SEARCH · Search NASA

Results for “Parallel Optimization Algorithms”

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.

SourceResearch connectionFrom yearThrough year

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

At least 667 records · Page 37

All sources NASA NTRS DOE OSTI

DOE OSTI2020

SQUARE: Strategic Quantum Ancilla Reuse for Modular Quantum Programs via Cost-Effective Uncomputation

Compiling high-level quantum programs to machines that are size constrained (i.e. limited number of quantum bits) and time constrained (i.e. limited number of quantum operations) is challenging. In this paper, we present SQUARE (Strategic QUantum Ancilla REuse), a compilation infrastructure that tackles allocation and reclamation of scratch qubits (called ancilla) in modular quantum programs. At its core, SQUARE strategically performs uncomputation to create opportunities for qubit reuse. Current Noisy Intermediate-Scale Quantum (NISQ) computers and forward-looking Fault-Tolerant (FT) quantum computers have fundamentally different constraints such as data locality, instruction parallelism, and communication overhead. Our heuristic-based ancilla-reuse algorithm balances these considerations and fits computations into resource-constrained NISQ or FT quantum machines, throttling parallelism when necessary. To precisely capture the workload of a program, we propose an improved metric, the "active quantum volume," and use this metric to evaluate the effectiveness of our algorithm. Furthermore, our results show that SQUARE improves the average success rate of NISQ applications by 1.47X. Surprisingly, the additional gates for uncomputation create ancilla with better locality, and result in substantially fewer swap gates and less gate noise overall. SQUARE also achieves an average reduction of 1.5X (and up to 9.6X) in active quantum volume for FT machines.

compiler optimization↗

NASA NTRS1996

Multigrid with Overlapping Patches

Solving boundary value problems with optimal efficiency requires adaptivity and multilevel techniques. Previously, an implementation of the AFACx algorithm is presented that is based on rectangular Cartesian grids. This implementation does not allow for the over]ap of grids that lie on the same level of refinement. We investigate the case in which these grids overlap. A standard technique for overlapping grids is the Schwarz algorithm. Some ways of using the Schwarz algorithm in a standard multigrid scheme are presented. Also, a problem that arises in some situations with non-aligned, overlapping grids is described. This situation comes up in a natural way when the Schwarz algorithm is used as a relaxation scheme within a multilevel algorithm. We identify the reason for the bad convergence and show that by more sophisticated interpolation the difficulties can be overcome. Then we present a multiplicative Schwarz algorithm for a large number of grids that has a high potential for parallelization. Finally we give some numerical results for the FACx algorithm with overlapping grids on each refinement level. The implementation of the described codes uses C++ and the array class libraries A++ and P++. Using the A++/P++ programming environment, it was possible to move from a serial code to a parallel code within a few days.

Berndt, Markus↗

DOE OSTI2025

A scalable multidimensional fully implicit solver for Hall magnetohydrodynamics

We propose an optimally performant fully implicit algorithm for the Hall magnetohydrodynamics (HMHD) equations based on multigrid-preconditioned Jacobian-free Newton-Krylov methods. HMHD is a challenging system to solve numerically because it supports stiff fast dispersive waves. The preconditioner is formulated using an operator-split approximate block factorization (Schur complement), informed by physics insight. We use a vector-potential formulation (instead of a magnetic field one) to allow a clean segregation of the problematic $\nabla$ x $\nabla$ x operator in the electron Ohm's law subsystem. This segregation allows the formulation of an effective damped block-Jacobi smoother for multigrid. We demonstrate by analysis that our proposed block-Jacobi iteration is convergent and has the smoothing property. The resulting HMHD solver is verified linearly with wave propagation examples, and nonlinearly with the GEM challenge reconnection problem by comparison against another HMHD code. We demonstrate the excellent algorithmic and parallel performance of the algorithm up to 16384 MPI tasks in two dimensions.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

NASA NTRS2017

Data Reduction Pipeline for the CHARIS Integral-Field Spectrograph I: Detector Readout Calibration and Data Cube Extraction

