel archive solutions is very small especially those that will move a single large striped parallel disk file onto many tapes in parallel. We believe that a hybrid storage approach of using COTS components and innovative software technology can bring new capabilities into a production environment for the HPC community much faster than the approach of creating and maintaining a complete end-to-end unique parallel archive software solution. In this paper, we relay our experience of integrating a global parallel file system and a standard backup/archive product with a very small amount of additional code to provide a scalable, parallel archive. Our solution has a high degree of overlap with current parallel archive products including (a) doing parallel movement to/from tape for a single large parallel file, (b) hierarchical storage management, (c) ILM features, (d) high volume (non-single parallel file) archives for backup/archive/content management, and (e) leveraging all free file movement tools in Linux such as copy, move, Is, tar, etc. We have successfully applied our working COTS Parallel Archive System to the current world's first petafiop/s computing system, LANL's Roadrunner machine, and demonstrated its capability to address requirements of
Integration experiences and performance studies of A COTS parallel archive systems
Chen, Hsing-bung; Scott, Cody; Grider, Bary; Torres, Aaron; Turley, Milton; Sanchez, Kathy; Bremer, John
2010-01-01
Current and future Archive Storage Systems have been asked to (a) scale to very high bandwidths, (b) scale in metadata performance, (c) support policy-based hierarchical storage management capability, (d) scale in supporting changing needs of very large data sets, (e) support standard interface, and (f) utilize commercial-off-the-shelf(COTS) hardware. Parallel file systems have been asked to do the same thing but at one or more orders of magnitude faster in performance. Archive systems continue to move closer to file systems in their design due to the need for speed and bandwidth, especially metadata searching speeds such as more caching and less robust semantics. Currently the number of extreme highly scalable parallel archive solutions is very small especially those that will move a single large striped parallel disk file onto many tapes in parallel. We believe that a hybrid storage approach of using COTS components and innovative software technology can bring new capabilities into a production environment for the HPC community much faster than the approach of creating and maintaining a complete end-to-end unique parallel archive software solution. In this paper, we relay our experience of integrating a global parallel file system and a standard backup/archive product with a very small amount of additional code to provide a scalable, parallel archive. Our solution has a high degree of overlap with current parallel archive products including (a) doing parallel movement to/from tape for a single large parallel file, (b) hierarchical storage management, (c) ILM features, (d) high volume (non-single parallel file) archives for backup/archive/content management, and (e) leveraging all free file movement tools in Linux such as copy, move, ls, tar, etc. We have successfully applied our working COTS Parallel Archive System to the current world's first petaflop/s computing system, LANL's Roadrunner, and demonstrated its capability to address requirements of future
Yokohama, Noriya
2013-07-01
This report was aimed at structuring the design of architectures and studying performance measurement of a parallel computing environment using a Monte Carlo simulation for particle therapy using a high performance computing (HPC) instance within a public cloud-computing infrastructure. Performance measurements showed an approximately 28 times faster speed than seen with single-thread architecture, combined with improved stability. A study of methods of optimizing the system operations also indicated lower cost. PMID:23877155
Yokohama, Noriya
2013-07-01
This report was aimed at structuring the design of architectures and studying performance measurement of a parallel computing environment using a Monte Carlo simulation for particle therapy using a high performance computing (HPC) instance within a public cloud-computing infrastructure. Performance measurements showed an approximately 28 times faster speed than seen with single-thread architecture, combined with improved stability. A study of methods of optimizing the system operations also indicated lower cost.
Parallel computer graphics algorithms for the Connection Machine
Richardson, J.F.
1990-01-01
Many of the classes of computer graphics algorithms and polygon storage schemes can be adapted for parallel execution on various parallel architectures. The connection machine is one such architecture that should be thought of as a multiprocessor grid that can be reconfigured into standard 2-dimensional mesh and n-dimensional hypercube architectures. The classes of algorithms considered in this paper are SPLINES; POLYGON STORAGE; TRIANGULARIZATION; and SYMBOLIC INPUT. The target Connection Machine (hearafter designated as CM) for the algorithms of this paper has 8192 physical processors. Each physical processor has 8 kilobytes of local memory plus an arithmetic-logic unit. All processors can communicate with any other processor through a router. Thus this CM has a shared memory of 64 megabytes when used as a standard multiprocessor (MIMD) architecture. In addition, the CM interconnection structure can simulate a 2-dimensional mesh and n-dimensional hypercube (SIMD) architecture with the mesh being the default architecture. The front end for the CM is a Symbolics and the high level language is LISP or FORTRAN.
Application of parallel computing techniques to a large-scale reservoir simulation
Zhang, Keni; Wu, Yu-Shu; Ding, Chris; Pruess, Karsten
2001-02-01
Even with the continual advances made in both computational algorithms and computer hardware used in reservoir modeling studies, large-scale simulation of fluid and heat flow in heterogeneous reservoirs remains a challenge. The problem commonly arises from intensive computational requirement for detailed modeling investigations of real-world reservoirs. This paper presents the application of a massive parallel-computing version of the TOUGH2 code developed for performing large-scale field simulations. As an application example, the parallelized TOUGH2 code is applied to develop a three-dimensional unsaturated-zone numerical model simulating flow of moisture, gas, and heat in the unsaturated zone of Yucca Mountain, Nevada, a potential repository for high-level radioactive waste. The modeling approach employs refined spatial discretization to represent the heterogeneous fractured tuffs of the system, using more than a million 3-D gridblocks. The problem of two-phase flow and heat transfer within the model domain leads to a total of 3,226,566 linear equations to be solved per Newton iteration. The simulation is conducted on a Cray T3E-900, a distributed-memory massively parallel computer. Simulation results indicate that the parallel computing technique, as implemented in the TOUGH2 code, is very efficient. The reliability and accuracy of the model results have been demonstrated by comparing them to those of small-scale (coarse-grid) models. These comparisons show that simulation results obtained with the refined grid provide more detailed predictions of the future flow conditions at the site, aiding in the assessment of proposed repository performance.
