Search NASA⌕ Search

SEARCH · Search NASA

Results for “accelerated computing”

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 1,009 records · Page 56

Experience and Practice of Batch Scheduling on Leadership Supercomputers at Argonne

The mission of the DOE Argonne Leadership Computing Facility (ALCF) is to accelerate major scientific discoveries and engineering breakthroughs for humanity by designing and providing world-leading computing facilities in partnership with the computational science community. The ALCF operates supercomputers that are generally amongst the Top 5 fastest machines in the world. Specifically, ALCF is looking for the science that is either too big to run anywhere else, or it would take so long as to be impractical (i.e., “capability jobs”). At ALCF, batch scheduling plays a critical role for achieving a set of site goals within a set of constraints. While system utilization is an important goal at ALCF, its largest mission constraint is to enable extreme scale parallel jobs to take precedence. In this paper, we will describe the specific scheduling goals and constraints, analyze the workload traces collected in 2013–2017 from the 48-rack petascale supercomputer Mira, and discuss the upcoming scheduling challenges at ALCF.

Allcock, William↗

Accelerated development of Zr-containing new generation ferritic steels for advanced nuclear reactors

The mission of the Nuclear Energy Enabling Technologies (NEET) program is to develop crosscutting technologies for nuclear energy applications. Advanced structural materials with superior performance at elevated temperatures are always desired for nuclear reactors, which can improve reactor economics, safety margins, and design flexibility. They benefit not only new reactors, including advanced light water reactors (LWRs) and fast reactors such as the sodium-cooled fast reactor (SFR) that is primarily designed for management of high-level wastes, but also life extension of the existing fleet when component exchange is needed. Developing and utilizing the modern materials science tools (experimental, theoretical, and computational tools) is an important path to more efficient alloy development and process optimization. The ultimate goal of this project is, with the aid of computational modeling tools, to accelerate the development of Zr-bearing ferritic alloys that can be fabricated using conventional steelmaking methods. The new alloys are expected to have superior high-temperature creep performance and excellent radiation resistance as compared to Grade 91. The designed alloys were fabricated using arc-melting and drop-casting, followed by hot rolling and conventional heat treatments. Comprehensive experimental studies have been conducted on the developed alloys to evaluate their hardness, tensile properties, creep resistance, Charpy impact toughness, and aging resistance, as well as resistance to proton and heavy ion (Fe 2+ ) irradiation.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Application of Plasma Waveguides to High Energy Accelerators

This grant supported basic experimental, theoretical and computer simulation research into developing a compact, high pulse repetition rate laser accelerator using the direct laser acceleration mechanism in plasma-based slow wave structures.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Computer-Vision-Based Vibration Tracking Using a Digital Camera: A Sparse-Optical-Flow-Based Target Tracking Method

Computer-vision-based target tracking is a technology applied to a wide range of research areas, including structural vibration monitoring. However, current target tracking methods suffer from noise in digital image processing. In this paper, a new target tracking method based on the sparse optical flow technique is introduced for improving the accuracy in tracking the target, especially when the target has a large displacement. The proposed method utilizes the Oriented FAST and Rotated BRIEF (ORB) technique which is based on FAST (Features from Accelerated Segment Test), a feature detector, and BRIEF (Binary Robust Independent Elementary Features), a binary descriptor. ORB maintains a variety of keypoints and combines the multi-level strategy with an optical flow algorithm to search the keypoints with a large motion vector for tracking. Then, an outlier removal method based on Hamming distance and interquartile range (IQR) score is introduced to minimize the error. The proposed target tracking method is verified through a lab experiment—a three-story shear building structure subjected to various harmonic excitations. It is compared with existing sparse-optical-flow-based target tracking methods and target tracking methods based on three other types of techniques, i.e., feature matching, dense optical flow, and template matching. The results show that the performance of target tracking is greatly improved through the use of a multi-level strategy and the proposed outlier removal method. The proposed sparse-optical-flow-based target tracking method achieves the best accuracy compared to other existing target tracking methods.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

