Search NASA⌕ Search

SEARCH · Search NASA

Results for “Parallelization Strategy”

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 505 records · Page 28
All sourcesNASA NTRSDOE OSTI
DOE OSTI2013

Reducing the bulk of the bulk synchronous parallel model

For over two decades the dominant means for enabling portable performance of computational science and engineering applications on parallel processing architectures has been the bulk-synchronous parallel programming (BSP) model. Code developers, motivated by performance considerations to minimize the number of messages transmitted, have typically pursued a strategy of aggregating message data into fewer, larger messages. Emerging and future high-performance architectures, especially those seen as targeting Exascale capabilities, provide motivation and capabilities for revisiting this approach. In this paper we explore alternative configurations within the context of a large-scale complex multi-physics application and a proxy that represents its behavior, presenting results that demonstrate some important advantages as the number of processors increases in scale.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2013

From Physics Model to Results: An Optimizing Framework for Cross-Architecture Code Generation

Starting from a high-level problem description in terms of partial differential equations using abstract tensor notation, the Chemora framework discretizes, optimizes, and generates complete high performance codes for a wide range of compute architectures. Chemora extends the capabilities of Cactus, facilitating the usage of large-scale CPU/GPU systems in an efficient manner for complex applications, without low-level code tuning. Chemora achieves parallelism through MPI and multi-threading, combining OpenMP and CUDA. Optimizations include high-level code transformations, efficient loop traversal strategies, dynamically selected data and instruction cache usage strategies, and JIT compilation of GPU code tailored to the problem characteristics. The discretization is based on higher-order finite differences on multi-block domains. Chemora's capabilities are demonstrated by simulations of black hole collisions. This problem provides an acid test of the framework, as the Einstein equations contain hundreds of variables and thousands of terms.

Blazewicz, Marek↗
DOE OSTI2018

Slycat™ User Manual

Slycat™ is a web-based system for performing data analysis and visualization of potentially large quantities of remote, high-dimensional data. Slycat™ specializes in working with ensemble data. An ensemble is a group of related data sets, which typically consists of a set of simulation runs exploring the same problem space. An ensemble can be thought of as a set of samples within a multi-variate domain, where each sample is a vector whose value defines a point in high-dimensional space. To understand and describe the underlying problem being modeled in the simulations, ensemble analysis looks for shared behaviors and common features across the group of runs. Additionally, ensemble analysis tries to quantify differences found in any members that deviate from the rest of the group. The Slycat™ system integrates data management, scalable analysis, and visualization. Results are viewed remotely on a user’s desktop via commodity web clients using a multi-tiered hierarchy of computation and data storage, as shown in Figure 1. Our goal is to operate on data as close to the source as possible, thereby reducing time and storage costs associated with data movement. Consequently, we are working to develop parallel analysis capabilities that operate on High Performance Computing (HPC) platforms, to explore approaches for reducing data size, and to implement strategies for staging computation across the Slycat™ hierarchy. Within Slycat™, data and visual analysis are organized around projects, which are shared by a project team. Project members are explicitly added, each with a designated set of permissions. Although users sign-in to access Slycat™, individual accounts are not maintained. Instead, authentication is used to determine project access. Within projects, Slycat™ models capture analysis results and enable data exploration through various visual representations. Although for scientists each simulation run is a model of real-world phenomena given certain conditions, we use the term model to refer to our modeling of the ensemble data, not the physics. Different model types often provide complementary perspectives on data features when analyzing the same data set. Each model visualizes data at several levels of abstraction, allowing the user to range from viewing the ensemble holistically to accessing numeric parameter values for a single run. Bookmarks provide a mechanism for sharing results, enabling interesting model states to be labeled and saved.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Stripmap SAR Data Collection Geometry on a Round Earth

A stripmap Synthetic Aperture Radar (SAR) image is a long SAR image along some centerline, and formed from multiple synthetic apertures. At issue is that the centerline in the image actually corresponds to an arc on a round earth, and multiple strategies exist for fitting the image centerline to the round earth. Some of those strategies involve Rhumb lines, great circle paths, and great ellipse paths. Some are better than others in polar regions. Notions of parallel flight paths for the radar during data collection also require careful consideration of the geometry of a round earth.

47 OTHER INSTRUMENTATION↗
NASA NTRS1993

Feed-forward volume rendering algorithm for moderately parallel MIMD machines

Algorithms for direct volume rendering on parallel and vector processors are investigated. Volumes are transformed efficiently on parallel processors by dividing the data into slices and beams of voxels. Equal sized sets of slices along one axis are distributed to processors. Parallelism is achieved at two levels. Because each slice can be transformed independently of others, processors transform their assigned slices with no communication, thus providing maximum possible parallelism at the first level. Within each slice, consecutive beams are incrementally transformed using coherency in the transformation computation. Also, coherency across slices can be exploited to further enhance performance. This coherency yields the second level of parallelism through the use of the vector processing or pipelining. Other ongoing efforts include investigations into image reconstruction techniques, load balancing strategies, and improving performance.

