SEARCH · Search NASA
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.
The parallel and pipeline organization of fast unitary transform algorithms such as the Fast Fourier Transform are discussed. The efficiency is pointed out of a combined parallel-pipeline processor of a transform such as the Haar transform in which 2 to the n minus 1 power hardware butterflies generate a transform of order 2 to the n power every computation cycle.
This paper is concerned with the implementation of a 3D unstructured-grid Euler-solver on massively parallel distributed-memory computer architectures. The goal is to minimize solution time by achieving high computational rates with a numerically efficient algorithm. An unstructured multigrid algorithm with an edge-based data-structure has been adopted, and a number of optimizations have been devised and implemented in order to accelerate the parallel computational rates. The implementation is carried out by creating a set of software tools, which ease the implementation of computational problems on parallel architecture machines by relieving the user of the low-level machine specific issues. The quantitative effect of the various optimizations are demonstrated, and we show that the combined effect of these optimizations leads to roughly a factor of three performance improvement. The overall solution efficiency is compared with that obtained on the CRAY-YMP vector supercomputer.
A three-dimensional electromagnetic PIC code has been developed on the 512 node Intel Touchstone Delta MIMD parallel computer. This code is based on the General Concurrent PIC algorithm which uses a domain decomposition to divide the computation among the processors. The 3D simulation domain can be partitioned into 1-, 2-, or 3-dimensional subdomains. Particles must be exchanged between processors as they move among the subdomains.
There has been significant improvement in the performance of VLSI devices, in terms of size, power consumption, and speed, in recent years and this trend may also continue for some near future. However, it is a well known fact that there are major obstacles, i.e., physical limitation of feature size reduction and ever increasing cost of foundry, that would prevent the long term continuation of this trend. This has motivated the exploration of some fundamentally new technologies that are not dependent on the conventional feature size approach. Such technologies are expected to enable scaling to continue to the ultimate level, i.e., molecular and atomistic size. Quantum computing, quantum dot-based computing, DNA based computing, biologically inspired computing, etc., are examples of such new technologies. In particular, quantum-dots based computing by using Quantum-dot Cellular Automata (QCA) has recently been intensely investigated as a promising new technology capable of offering significant improvement over conventional VLSI in terms of reduction of feature size (and hence increase in integration level), reduction of power consumption, and increase of switching speed. Quantum dot-based computing and memory in general and QCA specifically, are intriguing to NASA due to their high packing density (10(exp 11) - 10(exp 12) per square cm ) and low power consumption (no transfer of current) and potentially higher radiation tolerant. Under Revolutionary Computing Technology (RTC) Program at the NASA/JPL Center for Integrated Space Microelectronics (CISM), we have been investigating the potential applications of QCA for the space program. To this end, exploiting the intrinsic features of QCA, we have designed novel QCA-based circuits for co-planner (i.e., single layer) and compact implementation of a class of data permutation matrices, a class of interconnection networks, and a bit-serial processor. Building upon these circuits, we have developed novel algorithms and QCA-based architectures for highly parallel and systolic computation of signal/image processing applications, such as FFT and Wavelet and Wlash-Hadamard Transforms.
Most of the recently proposed computational methods for solving partial differential equations on multiprocessor architectures stem from the 'divide and conquer' paradigm and involve some form of domain decomposition. For those methods which also require grids of points or patches of elements, it is often necessary to explicitly partition the underlying mesh, especially when working with local memory parallel processors. In this paper, a family of cost-effective algorithms for the automatic partitioning of arbitrary two- and three-dimensional finite element and finite difference meshes is presented and discussed in view of a domain decomposed solution procedure and parallel processing. The influence of the algorithmic aspects of a solution method (implicit/explicit computations), and the architectural specifics of a multiprocessor (SIMD/MIMD, startup/transmission time), on the design of a mesh partitioning algorithm are discussed. The impact of the partitioning strategy on load balancing, operation count, operator conditioning, rate of convergence and processor mapping is also addressed. Finally, the proposed mesh decomposition algorithms are demonstrated with realistic examples of finite element, finite volume, and finite difference meshes associated with the parallel solution of solid and fluid mechanics problems on the iPSC/2 and iPSC/860 multiprocessors.