We present the data reduction pipeline for CHARIS, a high-contrast integral-field spectrograph for the Subaru Telescope. The pipeline constructs a ramp from the raw reads using the measured nonlinear pixel response and reconstructs the data cube using one of three extraction algorithms: aperture photometry, optimal extraction, or chi-squared fitting. We measure and apply both a detector flatfield and a lenslet flatfield and reconstruct the wavelength- and position-dependent lenslet point-spread function (PSF) from images taken with a tunable laser. We use these measured PSFs to implement a chi-squared-based extraction of the data cube, with typical residuals of approximately 5 percent due to imperfect models of the under-sampled lenslet PSFs. The full two-dimensional residual of the chi-squared extraction allows us to model and remove correlated read noise, dramatically improving CHARIS's performance. The chi-squared extraction produces a data cube that has been deconvolved with the line-spread function and never performs any interpolations of either the data or the individual lenslet spectra. The extracted data cube also includes uncertainties for each spatial and spectral measurement. CHARIS's software is parallelized, written in Python and Cython, and freely available on github with a separate documentation page. Astrometric and spectrophotometric calibrations of the data cubes and PSF subtraction will be treated in a forthcoming paper.

Brandt, Timothy D.↗

DOE OSTI2018

Center for Extended Magnetohydrodynamic Modeling (CEMM), Final Technical Report for UW-Madison Engineering Physics

The Center for Extended Magnetohydrodynamic Modeling (CEMM) focused on the further development, verification, validation, and application of two scientific computation programs, the NIMROD and M3D-C1 codes, for modeling macroscopic dynamics in magnetized plasma. Both NIMROD and M3D-C1 solve the mathematical equations that model two-fluid 3D magnetohydrodynamic (MHD). They are designed and optimized to be efficient and accurate for toroidal confinement physics calculations. The effort improved the fidelity of the physics models, the realism of the treatment of the surrounding vessel, coils, and other structures, the accuracy and efficiency of the algorithms, and the ability to scale calculations to thousands of processors in order to take advantage of leading-edge computers. The primary application areas of Resonant Magnetic Perturbations (RMP), Edge Harmonic Oscillations (EHO), Edge Localized Modes (ELMs), Disruption Studies, Sawteeth and Stationary States, and Kinetic MHD and Neoclassical Tearing Modes (NTM) are of concern for the international ITER experiment, which is under construction in France. The effort by the UW-Madison Engineering Physics component of CEMM improved the numerical methods in the NIMROD code with respect to modeling MHD interchange and two-fluid effects. It also improved the algorithm for solving large algebraic systems to enhance NIMROD's computational performance when running the code in parallel on increasing numbers of processor cores. The modeling efforts included a study of the nonlinear evolution of ballooning, and the findings were supported by analytical theory and computation. Numerical MHD modeling also examined the influences of two-fluid effects on ELMs and RMPs, where "two-fluid" refers to separation of electron dynamics from ion dynamics. Another related aspect considered the effects of a localized layer of electrical current density near the edge of the plasma confinement region. Two-fluid effects on resonant reconnecting instabilities were verified and applied to computations of sawtooth and drift-tearing dynamics.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2018

Big Data Mining for Assessing Calibration of Building Energy Models

Residential and commercial buildings in China, India, the United States (US), United Kingdom (UK), and Italy consume 39-45% of each nation's primary energy. Building energy models can be used to automatically optimize the return-on-investment for retrofits to improve a building’s energy efficiency. However, with an average of 3,000 building descriptors necessary to accurately simulate a single building, there is a market need to reduce the transaction cost for creating a more robust model for simulating every building in a city and accurately estimate savings prior to capital expenditures.We used two of the world’s fastest supercomputers, assembled unique datasets, and developed innovative algorithms for big data mining to assess different methods to create an accurate building energy model. The team has leveraged a total of eight high performance computing resources and 218 servers to analyze the best methods, metrics, and algorithms for creating an accurate building energy model beyond current industry standards necessary for private-sector financing. The project developed the world’s fastest buildings simulator, has completed over 8 million simulations totaling over 300TB, and mined this data with over 130,000 parallel artificial intelligence algorithms. This was used to quantify accuracy of AI-developed calibration algorithms with results that surpass industry standard guidelines and can identify individual building parameters to between 15% and 32% of their actual value.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

DOE OSTI2019

O(N) Quantum Molecular Dynamics (QMD) Algorithms, Code Requirements, Release Notes, and Library Deployment