Yagel, Roni↗
DOE OSTI2022

Accelerating Collective Communication in Data Parallel Training across Deep Learning Frameworks

This work develops new techniques within Horovod, a generic communication library supporting data parallel training across deep learning frameworks. In particular, we improve the Horovod control plane by implementing a new coordination scheme that takes advantage of the characteristics of the typical data parallel training paradigm, namely the repeated execution of collectives on the gradients of a fixed set of tensors. Using a caching strategy, we execute Horovod’s existing coordinator-worker logic only once during a typical training run, replacing it with a more efficient decentralized orchestration strategy using the cached data and a global intersection of a bitvector for the remaining training duration. Next, we introduce a feature for end users to explicitly group collective operations, enabling finer grained control over the communication buffer sizes. To evaluate our proposed strategies, we conduct experiments on a world-class supercomputer — Summit. We compare our proposals to Horovod’s original design and observe 2x performance improvement at a scale of 6000 GPUs; we also compare them against tf.distribute and torch.DDP and achieve 12% better and comparable performance, respectively, using up to 1536 GPUs; we compare our solution against BytePS in typical HPC settings and achieve about 20% better performance on a scale of 768 GPUs. Finally, we test our strategies on a scientific application (STEMDL) using up to 27,600 GPUs (the entire Summit) and show that we achieve a near-linear scaling of 0.93 with a sustained performance of 1.54 exaflops (with standard error +- 0.02) in FP16 precision.

Romero, Joshua↗
DOE OSTI2025

Safe Reinforcement Learning-Based Transient Stability Control for Islanded Microgrids With Topology Reconfiguration

This paper proposes a safe reinforcement learning (RL)-based transient stability emergency control (TSEC) method for islanded microgrids. RL requires extensive interaction with the environment to learn control strategies, hence, a data-driven approach is used as a substitute for time-consuming time-domain simulation calculations. Deep sigma point processes (DSPP), which is a Gaussian process model, is utilized to predict the normal distribution of transient stability of microgrids and to construct a transient stability chance constraint. Reward-constrained policy optimization (RCPO) can simultaneously achieve objective prediction, policy learning, and constraint cost coefficient update across multiple timescales. RCPO interacts with the DSPP-based microgrid environment through a multi-process parallel manner, greatly increasing the training speed. Case studies on a real islanded microgrid demonstrate that the proposed method can efficiently and quickly obtain the optimal emergency control strategy while adhering to all hard constraints.

14 SOLAR ENERGY↗
DOE OSTI2016

Numerical modeling of heat-transfer and the influence of process parameters on tailoring the grain morphology of IN718 in electron beam additive manufacturing

The fabrication of 3-D parts from CAD models by additive manufacturing (AM) is a disruptive technology that is transforming the metal manufacturing industry. The correlation between solidification microstructure and mechanical properties has been well understood in the casting and welding processes over the years. This paper focuses on extending these principles to additive manufacturing to understand the transient phenomena of repeated melting and solidification during electron beam powder melting process to achieve site-specific microstructure control within a fabricated component. In this paper, we have developed a novel melt scan strategy for electron beam melting of nickel-base superalloy (Inconel 718) and also analyzed 3-D heat transfer conditions using a parallel numerical solidification code (Truchas) developed at Los Alamos National Laboratory. The spatial and temporal variations of temperature gradient (G) and growth velocity (R) at the liquid-solid interface of the melt pool were calculated as a function of electron beam parameters. By manipulating the relative number of voxels that lie in the columnar or equiaxed region, the crystallographic texture of the components can be controlled to an extent. The analysis of the parameters provided optimum processing conditions that will result in columnar to equiaxed transition (CET) during the solidification. Furthermore, the results from the numerical simulations were validated by experimental processing and characterization thereby proving the potential of additive manufacturing process to achieve site-specific crystallographic texture control within a fabricated component.

43 PARTICLE ACCELERATORS↗
DOE OSTI2017

Stress Relaxation in Epoxy Thermosets via a Ferrocene-Based Amine Curing Agent

Physical stress relaxation in rubbery, thermoset polymers is limited by cross-links, which impede segmental motion and restrict relaxation to network defects, such as chain ends. In parallel, the cure shrinkage associated with thermoset polymerizations leads to the development of internal residual stress that cannot be effectively relaxed. Recent strategies have reduced or eliminated such cure stress in thermoset polymers largely by exploiting chemical relaxation processes, wherein temporary cross-links or otherwise transient bonds are incorporated into the polymer network. In this paper, we explore an alternative approach, wherein physical relaxation is enhanced by the incorporation of organometallic sandwich moieties into the backbone of the polymer network. A standard epoxy resin is cured with a diamine derivative of ferrocene and compared to conventional diamine curing agents. The ferrocene-based thermoset is clearly distinguished from the conventional materials by reduced cure stress with increasing cure temperature as well as unique stress relaxation behavior above its glass transition in the fully cured state. The relaxation experiments exhibit features characteristic of a physical relaxation process. Furthermore, the cure stress is observed to vanish precipitously upon deliberate introduction of network defects through an increasing imbalance of epoxy and amine functional groups. Finally, we postulate that these beneficial properties arise from fluxional motion of the cyclopentadienyl ligands on the polymer backbone.