The Lanczos (1950), multisectioning, and subspace iteration sequential methods for vibration analysis presently used as bases for three parallel algorithms are noted, in the aftermath of three example problems, to maintain reasonable accuracy in the computation of vibration frequencies. Significant computation time reductions are obtained as the number of processors increases. An analysis is made of the performance of each method, in order to characterize relative strengths and weaknesses as well as to identify those parameters that most strongly affect computation efficiency.
Multidisciplinary design optimization of aircraft will require unprecedented capabilities of both analysis software and computer hardware. The speed and accuracy of the analysis will depend heavily on the computational fluid dynamics (CFD) module which is used. A new CFD module has been developed to combine the robust accuracy of conventional codes with the ability to run on parallel architectures. This is achieved by parallelizing the ARC3D algorithm, a central-differenced Navier-Stokes method, on the Intel iPSC/860. The computed solutions are identical to those from conventional machines. Computational speed on 64 processors is comparable to the rate on one Cray Y-MP processor and will increase as new generations of parallel computers become available.
Over recent years, Adaptive Mesh Refinement (AMR) algorithms which dynamically match the local resolution of the computational grid to the numerical solution being sought have emerged as powerful tools for solving problems that contain disparate length and time scales. In particular, several workers have demonstrated the effectiveness of employing an adaptive, block-structured hierarchical grid system for simulations of complex shock wave phenomena. Unfortunately, from the parallel algorithm developer's viewpoint, this class of scheme is quite involved; these schemes cannot be distilled down to a small kernel upon which various parallelizing strategies may be tested. However, because of their block-structured nature such schemes are inherently parallel, so all is not lost. In this paper we describe the method by which Quirk's AMR algorithm has been parallelized. This method is built upon just a few simple message passing routines and so it may be implemented across a broad class of MIMD machines. Moreover, the method of parallelization is such that the original serial code is left virtually intact, and so we are left with just a single product to support. The importance of this fact should not be underestimated given the size and complexity of the original algorithm.
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.
In geophysics, we are confronted with many under-determined inverse problems. For example, all of our observations of earthquakes are made at the Earth’s surface. So, when we try to infer how slip during an earthquake evolves in space and time, we find that there are many potential slip histories that are consistent with our limited observations and our understanding of earthquake physics. One way to approach these problems is with Bayesian analysis which allows us to infer the ensemble of all potential slip models that satisfy the observations and our prior knowledge of earthquake physics. In Bayesian analysis, our prior knowledge is known as the prior probability density function or prior PDF, the fit to the data is known as the data likelihood, and the target PDF that satisfies both the prior PDF and data likelihood is known as the posterior PDF. However, simulating the posterior PDF typically requires using Markov Chain Monte Carlo (MCMC) to draw tens of billions of random realizations of earthquake slip models, which may not be computationally feasible. To make this and similar geophysical inversions computationally tractable, we developed the Cascading Adaptive Transitional Metropolis In Parallel (CATMIP) algorithm. CATMIP is an efficient parallel Markov Chain Monte Carlo (MCMC) sampler that is used for model fitting and uncertainty quantification in geophysics. Example use cases are earthquake rupture modeling, determining mineral composition on Mars, reconstructing the history of ocean salinity, and historical earthquake relocation. CATMIP employs many parallel instances of the Metropolis algorithm for sampling in a transitioning framework. Transitioning is a process in which a set of random samples at equilibrium with a known probability density function (PDF) are used as seeds for the Markov chains to sample successive target PDFs that incrementally move the distribution from the starting seeds to the final desired PDF that describes the relative plausibility of potential values for the model parameters. The algorithm is implemented as a Master-Worker model employing MPI for communication. The worker processes are loosely coupled with global parameters periodically optimized by the master process. This provides a very high amount of parallelism with little communication between updates. During the presentation we will discuss the history of the algorithm and elaborate the earthquake rupture modeling use case for the CATMIP package. Our first step toward GPU optimization was to optimize the code for the CPU. CPU profiling revealed that most of the compute time is spent in calls to level 2 BLAS routines and calls to GSL random number generators. We revised the algorithm to employ level 3 BLAS routines instead. In our presentation we will describe how this was accomplished. Adding GPU support to CATMIP consisted mostly of replacing the calls to GSL with calls to GPU vendor-provided library routines. A small number of loops were directly implemented in CUDA. In the presentation will provide implementation details. Finally, we will discuss methods for profiling and opportunities for further optimizing GPU execution. By creating a code with the flexibility to run on either a CPU or GPU architecture, CATMIP can be used on systems ranging from large CPU-based HPC environments to single servers with GPU acceleration and everything in between.