[Office of Basic Energy Sciences (BES)] Roundtable on Producing and Managing Large Scientific Data with Artificial Intelligence and Machine Learning

To identify specific Priority Research Opportunities (PROs) for artificial intelligence and machine learning (AI/ML) at the US Department of Energy’s scientific user facilities, the Office of Basic Energy Sciences (BES) convened a roundtable of facility experts encompassing the fields of physics, chemistry, materials synthesis science, computational science, detector and accelerator technology, theory, modeling, simulation, and atomic-scale characterization techniques. The roundtable met on October 22–23, 2019, to identify coordinated, long-term AI/ML research efforts that will drive major advances in neutron, photon, and nanoscale sciences. This report describes the four PROs identified at the roundtable: PRO 1 on how AI/ML can extract high-value information from the large datasets; PRO 2 on how AI/ML can use such information in real time to maximize the facilities’ scientific output; PRO 3 on using AI/ML virtual laboratories (i.e., computational models of experimental facilities) to aid the facilities and user community in design and control of machine parameters and the design and execution of experiments, including training AI/ML models for PROs 1 and 2; and PRO 4 on how a shared scientific data infrastructure can provide tools to assemble and analyze the totality of data coming from user facilities. A section of the end of the report provides a summary on computer science and mathematics that highlights where enhanced AI/ML capabilities could be particularly impactful for BES user facilities.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

An assessment of coupling algorithms for nuclear reactor core physics simulations

This paper evaluates the performance of multiphysics coupling algorithms applied to a light water nuclear reactor core simulation. The simulation couples the k-eigenvalue form of the neutron transport equation with heat conduction and subchannel flow equations. We compare Picard iteration (block Gauss–Seidel) to Anderson acceleration and multiple variants of preconditioned Jacobian-free Newton–Krylov (JFNK). The performance of the methods are evaluated over a range of energy group structures and core power levels. A novel physics-based approximation to a Jacobian-vector product has been developed to mitigate the impact of expensive on-line cross section processing steps. Numerical simulations demonstrating the efficiency of JFNK and Anderson acceleration relative to standard Picard iteration are performed on a 3D model of a nuclear fuel assembly. Both criticality (k-eigenvalue) and critical boron search problems are considered.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Toward Large-Scale Image Segmentation on Summit

Semantic segmentation of images is an important computer vision task that emerges in a variety of application domains such as medical imaging, robotic vision and autonomous vehicles to name a few. While these domain-specific image analysis tasks involve relatively small image sizes (~ 10 2 × 10 2 ), there are many applications that need to train machine learning models on image data with extents that are orders of magnitude larger (~10 4 × 10 4 ). Training deep neural network (DNN) models on large extent images is extremely memory-intensive and often exceeds the memory limitations of a single graphical processing unit, a hardware accelerator of choice for computer vision workloads. Here, an efficient, sample parallel approach to train U-Net models on large extent image data sets is presented. Its advantages and limitations are analyzed and near-linear strong-scaling speedup demonstrated on 256 nodes (1536 GPUs) of the Summit supercomputer. Using a single node of the Summit supercomputer, an early evaluation of a recently released model parallel framework called GPipe is demonstrated to deliver ~ 2X speedup in executing a U-Net model with an order of magnitude larger number of trainable parameters than reported before. Performance bottlenecks for pipelined training of U-Net models are identified and mitigation strategies to improve the speedups are discussed. Together, these results open up the possibility of combining both approaches into a unified scalable pipelined and data parallel algorithm to efficiently train U-Net models with very large receptive fields on data sets of ultra-large extent images.

