Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallel computers.”

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

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

At least 1,063 records · Page 59

Parallel and Distributed Computational Fluid Dynamics: Experimental Results and Challenges

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.

Djomehri, Mohammad Jahed↗

Robust and efficient overset grid assembly for partitioned unstructured meshes

This paper presents a method to perform efficient and automated Overset Grid Assembly (OGA) on a system of overlapping unstructured meshes in a parallel computing environment where all meshes are partitioned into multiple mesh-blocks and processed on multiple cores. The main task of the overset grid assembler is to identify, in parallel, among all points in the overlapping mesh system, at which points the flow solution should be computed (field points), interpolated (receptor points), or ignored (hole points). Point containment search or donor search, an algorithm to efficiently determine the cell that contains a given point, is the core procedure necessary for accomplishing this task. Donor search is particularly challenging for partitioned unstructured meshes because of the complex irregular boundaries that are often created during partitioning. Another challenge arises because of the large variation in the type of mesh-block overlap and the resulting large load imbalance on multiple processors. Desirable traits for the grid assembly method are efficiency (requiring only a small fraction of the solver time), robustness (correct identification of all point types), and full automation (no user input required other than the mesh system). Additionally, the method should be scalable, which is an important challenge due to the inherent load imbalance. This paper describes a fully-automated grid assembly method, which can use two different donor search algorithms. One is based on the use of auxiliary grids and Exact Inverse Maps (EIM), and the other is based on the use of Alternating Digital Trees (ADT). The EIM method is demonstrated to be more efficient than the ADT method, while retaining robustness. An adaptive load re-balance algorithm is also designed and implemented, which considerably improves the scalability of the method.

97 MATHEMATICS AND COMPUTING↗

Electric Grid Expansion Planning with High Levels of Variable Generation

Renewables are taking a large proportion of generation capacity in U.S. power grids. As their randomness has increasing influence on power system operation, it is necessary to consider their impact on system expansion planning. To this end, this project studies the generation and transmission expansion co-optimization problem of the US Eastern Interconnection (EI) power grid with a high wind power penetration rate. In this project, the generation and transmission expansion problem for the EI system is modeled as a mixed-integer programming (MIP) problem. This study analyzed a time series creation method to capture the diversity of load and wind power across balancing regions in the EI system. The obtained time series can be easily introduced into the MIP co-optimization problem and then solved robustly through available MIP solvers. Simulation results show that the proposed time series generation method and the expansion co-optimization model and can improve the expansion result significantly after considering the diversity of wind and load across EI regions. The improved expansion plan that combines generation and transmission will aid system planners and policy makers to maximize the social welfare. This study shows that modelling load and wind variations and diversities across balancing regions will produce significantly different expansion result compared with former studies. For example, if wind is modeled in more details (by increasing the number of wind output levels) so that more wind blocks are considered in expansion planning, transmission expansion will be larger and the expansion timing will be earlier. Regarding generation expansion, more wind scenarios will slightly reduce wind generation expansion in the EI system and increase the expansion of other generation such as gas. Also, adopting detailed wind scenarios will reveal that it may be uneconomic to expand transmission networks for transmitting a large amount of wind power through a long distance in the EI system. Incorporating more details of renewables in expansion planning will inevitably increase the computational burden. Therefore, high performance computing (HPC) techniques are urgently needed for power system operation and planning optimization. As a scoping study task, this project tested some preliminary parallel computation techniques such as breaking down the simulation task into several sub-tasks based on chronology splitting or sample splitting, and then assigning these sub-tasks to different cores. Testing results show significant time reduction when a simulation task is split into several sub-tasks for parallel execution.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Generation and analysis of large reliability models

An effort has been underway for several years at NASA's Langley Research Center to extend the capability of Markov modeling techniques for reliability analysis to the designers of highly reliable avionic systems. This effort has been focused in the areas of increased model abstraction and increased computational capability. The reliability model generator (RMG), a software tool which uses as input a graphical, object-oriented block diagram of the system, is discussed. RMG uses an automated failure modes-effects analysis algorithm to produce the reliability model from the graphical description. Also considered is the ASSURE software tool, a parallel processing program which uses the ASSIST modeling language and SURE semi-Markov solution technique. An executable failure modes-effects analysis is used by ASSURE. The successful combination of the power of graphical representation, automated model generation, and parallel computation leads to the conclusion that large system architectures can now be analyzed.

