Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallel optimization”

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 1,081 records · Page 60

Improving I/O Performance for Exascale Applications through Online Data Layout Reorganization

The applications being developed within the U.S. Exascale Computing Project (ECP) to run on imminent Exascale computers will generate scientific results with unprecedented fidelity and record turn-around time. Many of these codes are based on particle-mesh methods and use advanced algorithms, especially dynamic load-balancing and mesh-refinement, to achieve high performance on Exascale machines. Yet, as such algorithms improve parallel application efficiency, they raise new challenges for I/O logic due to their irregular and dynamic data distributions. Thus, while the enormous data rates of Exascale simulations already challenge existing file system write strategies, the need for efficient read and processing of generated data introduces additional constraints on the data layout strategies that can be used when writing data to secondary storage. We review these I/O challenges and introduce two online data layout reorganization approaches for achieving good tradeoffs between read and write performance. We demonstrate the benefits of using these two approaches for the ECP particle-in-cell simulation WarpX, which serves as a motif for a large class of important Exascale applications. Here, we show that by understanding application I/O patterns and carefully designing data layouts we can increase read performance by more than 80 percent.

97 MATHEMATICS AND COMPUTING↗

PolyCheck: Dynamic Verification of Iteration Space Transformations on Affine Programs

High-level compiler transformations, especially loop transformations, are widely recognized as critical optimizations to restructure programs to improve data locality and expose parallelism. Guaranteeing the correctness of program transformations is essential, and to date three main approaches have been developed: proof of equivalence of affine programs, matching the execution traces of programs, and checking bit-by-bit equivalence of the outputs of the programs. Each technique suffers from limitations in either the kind of transformations supported, space complexity, or the sensitivity to the testing dataset. In this paper, we take a novel approach addressing all three limitations to provide an automatic bug checker to verify any iteration reordering transformations on affine programs, including non-affine transformations, with space consumption proportional to the original program data, and robust to arbitrary datasets of a given size. We achieve this by exploiting the structure of affine program control- and data-flow to generate at compile-time lightweight checker code to be executed within the transformed program. Experimental results assess the correctness and effectiveness of our method, and its increased coverage over previous approaches.

affine programs↗

Distributed Halide

Many image processing tasks are naturally expressed as a pipeline of small computational kernels known as stencils. Halide is a popular domain-specific language and compiler designed to implement image processing algorithms. Halide uses simple language constructs to express what to compute and a separate scheduling co-language for expressing when and where to perform the computation. This approach has demonstrated performance comparable to or better than hand-optimized code. Until now, however, Halide has been restricted to parallel shared memory execution, limiting its performance for memory-bandwidth-bound pipelines or large-scale image processing tasks. We present an extension to Halide to support distributed-memory parallel execution of complex stencil pipelines. These extensions compose with the existing scheduling constructs in Halide, allowing expression of complex computation and communication strategies. Existing Halide applications can be distributed with minimal changes, allowing programmers to explore the tradeoff between recomputation and communication with little effort. Approximately 10 new of lines code are needed even for a 200 line, 99 stage application. On nine image processing benchmarks, our extensions give up to a 1.4× speedup on a single node over regular multithreaded execution with the same number of cores, by mitigating the effects of non-uniform memory access. The distributed benchmarks achieve up to 18× speedup on a 16 node testing machine and up to 57× speedup on 64 nodes of the NERSC Cori supercomputer.

97 MATHEMATICS AND COMPUTING↗

Automatic Differentiation of C++ Codes on Emerging Manycore Architectures with Sacado

Automatic differentiation (AD) is a well-known technique for evaluating analytic derivatives of calculations implemented on a computer, with numerous software tools available for incorporating AD technology into complex applications. However, a growing challenge for AD is the efficient differentiation of parallel computations implemented on emerging manycore computing architectures such as multicore CPUs, GPUs, and accelerators as these devices become more pervasive. In this work, we explore forward mode, operator overloading-based differentiation of C++ codes on these architectures using the widely available Sacado AD software package. In particular, we leverage Kokkos, a C++ tool providing APIs for implementing parallel computations that is portable to a wide variety of emerging architectures. Here we describe the challenges that arise when differentiating code for these architectures using Kokkos, and two approaches for overcoming them that ensure optimal memory access patterns as well as expose additional dimensions of fine-grained parallelism in the derivative calculation. We describe the results of several computational experiments that demonstrate the performance of the approach on a few contemporary CPU and GPU architectures. We then conclude with applications of these techniques to the simulation of discretized systems of partial differential equations.