Seal, Sudip↗

Accelerating the Gauss-Seidel Power Flow Solver on a High Performance Reconfigurable Computer

The computationally intensive power flow problem determines the voltage magnitude and phase angle at each bus in a power system for hundreds of thousands of buses under balanced three-phase steady-state conditions. We report an FPGA acceleration of the Gauss-Seidel based power flow solver employed in the transmission module of the GridLAB-D power distribution simulator and analysis tool. The prototype hardware is implemented on an SGI Altix-RASC system equipped with a Xilinx Virtex II 6000 FPGA. Due to capacity limitations of the FPGA, only the bus voltage calculations of the power network are implemented on hardware while the branch current calculations are implemented in software. For a 200,000 bus system, the bus voltage calculation on the FPGA achieves a 48x speed-up with PQ buses and a 62 times for PV over an equivalent sequential software implementation. The average overall speed up of the FPGA-CPU implementation with 100 iterations of the Gauss-Seidel power solver is 2.6x over a software implementation, with the branch calculations on the CPU accounting for 85% of the total execution time. The FPGA-CPU implementation also shows linear scaling with increase in the size of the input power network.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Rich-Text based PSA Model Transformation and Verification in RiskA

Nuclear Energy is one of the most possible alternative to fossil energy in large-scale. Nuclear safety is the lifeline of nuclear energy development, and has been an important part of national security, especially after the accident of Fukushima. On the basis of computer codes, reliability engineering and Probabilistic Risk Assessment (PSA) is effective methods to improve the safety of nuclear power plants. After many years of development, the methodologies, applications and relevant soft programs have been quite sufficient in the area of nuclear safety. However, it is still difficult for the current PSA technologies and software to meet the needs of dealing with the problems of quantitatively public risk assessment in the digital society. Especially, there are so many software that adopted different modeling and analyzing methods, which brings difficulties to the comparison, evaluation or verification for different software and models. By using its own academic advantages, the advanced nuclear energy research team, i.e., FDS Team has invested a few hundreds person-years of independent research and development of a reliability and probabilistic safety assessment software system RiskA. The system consists of three main functions: device reliability optimization management, probabilistic safety assessment and instantaneous risk monitoring, and the comprehensive public risk assessment in digital society. RiskA establishes an integrated risk evaluation system for the device reliability, nuclear power plant (NPP) risk monitoring and public risk assessment. Currently, RiskA has successfully been applied in more than 600 organizations and institutes from a variety of fields in 39 countries, involving nuclear, aerospace, defense and so on. For example, RiskA has been used in the safety analysis of International Thermonuclear Experimental Reactor (ITER), FDS series fusion reactor, and the Strategic Priority Research Program of Chinese Academy of Sciences 'the Future Advanced Nuclear Fission Energy - Accelerator Driven Sub-critical (ADS) transmutation system'. Also, RiskA has been used in public risk communication and risk assessment. Among them, Third Qinshan nuclear power plant Risk Monitor (TQRM), which is the China's first domestic nuclear safety analysis software used in the actual online operation, has been highly apprised by renowned experts in the field. Meanwhile, it has been identified as the first domestic nuclear power plant risk monitoring system with completely independent intellectual property rights, and has reached the international advanced level. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Characteristics of an envelope model for laser-plasma accelerator simulation

Simulation of laser-plasma accelerator (LPA) experiments is computationally intensive due to the disparate length scales involved. Current experiments extend hundreds of laser wavelengths transversely and many thousands in the propagation direction, making explicit PIC simulations enormously expensive and requiring massively parallel execution in 3D. Simulating the next generation of LPA experiments is expected to increase the computational requirements yet further, by a factor of 1000. We can substantially improve the performance of LPA simulations by modeling the envelope evolution of the laser field rather than the field itself. This allows for much coarser grids, since we need only resolve the plasma wavelength and not the laser wavelength, and therefore larger timesteps can be used. Thus an envelope model can result in savings of several orders of magnitude in computational resources. By propagating the laser envelope in a Galilean frame moving at the speed of light, dispersive errors can be avoided and simulations over long distances become possible. The primary limitation to this envelope model is when the laser pulse develops large frequency shifts, and thus the slowly-varying envelope assumption is no longer valid. Here we describe the model and its implementation, and show rigorous benchmarks for the algorithm, establishing second-order convergence and correct laser group velocity. We also demonstrate simulations of LPA phenomena such as self-focusing and meter-scale acceleration stages using the model.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