Aerospace electronics↗

Neutronics Calculation of Fast Reactor Using Modern Computing Systems

The possibility of using the LUCKY-A three-dimensional multiprocessor transport code as a neutronics module for the COREMELT integrated non-stationary code for performing safety analysis, which is used for fast sodium reactor safety validation, is discussed. In this application, more stringent requirements are imposed on the computing speed of the LUCKY-A code. To test LUCKY-A possibilities, a typical computational problem is examined in a formulation characteristic for the proposed use of the code. The possibilities of the code for parallelized calculations are shown, and a comparison is made with calculations performed with the MMKKENO and TRIGEX codes. The efficacy of a parallel computational process is investigated as a function of the number of computing kernels used.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Poster — Thur Eve — 74: Distributed, asynchronous, reactive dosimetric and outcomes analysis using DICOMautomaton

Many have speculated about the future of computational technology in clinical radiation oncology. It has been advocated that the next generation of computational infrastructure will improve on the current generation by incorporating richer aspects of automation, more heavily and seamlessly featuring distributed and parallel computation, and providing more flexibility toward aggregate data analysis. In this report we describe how a recently created — but currently existing — analysis framework (DICOMautomaton) incorporates these aspects. DICOMautomaton supports a variety of use cases but is especially suited for dosimetric outcomes correlation analysis, investigation and comparison of radiotherapy treatment efficacy, and dose-volume computation. We describe: how it overcomes computational bottlenecks by distributing workload across a network of machines; how modern, asynchronous computational techniques are used to reduce blocking and avoid unnecessary computation; and how issues of out-of-date data are addressed using reactive programming techniques and data dependency chains. We describe internal architecture of the software and give a detailed demonstration of how DICOMautomaton could be used to search for correlations between dosimetric and outcomes data.

60 APPLIED LIFE SCIENCES↗

A Length Adaptive Algorithm-Hardware Co-design of Transformer on FPGA Through Sparse Attention and Dynamic Pipelining

Transformers are considered one of the most important deep learning models since 2018, in part because it establishes state-of-the-art (SOTA) records and could potentially replace existing Deep Neural Networks (DNNs). Despite the remarkable triumphs, the prolonged turnaround time of Transformer models is a widely recognized roadblock. The variety of sequence lengths imposes additional computing overhead where inputs need to be zero-padded to the maximum sentence length in the batch to accommodate the parallel computing platforms. This paper targets the field-programmable gate array (FPGA) and proposes a coherent sequence length adaptive algorithm–hardware co-design for Transformer acceleration. Particularly, we develop a hardware-friendly sparse attention operator and a length-aware hardware resource scheduling algorithm. The proposed sparse attention operator brings the complexity of attention-based models down to linear complexity and alleviates the off-chip memory traffic. The proposed length-aware resource hardware scheduling algorithm dynamically allocates the hardware resources to fill up the pipeline slots and eliminates bubbles for NLP tasks. Experiments show that our design has very small accuracy loss and has 80.2 × and 2.6 × speedup compared to CPU and GPU implementation, and 4 × higher energy efficiency than state-of-the-art GPU accelerator optimized via CUBLAS GEMM.

Peng, Hongwu↗

Random Number Generation for Petascale Quantum Monte Carlo

The quality of random number generators can affect the results of Monte Carlo computations, especially when a large number of random numbers are consumed. Furthermore, correlations present between different random number streams in a parallel computation can further affect the results. The SPRNG software, which the author had developed earlier, has pseudo-random number generators (PRNGs) capable of producing large numbers of streams with large periods. However, they had been empirically tested on only thousand streams earlier. In the work summarized here, we tested the SPRNG generators with over a hundred thousand streams, involving over 10^14 random numbers per test, on some tests. We also tested the popular Mersenne Twister. We believe that these are the largest tests of PRNGs, both in terms of the numbers of streams tested and the number of random numbers tested. We observed defects in some of these generators, including the Mersenne Twister, while a few generators appeared to perform well. We also corrected an error in the implementation of one of the SPRNG generators.