Optimized collectives using a DMA on a parallel computer
Chen, Dong; Gabor, Dozsa; Giampapa, Mark E.; Heidelberger; Phillip
2011-02-08
Optimizing collective operations using direct memory access controller on a parallel computer, in one aspect, may comprise establishing a byte counter associated with a direct memory access controller for each submessage in a message. The byte counter includes at least a base address of memory and a byte count associated with a submessage. A byte counter associated with a submessage is monitored to determine whether at least a block of data of the submessage has been received. The block of data has a predetermined size, for example, a number of bytes. The block is processed when the block has been fully received, for example, when the byte count indicates all bytes of the block have been received. The monitoring and processing may continue for all blocks in all submessages in the message.
On implicit Runge-Kutta methods for parallel computations
NASA Technical Reports Server (NTRS)
Keeling, Stephen L.
1987-01-01
Implicit Runge-Kutta methods which are well-suited for parallel computations are characterized. It is claimed that such methods are first of all, those for which the associated rational approximation to the exponential has distinct poles, and these are called multiply explicit (MIRK) methods. Also, because of the so-called order reduction phenomenon, there is reason to require that these poles be real. Then, it is proved that a necessary condition for a q-stage, real MIRK to be A sub 0-stable with maximal order q + 1 is that q = 1, 2, 3, or 5. Nevertheless, it is shown that for every positive integer q, there exists a q-stage, real MIRK which is I-stable with order q. Finally, some useful examples of algebraically stable MIRKs are given.
Parallel computation of a dam-break flow model using OpenMP on a multi-core computer
NASA Astrophysics Data System (ADS)
Zhang, Shanghong; Xia, Zhongxi; Yuan, Rui; Jiang, Xiaoming
2014-05-01
High-performance calculations are of great importance to the simulation of dam-break events, as discontinuous solutions and accelerated speed are key factors in the process of dam-break flow modeling. In this study, Roe's approximate Riemann solution of the finite volume method is adopted to solve the interface flux of grid cells and accurately simulate the discontinuous flow, and shared memory technology (OpenMP) is used to realize parallel computing. Because an explicit discrete technique is used to solve the governing equations, and there is no correlation between grid calculations in a single time step, the parallel dam-break model can be easily realized by adding OpenMP instructions to the loop structure of the grid calculations. The performance of the model is analyzed using six computing cores and four different grid division schemes for the Pangtoupao flood storage area in China. The results show that the parallel computing improves precision and increases the simulation speed of the dam-break flow, the simulation of 320 h flood process can be completed within 1.6 h on a 16-kernel computer; a speedup factor of 8.64× is achieved. Further analysis reveals that the models involving a larger number of calculations exhibit greater efficiency and a higher rate of acceleration. At the same time, the model has good extendibility, as the speedup increases with the number of processor cores. The parallel model based on OpenMP can make full use of multi-core processors, making it possible to simulate dam-break flows in large-scale watersheds on a single computer.
A parallel simulated annealing algorithm for standard cell placement on a hypercube computer
NASA Technical Reports Server (NTRS)
Jones, Mark Howard
1987-01-01
A parallel version of a simulated annealing algorithm is presented which is targeted to run on a hypercube computer. A strategy for mapping the cells in a two dimensional area of a chip onto processors in an n-dimensional hypercube is proposed such that both small and large distance moves can be applied. Two types of moves are allowed: cell exchanges and cell displacements. The computation of the cost function in parallel among all the processors in the hypercube is described along with a distributed data structure that needs to be stored in the hypercube to support parallel cost evaluation. A novel tree broadcasting strategy is used extensively in the algorithm for updating cell locations in the parallel environment. Studies on the performance of the algorithm on example industrial circuits show that it is faster and gives better final placement results than the uniprocessor simulated annealing algorithms. An improved uniprocessor algorithm is proposed which is based on the improved results obtained from parallelization of the simulated annealing algorithm.
Evaluation of high-performance computing software
Browne, S.; Dongarra, J.; Rowan, T.
1996-12-31
The absence of unbiased and up to date comparative evaluations of high-performance computing software complicates a user`s search for the appropriate software package. The National HPCC Software Exchange (NHSE) is attacking this problem using an approach that includes independent evaluations of software, incorporation of author and user feedback into the evaluations, and Web access to the evaluations. We are applying this approach to the Parallel Tools Library (PTLIB), a new software repository for parallel systems software and tools, and HPC-Netlib, a high performance branch of the Netlib mathematical software repository. Updating the evaluations with feed-back and making it available via the Web helps ensure accuracy and timeliness, and using independent reviewers produces unbiased comparative evaluations difficult to find elsewhere.
Low cost, highly effective parallel computing achieved through a Beowulf cluster.
Bitner, Marc; Skelton, Gordon
2003-01-01
A Beowulf cluster is a means of bringing together several computers and using software and network components to make this cluster of computers appear and function as one computer with multiple parallel computing processors. A cluster of computers can provide comparable computing power usually found only in very expensive super computers or servers.