A hardware overview of the RHIC LLRF platform

The RHIC Low Level RF (LLRF) platform is a flexible, modular system designed around a carrier board with six XMC daughter sites. The carrier board features a Xilinx FPGA with an embedded, hard core Power PC that is remotely reconfigurable. It serves as a front end computer (FEC) that interfaces with the RHIC control system. The carrier provides high speed serial data paths to each daughter site and between daughter sites as well as four generic external fiber optic links. It also distributes low noise clocks and serial data links to all daughter sites and monitors temperature, voltage and current. To date, two XMC cards have been designed: a four channel high speed ADC and a four channel high speed DAC. The new LLRF hardware was used to replace the old RHIC LLRF system for the 2009 run. For the 2010 run, the RHIC RF system operation was dramatically changed with the introduction of accelerating both beams in a new, common cavity instead of each ring having independent cavities. The flexibility of the new system was beneficial in allowing the low level system to be adapted to support this new configuration. This hardware was also used in 2009 to provide LLRF for the newly commissioned Electron Beam Ion Source.

43 PARTICLE ACCELERATORS↗

Surge-Inception Study in a Two-Spool Turbojet Engine

A two-spool turbojet engine was operated in the Lewis altitude wind tunnel to study the inception of compressor surge. In addition to the usual steady-state pressure and temperature measurements, the compressors were extensively instrumented with fast-response interstage pressure transducers. Thus it was possible to obtain maps for both compressors, pressure oscillations during rotating stall, effects of stall on efficiency, and stage-loading curves. In addition, with the transient measurements, it was possible to record interstage pressures and then compute stage performance during accelerations to the stall limit. Rotating stall was found to exist at low speeds in the outer spool. Although the stall arose from poor flow conditions at the inlet-stage blade tips, the low-energy air moved through the machine from the tip at the inlet to the outer spool to the hub at the inlet to the inner spool. This tip stall ultimately resulted in compressor surge in the mid-speed region, and necessitated inter-compressor air bleed. Interstage pressure measurements during acceleration to the compressor stall limit indicated that rotating stall was not a necessary condition for compressor surge and that, at the critical stall point, the circumferential interstage pressure distribution was uniform. The exit-stage group of the inner spool was first t o stall; then, the stages upstream stalled in succession until the inlet stage of the outer spool was stalled. With a sufficiently high fuel rate, the process repeated with a cycle time of about 0.1 second. It was possible to construct reproducible stage stall lines as a function of compressor speed from the stage stall points of several such compressor surges. This transient stall line was checked by computing the stall line from a steady-state stage-loading curve. Good agreement between the stage stall lines was obtained by these two methods.

Wallner, Lewis E.↗

Designing for aircraft structural crashworthiness

This report describes structural aviation crash dynamics research activities being conducted on general aviation aircraft and transport aircraft. The report includes experimental and analytical correlations of load-limiting subfloor and seat configurations tested dynamically in vertical drop tests and in a horizontal sled deceleration facility. Computer predictions using a finite-element nonlinear computer program, DYCAST, of the acceleration time-histories of these innovative seat and subfloor structures are presented. Proposed application of these computer techniques, and the nonlinear lumped mass computer program KRASH, to transport aircraft crash dynamics is discussed. A proposed FAA full-scale crash test of a fully instrumented radio controlled transport airplane is also described.

Thomson, R. G.↗

Local Improvement Results for Anderson Acceleration with Inaccurate Function Evaluations

Here, we analyze the convergence of Anderson acceleration when the fixed point map is corrupted with errors. We also consider uniformly bounded errors and stochastic errors with infinite tails. We prove local improvement results which describe the performance of the iteration up to the point where the accuracy of the function evaluation causes the iteration to stagnate. We illustrate the results with examples from neutronics.