97 MATHEMATICS AND COMPUTING↗

An O(log sup 2 N) parallel algorithm for computing the eigenvalues of a symmetric tridiagonal matrix

An O(log sup 2 N) parallel algorithm is presented for computing the eigenvalues of a symmetric tridiagonal matrix using a parallel algorithm for computing the zeros of the characteristic polynomial. The method is based on a quadratic recurrence in which the characteristic polynomial is constructed on a binary tree from polynomials whose degree doubles at each level. Intervals that contain exactly one zero are determined by the zeros of polynomials at the previous level which ensures that different processors compute different zeros. The exact behavior of the polynomials at the interval endpoints is used to eliminate the usual problems induced by finite precision arithmetic.

Swarztrauber, Paul N.↗

A computational framework for uncertainty quantification and stochastic optimization in unit commitment with wind power generation.

We present a computational framework for integrating a state-of-the-art numerical weather prediction (NWP) model in stochastic unit commitment/economic dispatch formulations that account for wind power uncertainty. We first enhance the NWP model with an ensemble-based uncertainty quantification strategy implemented in a distributed-memory parallel computing architecture. We discuss computational issues arising in the implementation of the framework and validate the model using real wind-speed data obtained from a set of meteorological stations. We build a simulated power system to demonstrate the developments.

17 WIND ENERGY↗

Predicting Cost/Performance Trade-Offs for Whitney: A Commodity Computing Cluster

Recent advances in low-end processor and network technology have made it possible to build a "supercomputer" out of commodity components. We develop simple models of the NAS Parallel Benchmarks version 2 (NPB 2) to explore the cost/performance trade-offs involved in building a balanced parallel computer supporting a scientific workload. We develop closed form expressions detailing the number and size of messages sent by each benchmark. Coupling these with measured single processor performance, network latency, and network bandwidth, our models predict benchmark performance to within 30%. A comparison based on total system cost reveals that current commodity technology (200 MHz Pentium Pros with 100baseT Ethernet) is well balanced for the NPBs up to a total system cost of around $1,000,000.

Becker, Jeffrey C.↗

Predicting Cost/Performance Trade-offs For Whitney: A Commodity Computing Cluster

Recent advances in low-end processor and network technology have made it possible to build a "supercomputer" out of commodity components. We develop simple models of the NAS Parallel Benchmarks version 2 (NPB 2) to explore the cost/performance trade-offs involved in building a balanced parallel computer supporting a scientific workload. By measuring single processor benchmark performance, network latency, and network bandwidth, and using closed form expressions detailing the number and size of messages sent by each benchmark, our models predict benchmark performance to within 30%. A comparison based on total system cost reveals that current commodity technology (200 MHz Pentium Pros with 100baseT Ethernet) is well balanced for the NPBs up to a total system cost of around $ 1,000,000.

Becker, Jeffrey C.↗

Porting hypre to heterogeneous computer architectures: Strategies and experiences

We report that linear systems are occurring in many applications, and solving them can take a large amount of the total simulation time. The high performance library hypre provides a variety of interfaces and linear solvers, including various multigrid methods, that have achieved good scalability on a variety of homogeneous parallel computer architectures. Heterogeneous architectures with nodes that have both CPUs and accelerators provide new challenges, since they require more fine-grained parallelism and reduced data movement between different memories on a single node as well as across nodes. We will discuss our experiences and strategies to port hypre to heterogeneous computers with accelerators, including the design of a new memory model, the use of abstractions, the BoxLoop macros in the structured and semi-structured interfaces, and the restructuring of algebraic multigrid (AMG) into modular components. We present numerical experiments comparing CPU and GPU performance for several test problems.