Parallel computing of overset grids for aerodynamic problems with moving objects
NASA Astrophysics Data System (ADS)
Prewitt, Nathan Coleman
When a store is dropped from a military aircraft at high subsonic, transonic, or supersonic speeds, the aerodynamic forces and moments acting on the store can be sufficient to send the store back into contact with the aircraft. Therefore, store separation analysis is used to certify the safety of any proposed drop. Time accurate computational fluid dynamics (CFD) offers the option of calculating store separation trajectories from first principles. In the Chimera grid scheme, a set of independent, overlapping, structured grids are used to decompose the domain of interest. This allows the use of efficient structured grid flow solvers and associated boundary conditions, and allows for grid motion without stretching or regridding. However, these advantages are gained in exchange for the requirement to establish communication links between the overlapping grids via a process referred to as "grid assembly." Relatively little work has been done to use parallel computing for time accurate, moving body problems. Thus, new techniques are presented for the parallel implementation of the assembly of overset, Chimera grids. This work is based on the grid assembly function defined in the Beggar code, currently under development at Eglin Air Force Base, FL. The parallel performance of each implementation is analyzed and equations are presented for estimating the parallel speedup. Each successive implementation attacks the weaknesses of the previous implementation in an effort to improve the parallel performance. The first implementation achieves the solution of moving body problems on multiple processors with minimum code changes. The second implementation improves the parallel performance by hiding the execution time of the grid assembly function behind the execution time of the flow solver. The third implementation uses coarse grain data decomposition to reduce the execution time of the grid assembly function. The final implementation demonstrates the fine grain decomposition
Parallel Computation of Orbit Determination for Space Debris Population
NASA Astrophysics Data System (ADS)
Olmedo, Estrella; Sanchez-Ortiz, Noelia; Ramos-Lerate, Mercedes
2009-03-01
In this work we present an algorithm for computing Orbit Determination for Space Debris population. The method presents a high degree of parallelism. That means that the number of available computers divides the computational effort. The context of this work and the later scope is to have the capability of cataloguing and correlating the Space Debris population. In this sense, as better the accuracy provided by the orbit determination is, more accurate will be the estimation of the state vectors corresponding to the debris objects and better will be the accuracy of the future catalogue of Space Debris. As more objects we can determinate the corresponding orbit, more complete will be the future catalogue. Therefore numerical tools for orbit determination are a key point in the development of a future ESSAS. The first time that a new object is observed, six measurements (these measurements may come from RADAR, Ground Based Telescope or Space Based Telescope) are required for computing an Initial Orbit Determination (IOD). After that, the Initial Estimated State Vector (IESV) is improved within the next-coming measurement. The idea of this method is the following. From six initial measurements, we compute the IOD following the same ideas of [1]. We compute also the initial knowledge covariance matrix (IKCM) corresponding to the IESV. In general, the numerical error of the IOD is too big for processing the following measurements with a conventional numerical filter (like the Square Root Information Filter (SRIF)). The problem is that the improvement of the accuracy in the IOD is not an easy task in those cases with large initial error. However the computed IKCM give a realistic approximation of the committed error in the IOD. The proposed algorithm uses the IKCM for generating a cloud of IESVs. All the IESV inside the cloud are processed with a new and much smaller IKCM by using SRIF. In such a way that the ones that are close enough to the real state vector (and thus
Parallel Computation of the Topology of Level Sets
Pascucci, V; Cole-McLaughlin, K
2004-12-16
we can compute the Contour Tree in linear time in many practical cases where t = O(n{sup 1-{epsilon}}). We report the running times for a parallel implementation, showing good scalability with the number of processors.
High Performance Photogrammetric Processing on Computer Clusters
NASA Astrophysics Data System (ADS)
Adrov, V. N.; Drakin, M. A.; Sechin, A. Y.
2012-07-01
Most cpu consuming tasks in photogrammetric processing can be done in parallel. The algorithms take independent bits as input and produce independent bits as output. The independence of bits comes from the nature of such algorithms since images, stereopairs or small image blocks parts can be processed independently. Many photogrammetric algorithms are fully automatic and do not require human interference. Photogrammetric workstations can perform tie points measurements, DTM calculations, orthophoto construction, mosaicing and many other service operations in parallel using distributed calculations. Distributed calculations save time reducing several days calculations to several hours calculations. Modern trends in computer technology show the increase of cpu cores in workstations, speed increase in local networks, and as a result dropping the price of the supercomputers or computer clusters that can contain hundreds or even thousands of computing nodes. Common distributed processing in DPW is usually targeted for interactive work with a limited number of cpu cores and is not optimized for centralized administration. The bottleneck of common distributed computing in photogrammetry can be in the limited lan throughput and storage performance, since the processing of huge amounts of large raster images is needed.