97 MATHEMATICS AND COMPUTING↗

HYPPO v1.0.0

HYPPO is a hyperparameter optimization (HPO) software that uses surrogate modeling and uncertainty quantification to provide reliable neural network architectures. Similar state-of-the-art HPO technologies do not use such features and mostly perform random searches to explore the parameter space, which can easily become computationally expensive. Using HYPPO, the user is able to reduce by an order of magnitude the number of evaluations necessary to identify the most optimal region in the hyperparameter space. Finally, using asynchronous nested parallelism, we are able to decrease by two orders of magnitude the amount of time required to complete the HPO process.

Dumont, Vincent↗

Miniaturized Air-to-Refrigerant Heat Exchangers

Air-to-refrigerant Heat eXchangers (HX) are an essential component of Heating, Ventilation, Air-Conditioning, and Refrigeration (HVAC&R) systems, serving as the main heat transfer component. The major limiting factor to HX performance is the large airside thermal resistance. Recent literature aims at improving heat transfer performance by utilizing enhancement methods such as fins and small tube diameters; this has lead to almost exhaustive research on the microchannel HX (MCHX). The objective of this project is to develop a miniaturized air-to-refrigerant HX with at least 20% reduction in volume, material volume, and approach temperature compared to current state-of-the-art multiport flat tube designs and also be capable of production within five years. Moreover, the proposed HX’s are expected to have good water drainage and should succeed in both evaporator and condenser applications. The project leveraged Parallel-Parametrized Computational Fluid Dynamics (PPCFD) and Approximation-Assisted Optimization (AAO) techniques to perform multi-scale analysis and shape optimization with the intent of developing novel HX designs whose thermal-hydraulic performance exceeds that of state-of-the-art MCHX. Nine heat exchanger geometries were initially chosen for detailed analysis, selected from 35+ geometries which were identified in previous work at the University of Maryland, College Park. The newly developed optimization framework was exercised for three design optimization problems: (DP I) 1.0kW radiator, (DP II) 10kW radiator and (DP III) 10kW two-phase HX. DP I consisted of the design and optimization of 1.0kW air-to-water HX’s which exceeded the project requirements of 20% volume/material reduction and 20% better performance. Two prototypes for the 1.0kW HX were prototyped, tested and validated using newly-designed airside and refrigerant side test facilities. DP II, a scaled version DP I for 10kW air-to-water HX applications, also yielded optimized HX designs which met project requirements. Attempts to prototype a 10kW have presented unique manufacturing challenges, especially regarding tube blockages and structural stability. DP III comprised optimizing two-phase HX’s for a 3.0Ton capacity in a heat pump / air-conditioning unit for cooling mode application using R410A as the working fluid. The HX’s theoretically address the project requirements. System-level analysis showed the HX’s achieved up to 15% improvement in COP while also reducing overall unit charge by 30-40%. The project methodology was capable of developing HX’s which can outperform current state-of-the-art MCHX by at least 20% reduction in volume, material volume, and approach temperature. Additionally, the capability for optimization using refrigerant charge as an objective function was developed. The five-year manufacturing feasibility of the proposed HX’s was shown to have a good outlook. Successful prototyping through both conventional manufacturing methods and next generation methods such as additive manufacturing was achieved.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Transient and Steady-State Inverse Problems in Sierra/Aria

Inverse problems arise in a wide range of applications, whenever unknown model parameters cannot be measured directly. Instead, the unknown parameters are estimated using experimental data and forward simulations. Thermal inverse problems, such as material characterization problems, are often large-scale and transient. Therefore, they require intrusive adjoint-based gradient implementations in order to be solved efficiently. The capability to solve large-scale transient thermal inverse problems using an adjoint-based approach was recently implemented in SNL Sierra Mechanics, a massively parallel capable multiphysics code suite. This report outlines the theory, optimization formulation, and path taken to implement thermal inverse capabilities in Sierra within a unit test framework. The capability utilizes Sierra/Aria and Sierra/Fuego data structures, the Rapid Optimization Library, and an interface to the Sierra/InverseOpt library. The existing Sierra/Aria time integrator is leveraged to implement a time-dependent adjoint solver.

