SEARCH · Engineering Papers
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.
A parallel distributed memory implementation of intergrid communication for dynamic overset grids is presented. Included are discussions of various options considered during development. Results are presented comparing an Intel iPSC/860 to a single processor Cray Y-MP. Results for grids in relative motion show the iPSC/860 implementation to be faster than the Cray implementation.
Many parallel and distributed computing research results are obtained in simulation, using simulators that mimic real-world executions on some target system. Each such simulator is configured by picking values for parameters that define the behavior of the underlying simulation models it implements. The main concern for a simulator is accuracy: simulated behaviors should be as close as possible to those observed in the real-world target system. This requires that values for each of the simulator's parameters be carefully picked, or “calibrated,” based on ground-truth real-world executions. Examining the current state of the art shows that simulator calibration, at least in the field of parallel and distributed computing, is often undocumented (and thus perhaps often not performed) and, when documented, is described as a labor-intensive, manual process. In this work we evaluate the benefit of automating simulation calibration using simple algorithms. Specifically, we use a real-world case study from the field of High Energy Physics and compare automated calibration to calibration performed by a domain scientist. Our main finding is that automated calibration is on par with or significantly outperforms the calibration performed by the domain scientist. Furthermore, automated calibration makes it straightforward to operate desirable tradeoffs between simulation accuracy and simulation speed.
The developed methodology is aimed at improving the efficiency of executing block-structured algorithms on parallel, distributed, heterogeneous computers. The basic approach of these algorithms is to divide the flow domain into many sub- domains called blocks, and solve the governing equations over these blocks. Dynamic load balancing problem is defined as the efficient distribution of the blocks among the available processors over a period of several hours of computations. In environments with computers of different architecture, operating systems, CPU speed, memory size, load, and network speed, balancing the loads and managing the communication between processors becomes crucial. Load balancing software tools for mutually dependent parallel processes have been created to efficiently utilize an advanced computation environment and algorithms. These tools are dynamic in nature because of the chances in the computer environment during execution time. More recently, these tools were extended to a second operating system: NT. In this paper, the problems associated with this application will be discussed. Also, the developed algorithms were combined with the load sharing capability of LSF to efficiently utilize workstation clusters for parallel computing. Finally, results will be presented on running a NASA based code ADPAC to demonstrate the developed tools for dynamic load balancing.
One issue which is central in developing a general purpose Fast Fourier Transform (FFT) subroutine on a distributed memory parallel machine is the data distribution. It is possible that different users would like to use the FFT routine with different data distributions. Thus, there is a need to design FFT schemes on distributed memory parallel machines which can support a variety of data distributions. An FFT implementation on a distributed memory parallel machine which works for a number of data distributions commonly encountered in scientific applications is presented. The problem of rearranging the data after computing the FFT is also addressed. The performance of the implementation on a distributed memory parallel machine Intel iPSC/860 is evaluated.
Viewgraphs on automatic management of parallel and distributed system resources are presented. Topics covered include: parallel applications; intelligent management of multiprocessing systems; performance evaluation of parallel architecture; dynamic concurrent programs; compiler-directed system approach; lattice gaseous cellular automata; and sparse matrix Cholesky factorization.
We introduce an efficient distributed sequence parallel approach for training transformer-based deep learning image segmentation models. The neural network models are comprised of a combination of a Vision Transformer encoder with a convolutional decoder to provide image segmentation mappings. The utility of the distributed sequence parallel approach is especially useful in cases where the tokenized embedding representation of image data are too large to fit into standard computing hardware memory. To demonstrate the performance and characteristics of our models trained in sequence parallel fashion compared to standard models, we evaluate our approach using a 3D MRI brain tumor segmentation dataset. We show that training with a sequence parallel approach can match standard sequential model training in terms of convergence. Furthermore, we show that our sequence parallel approach has the capability to support training of models that would not be possible on standard computing resources.
The Eikonal equation is used to calculate wave propagation and distance fields, and due to its complexity requires numerical treatment for its solution. In this work, we present a second-order distributed memory parallel fast sweeping method. The second-order solution switches on a two-point stencil when two upwind points are available, and reverts to first-order otherwise. In all examples, the second-order method improves the solution over the first-order, allowing for significant savings in memory while achieving the same accuracy. Parallelization over distributed memory saw good weak scaling with optimal convergence. The computational time for second-order was approximately 2.5 times slower than first-order, where the largest amount of mesh points ran on 144 cores (512 GB) was ≈20 billion. The savings in memory from the second-order method combined with the distributed memory algorithm result in the ability to solve problems much larger than are possible with the serial first-order method.
Performance degradation in scientific computing on parallel and distributed computer systems can be caused by numerous factors. In this half-day tutorial we explain what are the important methodological issues involved in obtaining codes that have good performance potential. Then we discuss what are the possible obstacles in realizing that potential on contemporary hardware platforms, and give an overview of the software tools currently available for identifying the performance bottlenecks. Finally, some realistic examples are used to illustrate the actual use and utility of such tools.
We develop the first distributed-memory parallel implementation of Symmetric Nonnegative Matrix Factorization (SymNMF), a key data analytics kernel for clustering and dimensionality reduction. Our implementation includes two different algorithms for SymNMF, which give comparable results in terms of time and accuracy. The first algorithm is a parallelization of an existing sequential approach that uses solvers for non symmetric NMF. The second algorithm is a novel approach based on the Gauss-Newton method. It exploits second-order information without incurring large computational and memory costs. We evaluate the scalability of our algorithms on the Summit system at Oak Ridge National Laboratory, scaling up to 128 nodes (4,096 cores) with 70% efficiency. Additionally, we demonstrate our software on an image segmentation task.