NASA Astrophysics Data System (ADS)
Louri, Ahmed; Sung, Hongki
1995-10-01
The interconnection network structure can be the deciding and limiting factor in the cost and the performance of parallel computers. One of the most popular point-to-point interconnection networks for parallel computers today is the hypercube. The regularity, logarithmic diameter, symmetry, high connectivity, fault tolerance, simple routing, and reconfigurability (easy embedding of other network topologies) of the hypercube make it a very attractive choice for parallel computers. Unfortunately the hypercube possesses a major drawback, which is the links per node increases as the network grows in size. As an alternative to the hypercube, the binary de Bruijn (BdB) network has recently received much attention. The BdB not only provides a logarithmic diameter, fault tolerance, and simple routing but also requires fewer links than the hypercube for the same network size. Additionally, a major advantage of the BdB edges per node is independent of the network size. This makes it very desirable for large-scale parallel systems. However, because of its asymmetrical nature and global connectivity, it poses a major challenge for VLSI technology. Optics, owing to its three-dimensional and global-connectivity nature, seems to be very suitable for implementing BdB networks. We present an implementation methodology for optical BdB networks. The distinctive feature of the proposed implementation methodology is partitionability of the network into a few primitive operations that can be implemented efficiently. We further show feasibility of the
High-performance computing in image registration
NASA Astrophysics Data System (ADS)
Zanin, Michele; Remondino, Fabio; Dalla Mura, Mauro
2012-10-01
Thanks to the recent technological advances, a large variety of image data is at our disposal with variable geometric, radiometric and temporal resolution. In many applications the processing of such images needs high performance computing techniques in order to deliver timely responses e.g. for rapid decisions or real-time actions. Thus, parallel or distributed computing methods, Digital Signal Processor (DSP) architectures, Graphical Processing Unit (GPU) programming and Field-Programmable Gate Array (FPGA) devices have become essential tools for the challenging issue of processing large amount of geo-data. The article focuses on the processing and registration of large datasets of terrestrial and aerial images for 3D reconstruction, diagnostic purposes and monitoring of the environment. For the image alignment procedure, sets of corresponding feature points need to be automatically extracted in order to successively compute the geometric transformation that aligns the data. The feature extraction and matching are ones of the most computationally demanding operations in the processing chain thus, a great degree of automation and speed is mandatory. The details of the implemented operations (named LARES) exploiting parallel architectures and GPU are thus presented. The innovative aspects of the implementation are (i) the effectiveness on a large variety of unorganized and complex datasets, (ii) capability to work with high-resolution images and (iii) the speed of the computations. Examples and comparisons with standard CPU processing are also reported and commented.
Bassetti, F.; Davis, K.; Quinlan, D.
1998-09-01
Application codes reliably achieve performance far less than the advertised capabilities of existing architectures, and this problem is worsening with increasingly-parallel machines. For large-scale numerical applications, stencil operations often impose the greater part of the computational cost, and the primary sources of inefficiency are the costs of message passing and poor cache utilization. This paper proposes and demonstrates optimizations for stencil and stencil-like computations for both serial and parallel environments that ameliorate these sources of inefficiency. Additionally, the authors argue that when stencil-like computations are encoded at a high level using object-oriented parallel array class libraries these optimizations, which are beyond the capability of compilers, may be automated.
Multicore Challenges and Benefits for High Performance Scientific Computing
Nielsen, Ida M. B.; Janssen, Curtis L.
2008-01-01
Until recently, performance gains in processors were achieved largely by improvements in clock speeds and instruction level parallelism. Thus, applications could obtain performance increases with relatively minor changes by upgrading to the latest generation of computing hardware. Currently, however, processor performance improvements are realized by using multicore technology and hardware support for multiple threads within each core, and taking full advantage of this technology to improve the performance of applications requires exposure of extreme levels of software parallelism. We will here discuss the architecture of parallel computers constructed from many multicore chips as well as techniques for managing the complexitymore » of programming such computers, including the hybrid message-passing/multi-threading programming model. We will illustrate these ideas with a hybrid distributed memory matrix multiply and a quantum chemistry algorithm for energy computation using Møller–Plesset perturbation theory.« less
Dynamic modeling of Tampa Bay urban development using parallel computing
NASA Astrophysics Data System (ADS)
Xian, George; Crane, Mike; Steinwand, Dan
2005-08-01
Urban land use and land cover has changed significantly in the environs of Tampa Bay, Florida, over the past 50 years. Extensive urbanization has created substantial change to the region's landscape and ecosystems. This paper uses a dynamic urban-growth model, SLEUTH, which applies six geospatial data themes (slope, land use, exclusion, urban extent, transportation, hillside), to study the process of urbanization and associated land use and land cover change in the Tampa Bay area. To reduce processing time and complete the modeling process within an acceptable period, the model is recoded and ported to a Beowulf cluster. The parallel-processing computer system accomplishes the massive amount of computation the modeling simulation requires. SLEUTH calibration process for the Tampa Bay urban growth simulation spends only 10 h CPU time. The model predicts future land use/cover change trends for Tampa Bay from 1992 to 2025. Urban extent is predicted to double in the Tampa Bay watershed between 1992 and 2025. Results show an upward trend of urbanization at the expense of a decline of 58% and 80% in agriculture and forested lands, respectively.
Dynamic modeling of Tampa Bay urban development using parallel computing
Xian, G.; Crane, M.; Steinwand, D.
2005-01-01
Urban land use and land cover has changed significantly in the environs of Tampa Bay, Florida, over the past 50 years. Extensive urbanization has created substantial change to the region's landscape and ecosystems. This paper uses a dynamic urban-growth model, SLEUTH, which applies six geospatial data themes (slope, land use, exclusion, urban extent, transportation, hillside), to study the process of urbanization and associated land use and land cover change in the Tampa Bay area. To reduce processing time and complete the modeling process within an acceptable period, the model is recoded and ported to a Beowulf cluster. The parallel-processing computer system accomplishes the massive amount of computation the modeling simulation requires. SLEUTH calibration process for the Tampa Bay urban growth simulation spends only 10 h CPU time. The model predicts future land use/cover change trends for Tampa Bay from 1992 to 2025. Urban extent is predicted to double in the Tampa Bay watershed between 1992 and 2025. Results show an upward trend of urbanization at the expense of a decline of 58% and 80% in agriculture and forested lands, respectively. ?? 2005 Elsevier Ltd. All rights reserved.