The Exascale Computing Project (ECP) Co-design Center for Particle Applications (CoPA) is addressing the challenges for particle applications to run on upcoming exascale computing architectures. This includes short- and long-range molecular dynamics (MD), particle-in-cell (PIC), and quantum molecular dynamics (QMD). This document addresses $\mathcal{O}(N)$ QMD. The goal of this CoPA team is to develop $\mathcal{O}(N)$ linear scaling methods for QMD optimized for new and next generation architectures. This will provide the capability to simulate larger systems and run for longer timeframes than is possible today. Focus is initially on the density matrix construction. This requires efficient matrix formats and linear algebra operations for distributed architectures. A significant reduction in the wallclock time for an MD timestep is necessary to achieve practical large scale QMD. Equally challenging is to develop scalable distributed implementations. This requires the exploration of efficient data formats (dense, sparse, etc.), data decomposition and distribution across compute nodes using block- and graph-based methods. Traditional programming models such as MPI/OpenMP, hybrid MPI/OpenMP/Accelerator, as well as asynchronous task-based models will be explored to achieve maximum parallelism. Parallelism at the level of a matrix operation, solver, and/or application needs to be explored to gain the most benefit on new and emerging architectures. This report serves as a “living document” of the additions and enhancements throughout the ECP CoPA project for $\mathcal{O}(N)$ QMD. The following sections address the applications we are working with, the quantum chemistry libraries (BML and PROGRESS), and our co-design exploration through the use of a proxy application.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Optimizing High-Throughput Inference on Graph Neural Networks at Shared Computing Facilities with the NVIDIA Triton Inference Server

Abstract With machine learning applications now spanning a variety of computational tasks, multi-user shared computing facilities are devoting a rapidly increasing proportion of their resources to such algorithms. Graph neural networks (GNNs), for example, have provided astounding improvements in extracting complex signatures from data and are now widely used in a variety of applications, such as particle jet classification in high energy physics (HEP). However, GNNs also come with an enormous computational penalty that requires the use of GPUs to maintain reasonable throughput. At shared computing facilities, such as those used by physicists at Fermi National Accelerator Laboratory (Fermilab), methodical resource allocation and high throughput at the many-user scale are key to ensuring that resources are being used as efficiently as possible. These facilities, however, primarily provide CPU-only nodes, which proves detrimental to time-to-insight and computational throughput for workflows that include machine learning inference. In this work, we describe how a shared computing facility can use the NVIDIA Triton Inference Server to optimize its resource allocation and computing structure, recovering high throughput while scaling out to multiple users by massively parallelizing their machine learning inference. To demonstrate the effectiveness of this system in a realistic multi-user environment, we use the Fermilab Elastic Analysis Facility augmented with the Triton Inference Server to provide scalable and high-throughput access to a HEP-specific GNN and report on the outcome.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

DOE OSTI2019

Synchronous and asynchronous optimized Schwarz methods for one-way subdivision of bounded domains

Convergence of both synchronous and asynchronous optimized Schwarz algorithms for the shifted Laplacian operator on a bounded rectangular domain, in a one-way subdivision of the computational domain, with overlap, is shown. Convergence results are obtained under very mild conditions on the size of the subdomains and on the amount of overlap. A couple of results are also given, relating the convergence rate of the asynchronous method to changes in the size of the domain. Numerical experiments illustrate the theoretical results.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Comparing Parallel Plastic‐to‐X Pathways and Their Role in a Circular Economy for PET Bottles

Abstract The United States generates the most plastic waste of any country and is a top contributor to global plastic pollution. Multiple end‐of‐life strategies must be implemented to minimize environmental impacts and retain valuable plastic material, but it is challenging to compare options that generate products with different lifetimes and utilities. Herein, they present a material flow model equipped with consequential life cycle assessment, cost analysis, and a plastic circularity indicator that considers product quality and lifetime. The model is used to estimate the greenhouse gas (GHG) emissions, circularity, and cost of polyethylene terephthalate (PET) bottle mechanical downcycling to lower‐quality resin, closed‐loop glycolysis to food‐grade PET, upcycling to glass fiber‐reinforced plastic, and conversion to non‐plastic products (electricity, oil) on a United States economy‐wide basis for the year 2020. A brute force algorithm suggests that a combination of 68% glycolysis, 11% mechanical recycling, 6% upcycling, 9% landfilling, and 5% incineration can minimize the cost and GHG emissions and maximize the circularity of the current PET economy. However, uncertainty around transportation distances, materials recovery facility efficiencies, and recycling yields can result in different “optimal” pathway mixes. This flexible framework enables informed decision‐making to move toward a cost‐ and environment‐conscious circular economy for plastic.