97 MATHEMATICS AND COMPUTING↗

Laser wakefield simulation using a speed-of-light frame envelope model

Simulation of laser wakefield accelerator (LWFA) experiments is computationally intensive due to the disparate length scales involved. Current experiments extend hundreds of laser wavelengths transversely and many thousands in the propagation direction, making explicit PIC simulations enormously expensive and requiring massively parallel execution in 3D. We can substantially improve the performance of laser wakefield simulations by modeling the envelope modulation of the laser field rather than the field itself. This allows for much coarser grids, since we need only resolve the plasma wavelength and not the laser wavelength, and therefore larger timesteps. Thus an envelope model can result in savings of several orders of magnitude in computational resources. By propagating the laser envelope in a frame moving at the speed of light, dispersive errors can be avoided and simulations over long distances become possible. Here we describe the model and its implementation, and show simulations and benchmarking of laser wakefield phenomena such as channel propagation, self-focusing, wakefield generation, and downramp injection using the model.

43 PARTICLE ACCELERATORS↗

Finite element computation of compressible flows with the SUPG formulation

Finite element computation of compressible Euler equations is presented in the context of the streamline-upwind/Petrov-Galerkin (SUPG) formulation. The SUPG formulation, which is based on adding stabilizing terms to the Galerkin formulation, is further supplemented with a shock capturing operator which addresses the difficulty in maintaining a satisfactory solution near discontinuities in the solution field. The shock capturing operator, which has been derived from work done in entropy variables for a similar operator, is shown to lead to an appropriate level of additional stabilization near shocks, without resulting in excessive numerical diffusion. An implicit treatment of the impermeable wall boundary condition is also presented. This treatment of the no-penetration condition offers increased stability for large Courant numbers, and accelerated convergence of the computations for both implicit and explicit applications. Several examples are presented to demonstrate the ability of this method to solve the equations governing compressible fluid flow.

Le Beau, G. J.↗

Ferroelectric capacitors and field-effect transistors as in-memory computing elements for machine learning workloads

Abstract This study discusses the feasibility of Ferroelectric Capacitors (FeCaps) and Ferroelectric Field-Effect Transistors (FeFETs) as In-Memory Computing (IMC) elements to accelerate machine learning (ML) workloads. We conducted an exploration of device fabrication and proposed system-algorithm co-design to boost performance. A novel FeCap device, incorporating an interfacial layer (IL) and$$\text {Hf}_{0.5}\text {Zr}_{0.5}\text {O}_2$$ Hf 0.5 Zr 0.5 O 2 (HZO), ensures a reduction in operating voltage and enhances HZO scaling while being compatible with CMOS circuits. The IL also enriches ferroelectricity and retention properties. When integrated into crossbar arrays, FeCaps and FeFETs demonstrate their effectiveness as IMC components, eliminating sneak paths and enabling selector-less operation, leading to notable improvements in energy efficiency and area utilization. However, it is worth noting that limited capacitance ratios in FeCaps introduced errors in multiply-and-accumulate (MAC) computations. The proposed co-design approach helps in mitigating these errors and achieves high accuracy in classifying the CIFAR-10 dataset, elevating it from a baseline of 10% to 81.7%. FeFETs in crossbars, with a higher on-off ratio, outperform FeCaps, and our proposed charge-based sensing scheme achieved at least an order of magnitude reduction in power consumption, compared to prevalent current-based methods.

Science & Technology - Other Topics↗

High-throughput virtual laboratory for drug discovery using massive datasets

Time-to-solution for structure-based screening of massive chemical databases for COVID-19 drug discovery has been decreased by an order of magnitude, and a virtual laboratory has been deployed at scale on up to 27,612 GPUs on the Summit supercomputer, allowing an average molecular docking of 19,028 compounds per second. Over one billion compounds were docked to two SARS-CoV-2 protein structures with full optimization of ligand position and 20 poses per docking, each in under 24 hours. GPU acceleration and high-throughput optimizations of the docking program produced 350× mean speedup over the CPU version (50× speedup per node). GPU acceleration of both feature calculation for machine-learning based scoring and distributed database queries reduced processing of the 2.4 TB output by orders of magnitude. The resulting 50× speedup for the full pipeline reduces an initial 43 day runtime to 21 hours per protein for providing high-scoring compounds to experimental collaborators for validation assays.

97 MATHEMATICS AND COMPUTING↗