97 MATHEMATICS AND COMPUTING↗

Configuration evaluation and criteria plan. Volume 2: Evaluation critera plan (preliminary). Space Transportation Main Engine (STME) configuration study

The unbiased selection of the Space Transportation Main Engine (STME) configuration requires that the candidate engines be evaluated against a predetermined set of criteria which must be properly weighted to emphasize critical requirements defined prior to the actual evaluation. The evaluation and selection process involves the following functions: (1) determining if a configuration can satisfy basic STME requirements (yes/no); (2) defining the evaluation criteria; (3) selecting the criteria relative importance or weighting; (4) determining the weighting sensitivities; and (5) establishing a baseline for engine evaluation. The criteria weighting and sensitivities are cost related and are based on mission models and vehicle requirements. The evaluation process is used as a coarse screen to determine the candidate engines for the parametric studies and as a fine screen to determine concept(s) for conceptual design. The criteria used for the coarse and fine screen evaluation process is shown. The coarse screen process involves verifying that the candidate engines can meet the yes/no screening requirements and a semi-subjective quantitative evaluation. The fine screen engines have to meet all of the yes/no screening gates and are then subjected to a detailed evaluation or assessment using the quantitative cost evaluation processes. The option exists for re-cycling a concept through the quantitative portion of the screening and allows for some degree of optimization. The basic vehicle is a two stage LOX/HC, LOX/LH2 parallel burn vehicle capable of placing 150,000 lbs in low Earth orbit (LEO).

Bair, E. K.↗

Scheduling with genetic algorithms

In many domains, scheduling a sequence of jobs is an important function contributing to the overall efficiency of the operation. At Boeing, we develop schedules for many different domains, including assembly of military and commercial aircraft, weapons systems, and space vehicles. Boeing is under contract to develop scheduling systems for the Space Station Payload Planning System (PPS) and Payload Operations and Integration Center (POIC). These applications require that we respect certain sequencing restrictions among the jobs to be scheduled while at the same time assigning resources to the jobs. We call this general problem scheduling and resource allocation. Genetic algorithms (GA's) offer a search method that uses a population of solutions and benefits from intrinsic parallelism to search the problem space rapidly, producing near-optimal solutions. Good intermediate solutions are probabalistically recombined to produce better offspring (based upon some application specific measure of solution fitness, e.g., minimum flowtime, or schedule completeness). Also, at any point in the search, any intermediate solution can be accepted as a final solution; allowing the search to proceed longer usually produces a better solution while terminating the search at virtually any time may yield an acceptable solution. Many processes are constrained by restrictions of sequence among the individual jobs. For a specific job, other jobs must be completed beforehand. While there are obviously many other constraints on processes, it is these on which we focussed for this research: how to allocate crews to jobs while satisfying job precedence requirements and personnel, and tooling and fixture (or, more generally, resource) requirements.

Fennel, Theron R.↗

Hybrid Optimization Software Suite

The hybrid optimization software suite (HOSS) is a general purpose fully 2D/3D parallel combined finite discrete element (FDEM) code that can be used to simulate problems involving fracture and fragmentation processes, large deformation and large rotations, discrete particle systems, solid-fluid interaction in hydro fracture problems, etc. HOSS uses a hybrid approach that combines finite-element and discrete-element methods with a novel computational fluid dynamics solver. The finite-element method is often used to analyze a material or object and how it responds to stress. The discrete-element method analyzes stresses and displacements in a volume containing a large number of particles, such as grains of sand. Fluid dynamics analyzes the fluid flow inside, around, or through solid domains. With these processes combined, HOSS represents a paradigm shift when it comes to generating accurate simulations of material deformations and failure. HOSS can perform the following: Resolves problems that consist of millions of deforming fracturing, and interacting solid particles; Requires no coupling—it naturally integrates with all regimes of fluid flow; Resolves all regimes of fluid flow, such as Stokes, low/high Reynolds, subsonic/transonic/supersonic/hypersonic, compressible/incompressible, viscid/inviscid, Newtonian/Non-Newtonian and turbulent/laminar flow; Combines all flow regimes in the same solver; Numerically stable for all flow regimes; Naturally combines different flow regimes in the same problem; Possesses material library for fluid behavior; Naturally matches fluid solver to solid solver; Includes non-inertial Eulerian formulation for fluids; Uses highly efficient parallel computing based on the Virtual Parallel Machine, thus enabling easy porting between different parallel architectures, including possible future architectures; Designed specifically for multi-physics problems in research and industry in virtual experimentation format, thereby complementing both theoretical-computational and experimental tools; Can be extended to address millions of deforming solid particles immersed in fluid flowing at all regimes; Can be extended to address biological applications, such as modeling bone fracture or red blood cells in blood plasma