09 BIOMASS FUELS↗

NASA NTRS1992

Improved interpretation of satellite altimeter data using genetic algorithms

Genetic algorithms (GA) are optimization techniques that are based on the mechanics of evolution and natural selection. They take advantage of the power of cumulative selection, in which successive incremental improvements in a solution structure become the basis for continued development. A GA is an iterative procedure that maintains a 'population' of 'organisms' (candidate solutions). Through successive 'generations' (iterations) the population as a whole improves in simulation of Darwin's 'survival of the fittest'. GA's have been shown to be successful where noise significantly reduces the ability of other search techniques to work effectively. Satellite altimetry provides useful information about oceanographic phenomena. It provides rapid global coverage of the oceans and is not as severely hampered by cloud cover as infrared imagery. Despite these and other benefits, several factors lead to significant difficulty in interpretation. The GA approach to the improved interpretation of satellite data involves the representation of the ocean surface model as a string of parameters or coefficients from the model. The GA searches in parallel, a population of such representations (organisms) to obtain the individual that is best suited to 'survive', that is, the fittest as measured with respect to some 'fitness' function. The fittest organism is the one that best represents the ocean surface model with respect to the altimeter data.

Messa, Kenneth↗

NASA NTRS2008

Automated Generation and Assessment of Autonomous Systems Test Cases

This slide presentation reviews some of the issues concerning verification and validation testing of autonomous spacecraft routinely culminates in the exploration of anomalous or faulted mission-like scenarios using the work involved during the Dawn mission's tests as examples. Prioritizing which scenarios to develop usually comes down to focusing on the most vulnerable areas and ensuring the best return on investment of test time. Rules-of-thumb strategies often come into play, such as injecting applicable anomalies prior to, during, and after system state changes; or, creating cases that ensure good safety-net algorithm coverage. Although experience and judgment in test selection can lead to high levels of confidence about the majority of a system's autonomy, it's likely that important test cases are overlooked. One method to fill in potential test coverage gaps is to automatically generate and execute test cases using algorithms that ensure desirable properties about the coverage. For example, generate cases for all possible fault monitors, and across all state change boundaries. Of course, the scope of coverage is determined by the test environment capabilities, where a faster-than-real-time, high-fidelity, software-only simulation would allow the broadest coverage. Even real-time systems that can be replicated and run in parallel, and that have reliable set-up and operations features provide an excellent resource for automated testing. Making detailed predictions for the outcome of such tests can be difficult, and when algorithmic means are employed to produce hundreds or even thousands of cases, generating predicts individually is impractical, and generating predicts with tools requires executable models of the design and environment that themselves require a complete test program. Therefore, evaluating the results of large number of mission scenario tests poses special challenges. A good approach to address this problem is to automatically score the results based on a range of metrics. Although the specific means of scoring depends highly on the application, the use of formal scoring - metrics has high value in identifying and prioritizing anomalies, and in presenting an overall picture of the state of the test program. In this paper we present a case study based on automatic generation and assessment of faulted test runs for the Dawn mission, and discuss its role in optimizing the allocation of resources for completing the test program.

Testing challenges↗

NASA NTRS2023

Easy, Scalable Subsetting of GEDI Point Clouds

The GEDI Subsetter, a Python tool developed for NASA’s Multi-mission Algorithm and Analysis Platform (MAAP), optimizes the accessibility and visualization of GEDI point clouds by enabling users to efficiently subset data in a convenient, scalable manner. Complex science data often requires users to learn new software skills and handle many large files. Handling and cleaning large data sets is tedious and error-prone. These challenges significantly impede analysis. One of the goals of NASA's MAAP is to provide a platform that lowers the barrier to conducting research and analysis at scale. When a group of MAAP users wanted to conduct above-ground biomass estimation using GEDI data, we found that their existing workflow for leveraging GEDI data suffered from the barriers mentioned above. Furthermore, their workflow did not scale easily beyond a small number of granules. We found that existing tools related to GEDI data retrieval and subsetting were too limiting, so the GEDI Subsetter was written to support MAAP users’ needs. Being able to run many subsetting jobs simultaneously in the MAAP, and parallelizing the code itself, has led to significant speed improvements in obtaining relevant data, reducing subsetting time from hours to minutes. MAAP users can now more quickly and easily obtain only the data relevant to their research, by choosing which GEDI collection they want to work with (L1A, L2A, L2B, or L4A), and how they want to subset it, by specifying an area of interest, a temporal range, and relevant attributes. This has significantly reduced the feedback loop for users, allowing them to much more quickly subset GEDI data and begin their analysis. Although the GEDI Subsetter originally targeted users of the MAAP, it is generalized such that it can also be used outside of the MAAP and includes a command-line interface for convenience. Furthermore, with minor modifications, it should be possible to use it with non-GEDI data as the general pattern should be applicable to other sparse/track-based sensors.