97 MATHEMATICS AND COMPUTING↗

Scalable Knowledge Graph Analytics at 136 Petaflop/s

We are motivated by newly proposed methods for data mining large-scale corpora of scholarly publications, such as the full biomedical literature, which may consist of tens of millions of papers spanning decades of research. In this setting, analysts seek to discover how concepts relate to one another. They construct graph representations from annotated text databases and then formulate the relationship-mining problem as one of computing all-pairs shortest paths (APSP), which becomes a significant bottleneck. In this context, we present a new high-performance algorithm and implementation of the Floyd-Warshall algorithm for distributed-memory parallel computers accelerated by GPUs, which we call DSNAPSHOT (Distributed Accelerated Semiring All-Pairs Shortest Path). For our largest experiments, we ran DSNAPSHOT on a connected input graph with millions of vertices using 4, 096nodes (24,576GPUs) of the Oak Ridge National Laboratory's Summit supercomputer system. We find DSNAPSHOT achieves a sustained performance of 136×1015 floating-point operations per second (136petaflop/s) at a parallel efficiency of 90% under weak scaling and, in absolute speed, 70% of the best possible performance given our computation (in the single-precision tropical semiring or “min-plus” algebra). Looking forward, we believe this novel capability will enable the mining of scholarly knowledge corpora when embedded and integrated into artificial intelligence-driven natural language processing workflows at scale.

Kannan, Ramakrishnan {ramki}↗

RIACS/USRA

The Research Institute for Advanced Computer Science (RIACS) was established by the Universities Space Research Association (USRA) at the NASA Ames Research Center (ARC) on 6 June 1983. RIACS is privately operated by USRA, a consortium of universities with research programs in the aerospace sciences, under contract with NASA. The primary mission of RIACS is to provide research and expertise in computer science and scientific computing to support the scientific missions of NASA ARC. The research carried out at RIACS must change its emphasis from year to year in response to NASA ARC's changing needs and technological opportunities. A flexible scientific staff is provided through a university faculty visitor program, a post doctoral program, and a student visitor program. Not only does this provide appropriate expertise but it also introduces scientists outside of NASA to NASA problems. A small group of core RIACS staff provides continuity and interacts with an ARC technical monitor and scientific advisory group to determine the RIACS mission. RIACS activities are reviewed and monitored by a USRA advisory council and ARC technical monitor. Research at RIACS is currently being done in the following areas: Parallel Computing, Advanced Methods for Scientific Computing, High Performance Networks and Technology, and Learning Systems. Parallel compiler techniques, adaptive numerical methods for flows in complicated geometries, and optimization were identified as important problems to investigate for ARC's involvement in the Computational Grand Challenges of the next decade.

Oliger, Joseph↗

Machine‐learning‐based predictive control of nonlinear processes. Part II: Computational implementation

Abstract Machine learning is receiving more attention in classical engineering fields, and in particular, recurrent neural networks (RNNs) coupled with ensemble regression tools have demonstrated the capability of modeling nonlinear dynamic processes. In Part I of this two‐article series, the Lyapunov‐based model predictive control (LMPC) method using a single RNN model and an ensemble of RNN models, respectively, was rigorously developed for a general class of nonlinear systems. In the present article, computational implementation issues of this new control method ranging from training of the RNN models, ensemble regression of the RNN models, and parallel computation for accelerating the real‐time model calculations are addressed. Furthermore, a chemical reactor example is used to demonstrate the implementation and effectiveness of these machine‐learning tools in LMPC as well as compare them with standard state‐space model identification tools.

Wu, Zhe↗

Projective Hedging Algorithms for Multistage Stochastic Programming, Supporting Distributed and Asynchronous Implementation

Here we propose a decomposition algorithm for multistage stochastic programming that resembles the progressive hedging method of Rockafellar and Wets but is provably capable of several forms of asynchronous operation. We derive the method from a class of projective operator splitting methods fairly recently proposed by Combettes and Eckstein, significantly expanding the known applications of those methods. Our derivation assures convergence for convex problems whose feasible set is compact, subject to some standard regularity conditions and a mild “fairness” condition on subproblem selection. The method’s convergence guarantees are deterministic and do not require randomization, in contrast to other proposed asynchronous variations of progressive hedging. Computational experiments described in an online appendix show the method to outperform progressive hedging on large-scale problems in a highly parallel computing environment.

97 MATHEMATICS AND COMPUTING↗