Knight, Earl E.↗

CASL COBRA-TF improvements for PWR DNB analysis

COBRA-TF (CTF) is a thermal hydraulic (T/H) subchannel code using either three-dimensional (3D) Cartesian or subchannel coordinate formulations for two-phase fluid flow and heat transfer solutions. CTF has been improved under the Consortium for Advanced Simulation of Light Water Reactors (CASL) program for Pressurized Water Reactor (PWR) applications, including software optimization, new closure models, pre- and post-processing and parallelization for modeling full reactor core T/H responses under normal operating and accident conditions. As a result of collaboration among CASL partners including the Westinghouse Electric Company, the Oak Ridge National Laboratory (ORNL), and the Sandia National Laboratories, additional modeling improvements were made to CTF specifically for PWR Departure from Nucleate Boiling (DNB) analysis, including a code option to evaluate fuel thermal margin in terms of DNB Ratio (DNBR) and an axial shape factor to account for effect of non-uniform axial power distribution on DNB. Multiple DNB correlations are now linked with CTF for different applications, including the Westinghouse proprietary WRB-1 correlation for fuel designs containing mixing vane grid spacers. The improved CTF code with the WRB-1 correlation (CTF/WRB-1) was validated using the DNB data from the PWR Subchannel Bundle Tests (PSBT). In addition to the comparison with the test data, the CTF/WRB-1 DNBR results and the associated local fluid conditions were compared to the results of the Westinghouse T/H design code, VIPRE-W, which is an enhanced version of the VIPRE-01 code originally developed by the Electric Power Research Institute (EPRI). The comparisons showed that CTF/WRB-1 DNBR predictions are in good agreement with the VIPRE-W results within the applicable range of the DNB correlation. A model sensitivity study was performed to confirm that the CTF void drift model had an insignificant effect on DNBR under the steam line break (SLB) low pressure condition. (authors)

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Terascale direct numerical simulations of turbulent combustion using S3D.

Computational science is paramount to the understanding of underlying processes in internal combustion engines of the future that will utilize non-petroleum-based alternative fuels, including carbon-neutral biofuels, and burn in new combustion regimes that will attain high efficiency while minimizing emissions of particulates and nitrogen oxides. Next-generation engines will likely operate at higher pressures, with greater amounts of dilution and utilize alternative fuels that exhibit a wide range of chemical and physical properties. Therefore, there is a significant role for high-fidelity simulations, direct numerical simulations (DNS), specifically designed to capture key turbulence-chemistry interactions in these relatively uncharted combustion regimes, and in particular, that can discriminate the effects of differences in fuel properties. In DNS, all of the relevant turbulence and flame scales are resolved numerically using high-order accurate numerical algorithms. As a consequence terascale DNS are computationally intensive, require massive amounts of computing power and generate tens of terabytes of data. Recent results from terascale DNS of turbulent flames are presented here, illustrating its role in elucidating flame stabilization mechanisms in a lifted turbulent hydrogen/air jet flame in a hot air co-flow, and the flame structure of a fuel-lean turbulent premixed jet flame. Computing at this scale requires close collaborations between computer and combustion scientists to provide optimized scaleable algorithms and software for terascale simulations, efficient collective parallel I/O, tools for volume visualization of multiscale, multivariate data and automating the combustion workflow. The enabling computer science, applied to combustion science, is also required in many other terascale physics and engineering simulations. In particular, performance monitoring is used to identify the performance of key kernels in the DNS code, S3D and especially memory intensive loops in the code. Through the careful application of loop transformations, data reuse in cache is exploited thereby reducing memory bandwidth needs, and hence, improving S3D's nodal performance. To enhance collective parallel I/O in S3D, an MPI-I/O caching design is used to construct a two-stage write-behind method for improving the performance of write-only operations. The simulations generate tens of terabytes of data requiring analysis. Interactive exploration of the simulation data is enabled by multivariate time-varying volume visualization. The visualization highlights spatial and temporal correlations between multiple reactive scalar fields using an intuitive user interface based on parallel coordinates and time histogram. Finally, an automated combustion workflow is designed using Kepler to manage large-scale data movement, data morphing, and archival and to provide a graphical display of run-time diagnostics.