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.
There are two sources of inaccuracy when simulating parallel and distributed computing systems: (i) a simulator implemented at an insufficient level of detail; and (ii) incorrectly calibrated simulation parameter values. Increasing the simulator’s level of detail can improve accuracy, but at the cost of higher space, time, and/or software complexity. Furthermore, evaluating the intrinsic accuracy of a simulator requires that its parameters be well-calibrated. Making decisions regarding the level of detail is thus challenging. We propose a methodology for instantiating the simulation calibration process and a framework for automating this process, which makes it possible to pick appropriate levels of detail for any simulator. We demonstrate the usefulness of our approach via two case studies for two different domains.
This thesis studies a forward recovery strategy using checkpointing and optimistic execution in parallel and distributed systems. The approach uses replicated tasks executing on different processors for forwared recovery and checkpoint comparison for error detection. To reduce overall redundancy, this approach employs a lower static redundancy in the common error-free situation to detect error than the standard N Module Redundancy scheme (NMR) does to mask off errors. For the rare occurrence of an error, this approach uses some extra redundancy for recovery. To reduce the run-time recovery overhead, look-ahead processes are used to advance computation speculatively and a rollback process is used to produce a diagnosis for correct look-ahead processes without rollback of the whole system. Both analytical and experimental evaluation have shown that this strategy can provide a nearly error-free execution time even under faults with a lower average redundancy than NMR.
The manual application of formal methods in system specification has produced successes, but in the end, despite any claims and assertions by practitioners, there is no provable relationship between a manually derived system specification or formal model and the customer's original requirements. Complex parallel and distributed system present the worst case implications for today s dearth of viable approaches for achieving system dependability. No avenue other than formal methods constitutes a serious contender for resolving the problem, and so recognition of requirements-based programming has come at a critical juncture. We describe a new, NASA-developed automated requirement-based programming method that can be applied to certain classes of systems, including complex parallel and distributed systems, to achieve a high degree of dependability.
We present a novel distributed union-find algorithm that features asynchronous parallelism and k-d tree based load balancing for scalable visualization and analysis of scientific data. Applications of union-find include level set extraction and critical point tracking, but distributed union-find can suffer from high synchronization costs and imbalanced workloads across parallel processes. In this study, we prove that global synchronizations in existing distributed union-find can be eliminated without changing final results, allowing overlapped communications and computations for scalable processing. We also use a k-d tree decomposition to redistribute inputs, in order to improve workload balancing. We benchmark the scalability of our algorithm with up to 1,024 processes using both synthetic and application data. Here, we demonstrate the use of our algorithm in critical point tracking and super-level set extraction with high-speed imaging experiments and fusion plasma simulations, respectively.
Recently, the DoD has sponsored work towards a new version of Ada, intended to support the construction of distributed systems. The revised version, often called Ada 9X, will become the new standard sometimes in the 1990s. It is intended that Ada 9X should provide language features giving limited support for distributed system construction. The requirements for such features are given. Many of the most advanced computer applications involve embedded systems that are comprised of parallel processors or networks of distributed computers. If Ada is to become the widely adopted language envisioned by many, it is essential that suitable compilers and tools be available to facilitate the creation of distributed and parallel Ada programs for these applications. The major languages issues impacting distributed and parallel programming are reviewed, and some principles upon which distributed/parallel language systems should be built are suggested. Based upon these, alternative language concepts for distributed/parallel programming are analyzed.
In this paper, we describe the design and implementation of a portable debugger for parallel and distributed programs. The design incorporates a client-server model in order to isolate non-portable debugger code from the user interface. The precise definition of a protocol for client-server interaction permits a high degree of portability of the client user interface. Replication of server components permits the implementation of a debugger for distributed computations. Portability across message passing implementations is achieved with a protocol that dictates the interaction between a message passing library and the debugger. This permits the same debugger to be used both on PVM and MTI programs. The process abstractions used for debugging message-passing programs can be easily adapted to debug HPF programs at the source level. This allows the debugger to present information hidden in tool-generated code in a meaningful manner.
The objective is to use the measurements of the equatorial particle distributions to obtain the parallel electric field structure and the evolution of the plasma distribution function along the field line. Appropriate uses of kinetic theory allows us to use the measured ( and inferred) particle distributions to obtain the electric field, and hence the variation on plasma density along the magnetic field line. The approach, here, is to utilize the adiabatic invariants, and assume the plasma distributions are in equilibrium.
Joint Nonnegative Matrix Factorization (JointNMF) is a hybrid method for mining information from datasets that contain both feature and connection information. We propose distributed-memory parallelizations of three algorithms for solving the JointNMF problem based on Alternating Nonnegative Least Squares, Projected Gradient Descent, and Projected Gauss-Newton. We extend well-known communication-avoiding algorithms using a single processor grid case to our coupled case on two processor grids. We demonstrate the scalability of the algorithms on up to 960 cores (40 nodes) with 60% parallel efficiency. The more sophisticated Alternating Nonnegative Least Squares (ANLS) and Gauss-Newton variants outperform the first-order gradient descent method in reducing the objective on large-scale problems. We perform a topic modelling task on a large corpus of academic papers that consists of over 37 million paper abstracts and nearly a billion citation relationships, demonstrating the utility and scalability of the methods.