Modeling the fracture of ice sheets on parallel computers.
Waisman, Haim; Bell, Robin; Keyes, David; Boman, Erik Gunnar; Tuminaro, Raymond Stephen
2010-03-01
The objective of this project is to investigate the complex fracture of ice and understand its role within larger ice sheet simulations and global climate change. At the present time, ice fracture is not explicitly considered within ice sheet models due in part to large computational costs associated with the accurate modeling of this complex phenomena. However, fracture not only plays an extremely important role in regional behavior but also influences ice dynamics over much larger zones in ways that are currently not well understood. Dramatic illustrations of fracture-induced phenomena most notably include the recent collapse of ice shelves in Antarctica (e.g. partial collapse of the Wilkins shelf in March of 2008 and the diminishing extent of the Larsen B shelf from 1998 to 2002). Other fracture examples include ice calving (fracture of icebergs) which is presently approximated in simplistic ways within ice sheet models, and the draining of supraglacial lakes through a complex network of cracks, a so called ice sheet plumbing system, that is believed to cause accelerated ice sheet flows due essentially to lubrication of the contact surface with the ground. These dramatic changes are emblematic of the ongoing change in the Earth's polar regions and highlight the important role of fracturing ice. To model ice fracture, a simulation capability will be designed centered around extended finite elements and solved by specialized multigrid methods on parallel computers. In addition, appropriate dynamic load balancing techniques will be employed to ensure an approximate equal amount of work for each processor.
Elliptical Orbit Performance Computer Program
NASA Technical Reports Server (NTRS)
Myler, T.
1984-01-01
Elliptical Orbit Performance (ELOPE) computer program for analyzing orbital performance of space boosters uses orbit insertion data obtained from trajectory simulation to generate parametric data on apogee and perigee altitudes as function of payload data. Data used to generate presentation plots that display elliptical orbit performance capability of space booster.
Distributed and parallel approach for handle and perform huge datasets
NASA Astrophysics Data System (ADS)
Konopko, Joanna
2015-12-01
Big Data refers to the dynamic, large and disparate volumes of data comes from many different sources (tools, machines, sensors, mobile devices) uncorrelated with each others. It requires new, innovative and scalable technology to collect, host and analytically process the vast amount of data. Proper architecture of the system that perform huge data sets is needed. In this paper, the comparison of distributed and parallel system architecture is presented on the example of MapReduce (MR) Hadoop platform and parallel database platform (DBMS). This paper also analyzes the problem of performing and handling valuable information from petabytes of data. The both paradigms: MapReduce and parallel DBMS are described and compared. The hybrid architecture approach is also proposed and could be used to solve the analyzed problem of storing and processing Big Data.
The path toward HEP High Performance Computing
NASA Astrophysics Data System (ADS)
Apostolakis, John; Brun, René; Carminati, Federico; Gheata, Andrei; Wenzel, Sandro
2014-06-01
High Energy Physics code has been known for making poor use of high performance computing architectures. Efforts in optimising HEP code on vector and RISC architectures have yield limited results and recent studies have shown that, on modern architectures, it achieves a performance between 10% and 50% of the peak one. Although several successful attempts have been made to port selected codes on GPUs, no major HEP code suite has a "High Performance" implementation. With LHC undergoing a major upgrade and a number of challenging experiments on the drawing board, HEP cannot any longer neglect the less-than-optimal performance of its code and it has to try making the best usage of the hardware. This activity is one of the foci of the SFT group at CERN, which hosts, among others, the Root and Geant4 project. The activity of the experiments is shared and coordinated via a Concurrency Forum, where the experience in optimising HEP code is presented and discussed. Another activity is the Geant-V project, centred on the development of a highperformance prototype for particle transport. Achieving a good concurrency level on the emerging parallel architectures without a complete redesign of the framework can only be done by parallelizing at event level, or with a much larger effort at track level. Apart the shareable data structures, this typically implies a multiplication factor in terms of memory consumption compared to the single threaded version, together with sub-optimal handling of event processing tails. Besides this, the low level instruction pipelining of modern processors cannot be used efficiently to speedup the program. We have implemented a framework that allows scheduling vectors of particles to an arbitrary number of computing resources in a fine grain parallel approach. The talk will review the current optimisation activities within the SFT group with a particular emphasis on the development perspectives towards a simulation framework able to profit best from
Payne, J.L.; Hassan, B.
1998-09-01
Massively parallel computers have enabled the analyst to solve complicated flow fields (turbulent, chemically reacting) that were previously intractable. Calculations are presented using a massively parallel CFD code called SACCARA (Sandia Advanced Code for Compressible Aerothermodynamics Research and Analysis) currently under development at Sandia National Laboratories as part of the Department of Energy (DOE) Accelerated Strategic Computing Initiative (ASCI). Computations were made on a generic reentry vehicle in a hypersonic flowfield utilizing three different distributed parallel computers to assess the parallel efficiency of the code with increasing numbers of processors. The parallel efficiencies for the SACCARA code will be presented for cases using 1, 150, 100 and 500 processors. Computations were also made on a subsonic/transonic vehicle using both 236 and 521 processors on a grid containing approximately 14.7 million grid points. Ongoing and future plans to implement a parallel overset grid capability and couple SACCARA with other mechanics codes in a massively parallel environment are discussed.
An Efficient Objective Analysis System for Parallel Computers
NASA Technical Reports Server (NTRS)
Stobie, J.