33 ADVANCED PROPULSION SYSTEMS↗

Angle-Resolved Photoemission Spectroscopy on Electronic Structure and Electron-Phonon Coupling in Cuprate Superconductors

In addition to the record high superconducting transition temperature (T{sub c}), high temperature cuprate superconductors are characterized by their unusual superconducting properties below T{sub c}, and anomalous normal state properties above T{sub c}. In the superconducting state, although it has long been realized that superconductivity still involves Cooper pairs, as in the traditional BCS theory, the experimentally determined d-wave pairing is different from the usual s-wave pairing found in conventional superconductors. The identification of the pairing mechanism in cuprate superconductors remains an outstanding issue. The normal state properties, particularly in the underdoped region, have been found to be at odd with conventional metals which is usually described by Fermi liquid theory; instead, the normal state at optimal doping fits better with the marginal Fermi liquid phenomenology. Most notable is the observation of the pseudogap state in the underdoped region above T{sub c}. As in other strongly correlated electrons systems, these unusual properties stem from the interplay between electronic, magnetic, lattice and orbital degrees of freedom. Understanding the microscopic process involved in these materials and the interaction of electrons with other entities is essential to understand the mechanism of high temperature superconductivity. Since the discovery of high-T{sub c} superconductivity in cuprates, angle-resolved photoemission spectroscopy (ARPES) has provided key experimental insights in revealing the electronic structure of high temperature superconductors. These include, among others, the earliest identification of dispersion and a large Fermi surface, an anisotropic superconducting gap suggestive of a d-wave order parameter, and an observation of the pseudogap in underdoped samples. In the mean time, this technique itself has experienced a dramatic improvement in its energy and momentum resolutions, leading to a series of new discoveries not thought possible only a decade ago. This revolution of the ARPES technique and its scientific impact result from dramatic advances in four essential components: instrumental resolution and efficiency, sample manipulation, high quality samples and well-matched scientific issues. The purpose of this treatise is to go through the prominent results obtained from ARPES on cuprate superconductors. Because there have been a number of recent reviews on the electronic structures of high-T{sub c} materials, we will mainly present the latest results not covered previously, with a special attention given on the electron-phonon interaction in cuprate superconductors. What has emerged is rich information about the anomalous electron-phonon interaction well beyond the traditional views of the subject. It exhibits strong doping, momentum and phonon symmetry dependence, and shows complex interplay with the strong electron-electron interaction in these materials. ARPES experiments have been instrumental in identifying the electronic structure, observing and detailing the electron-phonon mode coupling behavior, and mapping the doping evolution of the high-T{sub c} cuprates. The spectra evolve from the strongly coupled, polaronic spectra seen in underdoped cuprates to the Migdal-Eliashberg like spectra seen in the optimally and overdoped cuprates. In addition to the marked doping dependence, the cuprates exhibit pronounced anisotropy with direction in the Brillouin zone: sharp quasiparticles along the nodal direction that broaden significantly in the anti-nodal region of the underdoped cuprates, an anisotropic electron-phonon coupling vertex for particular modes identified in the optimal and overdoped compounds, and preferential scattering across the two parallel pieces of Fermi surface in the antinodal region for all doping levels. This also contributes to the pseudogap effect. To the extent that the Migdal-Eliashberg picture applies, the spectra of the cuprates bear resemblance to that seen in established strongly coupled electron-phonon superconductors such as Pb. On the other hand, the cuprates deviate from this conventional picture. In the underdoped regime, the carriers are best understood as small polarons in an antiferromagnetic, highly electron correlated background, while the doped compounds require an anisotropic electron-phonon vertex to detail the prominent mode coupling signatures in the superconducting state. Electronic vertex corrections to the electron-phonon coupling furthermore may enhance, and for certain phonons, determine, the anisotropy of the electron-phonon coupling. A consistent picture emerges of the cuprates, combining strong, anisotropic electron-phonon coupling, particular phonon modes that could give rise to such a coupling, and an electron-electron interaction modifying the el-ph vertex.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Optimizing multigrid reduction-in-time and Parareal coarse-grid operators for linear advection

Parallel-in-time methods, such as multigrid reduction-in-time (MGRIT) and Parareal, provide an attractive option for increasing concurrency when simulating time-dependent partial differential equations (PDEs) in modern high-performance computing environments. While these techniques have been very successful for parabolic equations, it has often been observed that their performance suffers dramatically when applied to advection-dominated problems or purely hyperbolic PDEs using standard rediscretization approaches on coarse grids. In this paper, we apply MGRIT or Parareal to the constant-coefficient linear advection equation, appealing to existing convergence theory to provide insight into the typically nonscalable or even divergent behavior of these solvers for this problem. To overcome these failings, we replace rediscretization on coarse grids with improved coarse-grid operators that are computed by applying optimization techniques to approximately minimize error estimates from the convergence theory. Therefore, one of our main findings is that, in order to obtain fast convergence as for parabolic problems, coarse-grid operators should take into account the behavior of the hyperbolic problem by tracking the characteristic curves. Our approach is tested for schemes of various orders using explicit or implicit Runge–Kutta methods combined with upwind-finite-difference spatial discretizations. In all cases, we obtain scalable convergence in just a handful of iterations, with parallel tests also showing significant speed-ups over sequential time-stepping.