Charles Daniels↗

DOE OSTI2019

Partitioned Coupling for Structural Acoustics

Here, we expand the second-order fluid–structure coupling scheme of Farhat et al. (1998, “Load and Motion Transfer Algorithms for 19 Fluid/Structure Interaction Problems With Non-Matching Discrete Interfaces: Momentum and Energy Conservation, Optimal Discretization and Application to Aeroelasticity,” Comput. Methods Appl. Mech. Eng., 157(1–2), pp. 95–114; 2006, “Provably Second-Order Time-Accurate Loosely-Coupled Solution Algorithms for Transient Nonlinear Computational Aeroelasticity,” Comput. Methods Appl. Mech. Eng., 195(17), pp. 1973–2001) to structural acoustics. The staggered structural acoustics solution method is demonstrated to be second-order accurate in time, and numerical results are compared to a monolithically coupled system. The partitioned coupling method is implemented in the Sierra Mechanics software suite, allowing for the loose coupling of time domain acoustics in SIERRA/SD to structural dynamics ( SIERRA/SD ) or solid mechanics ( SIERRA/SD ). The coupling is demonstrated to work for nonconforming meshes. Results are verified for a one-dimensional piston, and the staggered and monolithic results are compared to an exact solution. Huang, H. (1969, “Transient Interaction of Plane Acoustic Waves With a Spherical Elastic Shell,” J. Acoust. Soc. Am., 45(3), pp. 661–670) sphere scattering problem with a spherically spreading acoustic load demonstrates parallel capability on a complex problem. Our numerical results compare well for a bronze plate submerged in water and sinusoidally excited (Fahnline and Shepherd, 2017, “Transient Finite Element/Equivalent Sources Using Direct Coupling and Treating the Acoustic Coupling Matrix as Sparse,” J. Acoust. Soc. Am., 142(2), pp. 1011–1024).

42 ENGINEERING↗

NASA NTRS2000

Incremental Parallelization of Non-Data-Parallel Programs Using the Charon Message-Passing Library

Message passing is among the most popular techniques for parallelizing scientific programs on distributed-memory architectures. The reasons for its success are wide availability (MPI), efficiency, and full tuning control provided to the programmer. A major drawback, however, is that incremental parallelization, as offered by compiler directives, is not generally possible, because all data structures have to be changed throughout the program simultaneously. Charon remedies this situation through mappings between distributed and non-distributed data. It allows breaking up the parallelization into small steps, guaranteeing correctness at every stage. Several tools are available to help convert legacy codes into high-performance message-passing programs. They usually target data-parallel applications, whose loops carrying most of the work can be distributed among all processors without much dependency analysis. Others do a full dependency analysis and then convert the code virtually automatically. Even more toolkits are available that aid construction from scratch of message passing programs. None, however, allows piecemeal translation of codes with complex data dependencies (i.e. non-data-parallel programs) into message passing codes. The Charon library (available in both C and Fortran) provides incremental parallelization capabilities by linking legacy code arrays with distributed arrays. During the conversion process, non-distributed and distributed arrays exist side by side, and simple mapping functions allow the programmer to switch between the two in any location in the program. Charon also provides wrapper functions that leave the structure of the legacy code intact, but that allow execution on truly distributed data. Finally, the library provides a rich set of communication functions that support virtually all patterns of remote data demands in realistic structured grid scientific programs, including transposition, nearest-neighbor communication, pipelining, gather/scatter, and redistribution. At the end of the conversion process most intermediate Charon function calls will have been removed, the non-distributed arrays will have been deleted, and virtually the only remaining Charon functions calls are the high-level, highly optimized communications. Distribution of the data is under complete control of the programmer, although a wide range of useful distributions is easily available through predefined functions. A crucial aspect of the library is that it does not allocate space for distributed arrays, but accepts programmer-specified memory. This has two major consequences. First, codes parallelized using Charon do not suffer from encapsulation; user data is always directly accessible. This provides high efficiency, and also retains the possibility of using message passing directly for highly irregular communications. Second, non-distributed arrays can be interpreted as (trivial) distributions in the Charon sense, which allows them to be mapped to truly distributed arrays, and vice versa. This is the mechanism that enables incremental parallelization. In this paper we provide a brief introduction of the library and then focus on the actual steps in the parallelization process, using some representative examples from, among others, the NAS Parallel Benchmarks. We show how a complicated two-dimensional pipeline-the prototypical non-data-parallel algorithm- can be constructed with ease. To demonstrate the flexibility of the library, we give examples of the stepwise, efficient parallel implementation of nonlocal boundary conditions common in aircraft simulations, as well as the construction of the sequence of grids required for multigrid.