1999-01-01
A new atmospheric objective analysis system designed for parallel computers will be described. The system can produce a global analysis (on a 1 X 1 lat-lon grid with 18 levels of heights and winds and 10 levels of moisture) using 120,000 observations in 17 minutes on 32 CPUs (SGI Origin 2000). No special parallel code is needed (e.g. MPI or multitasking) and the 32 CPUs do not have to be on the same platform. The system is totally portable and can run on several different architectures at once. In addition, the system can easily scale up to 100 or more CPUS. This will allow for much higher resolution and significant increases in input data. The system scales linearly as the number of observations and the number of grid points. The cost overhead in going from 1 to 32 CPUs is 18%. In addition, the analysis results are identical regardless of the number of processors used. This system has all the characteristics of optimal interpolation, combining detailed instrument and first guess error statistics to produce the best estimate of the atmospheric state. Static tests with a 2 X 2.5 resolution version of this system showed it's analysis increments are comparable to the latest NASA operational system including maintenance of mass-wind balance. Results from several months of cycling test in the Goddard EOS Data Assimilation System (GEOS DAS) show this new analysis retains the same level of agreement between the first guess and observations (O-F statistics) as the current operational system.
An Efficient Objective Analysis System for Parallel Computers
NASA Technical Reports Server (NTRS)
Stobie, James G.
1999-01-01
A new objective analysis system designed for parallel computers will be described. The system can produce a global analysis (on a 2 x 2.5 lat-lon grid with 20 levels of heights and winds and 10 levels of moisture) using 120,000 observations in less than 3 minutes on 32 CPUs (SGI Origin 2000). No special parallel code is needed (e.g. MPI or multitasking) and the 32 CPUs do not have to be on the same platform. The system Ls totally portable and can run on -several different architectures at once. In addition, the system can easily scale up to 100 or more CPUS. This will allow for much higher resolution and significant increases in input data. The system scales linearly as the number of observations and the number of grid points. The cost overhead in going from I to 32 CPus is 18%. in addition, the analysis results are identical regardless of the number of processors used. T'his system has all the characteristics of optimal interpolation, combining detailed instrument and first guess error statistics to produce the best estimate of the atmospheric state. It also includes a new quality control (buddy check) system. Static tests with the system showed it's analysis increments are comparable to the latest NASA operational system including maintenance of mass-wind balance. Results from a 2-month cycling test in the Goddard EOS Data Assimilation System (GEOS DAS) show this new analysis retains the same level of agreement between the first guess and observations (0-F statistics) throughout the entire two months.
Massively parallel computation of RCS with finite elements
NASA Technical Reports Server (NTRS)
Parker, Jay
1993-01-01
One of the promising combinations of finite element approaches for scattering problems uses Whitney edge elements, spherical vector wave-absorbing boundary conditions, and bi-conjugate gradient solution for the frequency-domain near field. Each of these approaches may be criticized. Low-order elements require high mesh density, but also result in fast, reliable iterative convergence. Spherical wave-absorbing boundary conditions require additional space to be meshed beyond the most minimal near-space region, but result in fully sparse, symmetric matrices which keep storage and solution times low. Iterative solution is somewhat unpredictable and unfriendly to multiple right-hand sides, yet we find it to be uniformly fast on large problems to date, given the other two approaches. Implementation of these approaches on a distributed memory, message passing machine yields huge dividends, as full scalability to the largest machines appears assured and iterative solution times are well-behaved for large problems. We present times and solutions for computed RCS for a conducting cube and composite permeability/conducting sphere on the Intel ipsc860 with up to 16 processors solving over 200,000 unknowns. We estimate problems of approximately 10 million unknowns, encompassing 1000 cubic wavelengths, may be attempted on a currently available 512 processor machine, but would be exceedingly tedious to prepare. The most severe bottlenecks are due to the slow rate of mesh generation on non-parallel machines and the large transfer time from such a machine to the parallel processor. One solution, in progress, is to create and then distribute a coarse mesh among the processors, followed by systematic refinement within each processor. Elimination of redundant node definitions at the mesh-partition surfaces, snap-to-surface post processing of the resulting mesh for good modelling of curved surfaces, and load-balancing redistribution of new elements after the refinement are auxiliary
Speed Scaling for Energy and Performance with Instantaneous Parallelism
NASA Astrophysics Data System (ADS)
Sun, Hongyang; He, Yuxiong; Hsu, Wen-Jing
We consider energy-performance tradeoff for scheduling parallel jobs on multiprocessors using dynamic speed scaling. The objective is to minimize the sum of energy consumption and certain performance metric, including makespan and total flow time. We focus on designing algorithms that are aware of the jobs' instantaneous parallelism but not their characteristics in the future. For total flow time plus energy, it is known that any algorithm that does not rely on instantaneous parallelism is Ω(ln 1/α P)-competitive, where P is the total number of processors. In this paper, we demonstrate the benefits of knowing instantaneous parallelism by presenting an O(1)-competitive algorithm. In the case of makespan plus energy, which is considered in the literature for the first time, we present an O(ln 1 - 1/α P)-competitive algorithm for batched jobs consisting of fully-parallel and sequential phases. We show that this algorithm is asymptotically optimal by providing a matching lower bound.
Accelerating Dust Storm Simulation by Balancing Task Allocation in Parallel Computing Environment
NASA Astrophysics Data System (ADS)
Gui, Z.; Yang, C.; XIA, J.; Huang, Q.; YU, M.