The philosophy and data encoding possible in systolic array optical processor (SAOP) were reviewed. The multitude of linear algebraic operations achievable on this architecture is examined. These operations include such linear algebraic algorithms as: matrix-decomposition, direct and indirect solutions, implicit and explicit methods for partial differential equations, eigenvalue and eigenvector calculations, and singular value decomposition. This architecture can be utilized to realize general techniques for solving matrix linear and nonlinear algebraic equations, least mean square error solutions, FIR filters, and nested-loop algorithms for control engineering applications. The data flow and pipelining of operations, design of parallel algorithms and flexible architectures, application of these architectures to computationally intensive physical problems, error source modeling of optical processors, and matching of the computational needs of practical engineering problems to the capabilities of optical processors are emphasized.
A systematic procedure is developed for exploiting the parallel constructs of computation in a highly coupled, linear system application. An overall top-down design approach is adopted. Differential equations governing the application under consideration are partitioned into subtasks on the basis of a data flow analysis. The interconnected task units constitute a task graph which has to be computed in every update interval. Multiprocessing concepts utilizing parallel integration algorithms are then applied for efficient task graph execution. A simple scheduling routine is developed to handle task allocation while in the multiprocessor mode. Results of simulation and scheduling are compared on the basis of standard performance indices. Processor timing diagrams are developed on the basis of program output accruing to an optimal set of processors. Basic architectural attributes for implementing the system are discussed together with suggestions for processing element design. Emphasis is placed on flexible architectures capable of accommodating widely varying application specifics.
Described are four significant algorithms implemented on the massively parallel processor (MPP) at the Goddard Space Flight Center. Two are in the area of image analysis. Of the other two, one is a mathematical simulation experiment and the other deals with the efficient transfer of data between distantly separated processors in the MPP array. The first algorithm presented is the automatic determination of elevations from stereo pairs. The second algorithm solves mathematical logistic equations capable of producing both ordered and chaotic (or random) solutions. This work can potentially lead to the simulation of artificial life processes. The third algorithm is the automatic segmentation of images into reasonable regions based on some similarity criterion, while the fourth is an implementation of a bitonic sort of data which significantly overcomes the nearest neighbor interconnection constraints on the MPP for transferring data between distant processors.
The Intel Paragon is a mesh-connected distributed memory parallel computer. It uses an oblivious and deterministic message routing algorithm: this permits us to develop highly optimized schedules for frequently needed communication patterns. The complete exchange is one such pattern. Several approaches are available for carrying it out on the mesh. We study an algorithm developed by Scott. This algorithm assumes that a communication link can carry one message at a time and that a node can only transmit one message at a time. It requires global synchronization to enforce a schedule of transmissions. Unfortunately global synchronization has substantial overhead on the Paragon. At the same time the powerful interconnection mechanism of this machine permits 2 or 3 messages to share a communication link with minor overhead. It can also overlap multiple message transmission from the same node to some extent. We develop a generalization of Scott's algorithm that executes complete exchange with a prescribed contention. Schedules that incur greater contention require fewer synchronization steps. This permits us to tradeoff contention against synchronization overhead. We describe the performance of this algorithm and compare it with Scott's original algorithm as well as with a naive algorithm that does not take interconnection structure into account. The Bounded contention algorithm is always better than Scott's algorithm and outperforms the naive algorithm for all but the smallest message sizes. The naive algorithm fails to work on meshes larger than 12 x 12. These results show that due consideration of processor interconnect and machine performance parameters is necessary to obtain peak performance from the Paragon and its successor mesh machines.