VanderWijngaart, Rob F.↗

DOE OSTI2018

A non-intrusive parallel-in-time adjoint solver with the XBraid library

In this paper, an adjoint solver for the multigrid-in-time software library XBraid is presented. XBraid provides a non-intrusive approach for simulating unsteady dynamics on multiple processors while parallelizing not only in space but also in the time domain (XBraid: Parallel multigrid in time, http://llnl.gov/casc/xbraid). It applies an iterative multigrid reduction in time algorithm to existing spatially parallel classical time propagators and computes the unsteady solution parallel in time. Techniques from Automatic Differentiation are used to develop a consistent discrete adjoint solver which provides sensitivity information of output quantities with respect to design parameter changes. The adjoint code runs backwards through the primal XBraid actions and accumulates gradient information parallel in time. It is highly non-intrusive as existing adjoint time propagators can easily be integrated through the adjoint interface. The adjoint code is validated on advection-dominated flow with periodic upstream boundary condition. We report it provides similar strong scaling results as the primal XBraid solver and offers great potential for speeding up the overall computational costs for sensitivity analysis using multiple processors.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Fluid dynamic and thermal performance of a slotted cylinder at low Reynolds number

The fluid dynamic and thermal performance of a circular cylinder with a slot parallel to the flow is numerically investigated. The study utilized the semi-implicit finite volume multi-material algorithm MPM-ICE, a component of the Uintah framework. The normalized slot width s/D ranges from 0.1 - 0.3, introducing an additional heat transfer surface area between ~ 10 and ~ 50%, and a mass reduction between ~ 13 and ~ 38% in the cylinder. We assumed two-dimensional incompressible flow and simulated a Reynolds number Re D between 100 and 1000. The slotted cylinders are found to have a total drag force reduction up to ~ 45%, compared to a solid cylinder despite the additional viscous drag force in the slot. Convection heat transfer is enhanced up to ~ 70%. Further, the slotted cylinder performance index, defined as the ratio of the heat rate to the drag force, increases up to maximum of ~ 3, indicating better overall thermal fluid performance. An entropy analysis showed the best performance index occurs at the highest Re D . Correlations for drag coefficient and Nusselt number are proposed along with an entropy optimization method.

42 ENGINEERING↗

NASA NTRS2000

Use Computer-Aided Tools to Parallelize Large CFD Applications

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 Greenwich, to reduce potential errors made by users. Earlier tests on NAS Benchmarks and ARC3D have demonstrated good success of this tool. In this study, we have applied CAPO to parallelize three large applications in the area of computational fluid dynamics (CFD): OVERFLOW, TLNS3D and INS3D. These codes are widely used for solving Navier-Stokes equations with complicated boundary conditions and turbulence model in multiple zones. Each one comprises of from 50K to 1,00k lines of FORTRAN77. As an example, CAPO took 77 hours to complete the data dependence analysis of OVERFLOW on a workstation (SGI, 175MHz, R10K processor). A fair amount of effort was spent on correcting false dependencies due to lack of necessary knowledge during the analysis. Even so, CAPO provides an easy way for user to interact with the parallelization process. The OpenMP version was generated within a day after the analysis was completed. Due to sequential algorithms involved, code sections in TLNS3D and INS3D need to be restructured by hand to produce more efficient parallel codes. An included figure shows preliminary test results of the generated OVERFLOW with several test cases in single zone. The MPI data points for the small test case were taken from a handcoded MPI version. As we can see, CAPO's version has achieved 18 fold speed up on 32 nodes of the SGI O2K. For the small test case, it outperformed the MPI version. These results are very encouraging, but further work is needed. For example, although CAPO attempts to place directives on the outer- most parallel loops in an interprocedural framework, it does not insert directives based on the best manual strategy. In particular, it lacks the support of parallelization at the multi-zone level. Future work will emphasize on the development of methodology to work in a multi-zone level and with a hybrid approach. Development of tools to perform more complicated code transformation is also needed.

Jin, H.↗