2013-12-01
Dust storm has serious negative impacts on environment, human health, and assets. The continuing global climate change has increased the frequency and intensity of dust storm in the past decades. To better understand and predict the distribution, intensity and structure of dust storm, a series of dust storm models have been developed, such as Dust Regional Atmospheric Model (DREAM), the NMM meteorological module (NMM-dust) and Chinese Unified Atmospheric Chemistry Environment for Dust (CUACE/Dust). The developments and applications of these models have contributed significantly to both scientific research and our daily life. However, dust storm simulation is a data and computing intensive process. Normally, a simulation for a single dust storm event may take several days or hours to run. It seriously impacts the timeliness of prediction and potential applications. To speed up the process, high performance computing is widely adopted. By partitioning a large study area into small subdomains according to their geographic location and executing them on different computing nodes in a parallel fashion, the computing performance can be significantly improved. Since spatiotemporal correlations exist in the geophysical process of dust storm simulation, each subdomain allocated to a node need to communicate with other geographically adjacent subdomains to exchange data. Inappropriate allocations may introduce imbalance task loads and unnecessary communications among computing nodes. Therefore, task allocation method is the key factor, which may impact the feasibility of the paralleling. The allocation algorithm needs to carefully leverage the computing cost and communication cost for each computing node to minimize total execution time and reduce overall communication cost for the entire system. This presentation introduces two algorithms for such allocation and compares them with evenly distributed allocation method. Specifically, 1) In order to get optimized solutions, a
SequenceL: Automated Parallel Algorithms Derived from CSP-NT Computational Laws
NASA Technical Reports Server (NTRS)
Cooke, Daniel; Rushton, Nelson
2013-01-01
With the introduction of new parallel architectures like the cell and multicore chips from IBM, Intel, AMD, and ARM, as well as the petascale processing available for highend computing, a larger number of programmers will need to write parallel codes. Adding the parallel control structure to the sequence, selection, and iterative control constructs increases the complexity of code development, which often results in increased development costs and decreased reliability. SequenceL is a high-level programming language that is, a programming language that is closer to a human s way of thinking than to a machine s. Historically, high-level languages have resulted in decreased development costs and increased reliability, at the expense of performance. In recent applications at JSC and in industry, SequenceL has demonstrated the usual advantages of high-level programming in terms of low cost and high reliability. SequenceL programs, however, have run at speeds typically comparable with, and in many cases faster than, their counterparts written in C and C++ when run on single-core processors. Moreover, SequenceL is able to generate parallel executables automatically for multicore hardware, gaining parallel speedups without any extra effort from the programmer beyond what is required to write the sequen tial/singlecore code. A SequenceL-to-C++ translator has been developed that automatically renders readable multithreaded C++ from a combination of a SequenceL program and sample data input. The SequenceL language is based on two fundamental computational laws, Consume-Simplify- Produce (CSP) and Normalize-Trans - pose (NT), which enable it to automate the creation of parallel algorithms from high-level code that has no annotations of parallelism whatsoever. In our anecdotal experience, SequenceL development has been in every case less costly than development of the same algorithm in sequential (that is, single-core, single process) C or C++, and an order of magnitude less
A class of parallel algorithms for computation of the manipulator inertia matrix
NASA Technical Reports Server (NTRS)
Fijany, Amir; Bejczy, Antal K.
1989-01-01
Parallel and parallel/pipeline algorithms for computation of the manipulator inertia matrix are presented. An algorithm based on composite rigid-body spatial inertia method, which provides better features for parallelization, is used for the computation of the inertia matrix. Two parallel algorithms are developed which achieve the time lower bound in computation. Also described is the mapping of these algorithms with topological variation on a two-dimensional processor array, with nearest-neighbor connection, and with cardinality variation on a linear processor array. An efficient parallel/pipeline algorithm for the linear array was also developed, but at significantly higher efficiency.
Parallel calculations on shared memory, NUMA-based computers using MATLAB
NASA Astrophysics Data System (ADS)
Krotkiewski, Marcin; Dabrowski, Marcin
2014-05-01
Achieving satisfactory computational performance in numerical simulations on modern computer architectures can be a complex task. Multi-core design makes it necessary to parallelize the code. Efficient parallelization on NUMA (Non-Uniform Memory Access) shared memory architectures necessitates explicit placement of the data in the memory close to the CPU that uses it. In addition, using more than 8 CPUs (~100 cores) requires a cluster solution of interconnected nodes, which involves (expensive) communication between the processors. It takes significant effort to overcome these challenges even when programming in low-level languages, which give the programmer full control over data placement and work distribution. Instead, many modelers use high-level tools such as MATLAB, which severely limit the optimization/tuning options available. Nonetheless, the advantage of programming simplicity and a large available code base can tip the scale in favor of MATLAB. We investigate whether MATLAB can be used for efficient, parallel computations on modern shared memory architectures. A common approach to performance optimization of MATLAB programs is to identify a bottleneck and migrate the corresponding code block to a MEX file implemented in, e.g. C. Instead, we aim at achieving a scalable parallel performance of MATLABs core functionality. Some of the MATLABs internal functions (e.g., bsxfun, sort, BLAS3, operations on vectors) are multi-threaded. Achieving high parallel efficiency of those may potentially improve the performance of significant portion of MATLABs code base. Since we do not have MATLABs source code, our performance tuning relies on the tools provided by the operating system alone. Most importantly, we use custom memory allocation routines, thread to CPU binding, and memory page migration. The performance tests are carried out on multi-socket shared memory systems (2- and 4-way Intel-based computers), as well as a Distributed Shared Memory machine with 96 CPU
Identifying logical planes formed of compute nodes of a subcommunicator in a parallel computer
Davis, Kristan D.; Faraj, Daniel A.