97 MATHEMATICS AND COMPUTING↗

Parallel Multigrid in Time and Space for Extreme-Scale Computational Science: Chaotic and Hyperbolic Problems

The coming massive parallelism of exascale computing presents a pressing challenge for the many DOE simulations of time-dependent partial differential equations (PDEs), which typically use traditional sequential time stepping methods. Since this traditional approach is inherently serial, it presents a sequential bottleneck when moving to exascale computing, because future performance gains will come through greater concurrency, not faster clock speeds. Thus, the goal of this work is to research parallelism in time, i.e., methods that compute multiple time values simultaneously, not sequentially. The focus will be on hyperbolic and chaotic problems of interest to DOE, with the goal of enabling scalable simulations of time-dependent hyperbolic and chaotic problems on future architectures. The chosen methodology for solving these problems parallel-in-time is multigrid, because multigrid (when it works) is a powerful, optimal, and scalable solver for discretized PDEs. Multigrid is already commonly used in many DOE simulations for scalably and optimally solving space-only PDE problems. The areas of hyperbolic and chaotic problems are chosen because of their relevance to problems of programmatic interest to DOE. However, these problems are also well-known to be difficult for parallelin-time methods, with the most common method, parareal, diverging in many cases. The current stateof-the-art for parallel-in-time at LLNL is the multigrid reduction in time (MGRIT) XBraid package, which also struggles for such problems, while still showing some improvement over parareal. In summary, new methods are needed for an efficient parallel-in-time scheme for hyperbolic and chaotic problems, and this work shall research promising new multigrid methods in this area. In particular, we take inspiration from the Least Squares Shadowing (LSS by Wang) approach for solving chaotic problems. Here, an optimization approach is able to find “well-conditioned” shadow trajectories/solutions to the original “ill-conditioned” chaotic problem. Thus, the new multigrid methods researched here also arise in an optimization context.

97 MATHEMATICS AND COMPUTING↗

An Optimizing Compiler for Petascale I/O on Leadership-Class Architectures

In high-performance computing (HPC), parallel I/O architectures usually have very complex hierarchies with multiple layers that collectively constitute an I/O stack, including high-level I/O libraries such as PnetCDF and HDF5, I/O middleware such as MPI-IO, and parallel file systems such as PVFS and Lustre. Our DOE project explored automated instrumentation and compiler support for I/O intensive applications. Our project made significant progress towards understanding the complex I/O hierarchies of high-performance storage systems (including storage caches, HDDs, and SSDs), and designing and implementing state-of-the-art compiler/runtime system technology that targets I/O intensive HPC applications that target leadership class machine. This final report summarizes the major achievements of the project and also points out promising future directions Two new sections in this report compared to the previous report are IOGenie and SSD/NVM-specific optimizations.

97 MATHEMATICS AND COMPUTING↗

Parallel Finite Element Domain Decomposition for Structural/Acoustic Analysis

A domain decomposition (DD) formulation for solving sparse linear systems of equations resulting from finite element analysis is presented. The formulation incorporates mixed direct and iterative equation solving strategics and other novel algorithmic ideas that are optimized to take advantage of sparsity and exploit modern computer architecture, such as memory and parallel computing. The most time consuming part of the formulation is identified and the critical roles of direct sparse and iterative solvers within the framework of the formulation are discussed. Experiments on several computer platforms using several complex test matrices are conducted using software based on the formulation. Small-scale structural examples are used to validate thc steps in the formulation and large-scale (l,000,000+ unknowns) duct acoustic examples are used to evaluate the ORIGIN 2000 processors, and a duster of 6 PCs (running under the Windows environment). Statistics show that the formulation is efficient in both sequential and parallel computing environmental and that the formulation is significantly faster and consumes less memory than that based on one of the best available commercialized parallel sparse solvers.

Nguyen, Duc T.↗