This viewgraph presentation gives an overview of the future of computational fluid dynamics (CFD), which in the past has pioneered the field of flow simulation. Over time CFD has progressed as computing power. Numerical methods have been advanced as CPU and memory capacity increases. Complex configurations are routinely computed now and direct numerical simulations (DNS) and large eddy simulations (LES) are used to study turbulence. As the computing resources changed to parallel and distributed platforms, computer science aspects such as scalability (algorithmic and implementation) and portability and transparent codings have advanced. Examples of potential future (or current) challenges include risk assessment, limitations of the heuristic model, and the development of CFD and information technology (IT) tools.
This paper describes a new algorithm that improves the performance of application-controlled demand paging for the out-of-core visualization of data sets that are on either local disks or disks on remote servers. The performance improvements come from better overlapping the computation with the page reading process, and by performing multiple page reads in parallel. The new algorithm can be applied to many different visualization algorithms since application-controlled demand paging is not specific to any visualization algorithm. The paper includes measurements that show that the new multi-threaded paging algorithm decreases the time needed to compute visualizations by one third when using one processor and reading data from local disk. The time needed when using one processor and reading data from remote disk decreased by up to 60%. Visualization runs using data from remote disk ran about as fast as ones using data from local disk because the remote runs were able to make use of the remote server's high performance disk array.
Time-accurate simulation of turbulent flames in high Reynolds number flows is a challenging task since both fluid dynamics and combustion must be modeled accurately. To numerically simulate this phenomenon, very large computer resources (both time and memory) are required. Although current vector supercomputers are capable of providing adequate resources for simulations of this nature, the high cost and their limited availability, makes practical use of such machines less than satisfactory. At the same time, the explicit time integration algorithms used in unsteady flow simulations often possess a very high degree of parallelism, making them very amenable to efficient implementation on large-scale parallel computers. Under these circumstances, distributed memory parallel computers offer an excellent near-term solution for greatly increased computational speed and memory, at a cost that may render the unsteady simulations of the type discussed above more feasible and affordable.This paper discusses the study of unsteady turbulent flames using a simulation algorithm that is capable of retaining high parallel efficiency on distributed memory parallel architectures. Numerical studies are carried out using large-eddy simulation (LES). In LES, the scales larger than the grid are computed using a time- and space-accurate scheme, while the unresolved small scales are modeled using eddy viscosity based subgrid models. This is acceptable for the moment/energy closure since the small scales primarily provide a dissipative mechanism for the energy transferred from the large scales. However, for combustion to occur, the species must first undergo mixing at the small scales and then come into molecular contact. Therefore, global models cannot be used. Recently, a new model for turbulent combustion was developed, in which the combustion is modeled, within the subgrid (small-scales) using a methodology that simulates the mixing and the molecular transport and the chemical kinetics within each LES grid cell. Finite-rate kinetics can be included without any closure and this approach actually provides a means to predict the turbulent rates and the turbulent flame speed. The subgrid combustion model requires resolution of the local time scales associated with small-scale mixing, molecular diffusion and chemical kinetics and, therefore, within each grid cell, a significant amount of computations must be carried out before the large-scale (LES resolved) effects are incorporated. Therefore, this approach is uniquely suited for parallel processing and has been implemented on various systems such as: Intel Paragon, IBM SP-2, Cray T3D and SGI Power Challenge (PC) using the system independent Message Passing Interface (MPI) compiler. In this paper, timing data on these machines is reported along with some characteristic results.
In this study we propose a novel method to parallelize high-order compact numerical algorithms for the solution of three-dimensional PDEs (Partial Differential Equations) in a space-time domain. For this numerical integration most of the computer time is spent in computation of spatial derivatives at each stage of the Runge-Kutta temporal update. The most efficient direct method to compute spatial derivatives on a serial computer is a version of Gaussian elimination for narrow linear banded systems known as the Thomas algorithm. In a straightforward pipelined implementation of the Thomas algorithm processors are idle due to the forward and backward recurrences of the Thomas algorithm. To utilize processors during this time, we propose to use them for either non-local data independent computations, solving lines in the next spatial direction, or local data-dependent computations by the Runge-Kutta method. To achieve this goal, control of processor communication and computations by a static schedule is adopted. Thus, our parallel code is driven by a communication and computation schedule instead of the usual "creative, programming" approach. The obtained parallelization speed-up of the novel algorithm is about twice as much as that for the standard pipelined algorithm and close to that for the explicit DRP algorithm.