2016-03-01
In a parallel computer, a plurality of logical planes formed of compute nodes of a subcommunicator may be identified by: for each compute node of the subcommunicator and for a number of dimensions beginning with a first dimension: establishing, by a plane building node, in a positive direction of the first dimension, all logical planes that include the plane building node and compute nodes of the subcommunicator in a positive direction of a second dimension, where the second dimension is orthogonal to the first dimension; and establishing, by the plane building node, in a negative direction of the first dimension, all logical planes that include the plane building node and compute nodes of the subcommunicator in the positive direction of the second dimension.
Identifying logical planes formed of compute nodes of a subcommunicator in a parallel computer
Davis, Kristan D.; Faraj, Daniel
2016-05-03
In a parallel computer, a plurality of logical planes formed of compute nodes of a subcommunicator may be identified by: for each compute node of the subcommunicator and for a number of dimensions beginning with a first dimension: establishing, by a plane building node, in a positive direction of the first dimension, all logical planes that include the plane building node and compute nodes of the subcommunicator in a positive direction of a second dimension, where the second dimension is orthogonal to the first dimension; and establishing, by the plane building node, in a negative direction of the first dimension, all logical planes that include the plane building node and compute nodes of the subcommunicator in the positive direction of the second dimension.
Heterogeneous transmission and parallel computing platform (HTPCP) for remote sensing applications
NASA Astrophysics Data System (ADS)
Guo, Yi; Rius, Antonio; Ribò, Serni; Ferrer, Carles
2011-05-01
The increasing campaigns of GNSS-R scenario have put great pressure on high performance post-processing design into the space level instrumentation. Due to large scale of information acquisition and the intensive computation of cross-correlation waveform (CC-WAV), the overhead between the processing time and the storage of amount of data prior to downlink issues has lead us to get the solution of real-time parallel processing design on board. In this paper, we focus on the interaction of the chip level multiprocessing architecture and applications, which show that the unbalanced workload of the transmission and processing can be compensated on the novel architecture, Heterogeneous Transmission and Parallel Computing Platform (HTPCP). The intention of HTPCP is to get a solution for the bus congestion and memory allocation issues. The pros and cons of SMP and HTPCP are discussed, and the simulation results prove that HTPCP can highly improve the throughput of the GOLD-RTR system.
Choi, Jaeyoung; Walker, D.W.; Dongarra, J.J. |
1993-08-01
This paper describes the Parallel Universal Matrix Multiplication Algorithms (PUMMA) on distributed memory concurrent computers. The PUMMA package includes not only the non-transposed matrix multiplication routine C = A{center_dot}B, but also transposed multiplication routines C = A{sup T}{center_dot}B, C = A{center_dot}B{sup T}, and C = A{sup T}{center_dot}B{sup T}, for a block scattered data distribution. The routines perform efficiently for a wide range of processor configurations and block sizes. The PUMMA together provide the same functionality as the Level 3 BLAS routine xGEMM. Details of the parallel implementation of the routines are given, and results are presented for runs on the Intel Touchstone Delta computer.
NASA Technical Reports Server (NTRS)
Hsia, T. C.; Lu, G. Z.; Han, W. H.
1987-01-01
In advanced robot control problems, on-line computation of inverse Jacobian solution is frequently required. Parallel processing architecture is an effective way to reduce computation time. A parallel processing architecture is developed for the inverse Jacobian (inverse differential kinematic equation) of the PUMA arm. The proposed pipeline/parallel algorithm can be inplemented on an IC chip using systolic linear arrays. This implementation requires 27 processing cells and 25 time units. Computation time is thus significantly reduced.
McLay, R.T.; Carey, G.F.
1996-12-31
In this study we consider parallel solution of sparse linear systems arising from discretized PDE`s. As part of our continuing work on our parallel PCG Solver package, we have made improvements in two areas. The first is improving the performance of the matrix-vector product. Here on regular finite-difference grids, we are able to use the cache memory more efficiently for smaller domains or where there are multiple degrees of freedom. The second problem of interest in the present work is the construction of preconditioners in the context of the parallel PCG solver we are developing. Here the problem is partitioned over a set of processors subdomains and the matrix-vector product for PCG is carried out in parallel for overlapping grid subblocks. For problems of scaled speedup, the actual rate of convergence of the unpreconditioned system deteriorates as the mesh is refined. Multigrid and subdomain strategies provide a logical approach to resolving the problem. We consider the parallel trade-offs between communication and computation and provide a complexity analysis of a representative algorithm. Some preliminary calculations using the parallel package and comparisons with other preconditioners are provided together with parallel performance results.
Parallel computations on pedigree data through mapping to configurable computing devices
Henshall, John M; Little, Bryce Alvin
2006-01-01
Pedigree data structures have a number of applications in genetics, including the estimation of allelic or haplotype probabilities in humans and agricultural species, and the estimation of breeding values in agricultural species. Sequential algorithms for general purpose CPU-based computers are commonly used, but are inadequate for some tasks on large data sets. We show that pedigree data can be directly represented on Field Programmable Gate Arrays (FPGA), allowing highly efficient massively parallel simulation of the flow of genes. Operating on the whole pedigree in parallel, the transmission of genes can occur for all individuals in a single clock cycle. By using FPGA, the algorithms to estimate inbreeding coefficients and allelic probabilities are shown to operate hundreds to thousands of times faster than the corresponding sequentially based algorithms. Where problems can be largely represented in an integer form, FPGA provide an efficient platform for computations on pedigree data. PMID:16635449