36 MATERIALS SCIENCE↗
DOE OSTI2018

Novel functional and distributed approaches to data analysis available in ROOT

The bright future of particle physics at the Energy and Intensity frontiers poses exciting challenges to the scientific software community. The traditional strategies for processing and analyzing data are evolving in order to (i) offer higher-level programming model approaches and (ii) exploit parallelism to cope with the ever increasing complexity and size of the datasets. This contribution describes how the ROOT framework, a cornerstone of software stacks dedicated to particle physics, is preparing to provide adequate solutions for the analysis of large amount of scientific data on parallel architectures. The functional approach to parallel data analysis provided with the ROOT TDataFrame interface is then characterized. The design choices behind this new interface are described also comparing with other widely adopted tools such as Pandas and Apache Spark. The programming model is illustrated highlighting the reduction of boilerplate code, composability of the actions and data transformations as well as the capabilities of dealing with different data sources such as ROOT, JSON, CSV or databases. Details are given about how the functional approach allows transparent implicit parallelization of the chain of operations specified by the user. The progress done in the field of distributed analysis is examined. In particular, the power of the integration of ROOT with Apache Spark via the PyROOT interface is shown. In addition, the building blocks for the expression of parallelism in ROOT are briefly characterized together with the structural changes applied in the building and testing infrastructure which were necessary to put them in production.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
NASA NTRS1990

Computational technology for flight vehicles; Proceedings of the Symposium on Computational Technology on Flight Vehicles, Washington, DC, Nov. 5-7, 1990

The present conference on computational methods for aeronautics applications discusses topics in the fields of parallel computing, multidisciplinary computational methods, grid generation, visualization methods for CFD, probabilistic modeling, numerical simulations and methodologies for different flow regimes, computational strategies and adaptive methods in CFD, and computational strategies and dynamics and control. Attention is given to the MACH system-software kernel, a multidisciplinary approach to aeroelastic analysis, interactive grid generation with control points, interactive flow visualization using stream surfaces, numerical simulations of dynamic/aerodynamic interactions, implicit mathods for the Navier-Stokes equations, and the automatic phase-space analysis of dynamical systems.

Noor, Ahmed K.↗
NASA NTRS1998

Developing Information Power Grid Based Algorithms and Software

This exploratory study initiated our effort to understand performance modeling on parallel systems. The basic goal of performance modeling is to understand and predict the performance of a computer program or set of programs on a computer system. Performance modeling has numerous applications, including evaluation of algorithms, optimization of code implementations, parallel library development, comparison of system architectures, parallel system design, and procurement of new systems. Our work lays the basis for the construction of parallel libraries that allow for the reconstruction of application codes on several distinct architectures so as to assure performance portability. Following our strategy, once the requirements of applications are well understood, one can then construct a library in a layered fashion. The top level of this library will consist of architecture-independent geometric, numerical, and symbolic algorithms that are needed by the sample of applications. These routines should be written in a language that is portable across the targeted architectures.

Dongarra, Jack↗
NASA NTRSDate not supplied

ISS EVA80 EMU Water in the Helmet Failure and Associated Analytical Response

On March 23rd, 2022, during Extra-Vehicular Activity (EVA) 80 aboard the International Space Station (ISS) crew identified an 8-10-inch diameter thin film of water in the pressure bubble of the suit during repress. This launched an investigation to determine the cause of the water in the helmet as well as short- and long-term mitigation efforts to enable return to EVA as quickly as possible. The investigation analysis efforts in combination with hardware inspection and testing determined the most likely cause of the water in the helmet to be sublimator carryover as a result of high latent loading put on the system. In parallel with this investigation, efforts were being made to mitigate any future water in the helmet events. The short-term mitigation strategy developed is to install absorbent material in the pressure bubble to capture water as it enters the helmet before it can impact the astronauts. This hardware is referred to as the Helmet Absorption Band (HAB) and Helmet Absorption Pad – Extender (HAP-E). Longer-term mitigation strategies include a device to capture small water events before entering the helmet by installing a water capture system in the vent loop of the Extravehicular Mobility Unit (EMU). This effort is coined the T2 Water Capture System. Additionally, developing an on-orbit sublimator challenge test which will be able to verify sublimator performance before and after EVAs. The sublimator is the hardware which condenses water vapor and removes it from the vent loop, and this hardware is referred to as the EMU Moisture Injection Test System (EMITS). Based on the investigation and short-term water capture solutions the ISS Program decided to return to nominal EVAs on October 7, 2022. With the addition of the long-term mitigation strategies, the team hopes to be able to prevent any future water in the helmet events.

Veronica Lee Pizor↗
DOE OSTI2016

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↗
DOE OSTI2025

Sequence length scaling in vision transformers for scientific images on frontier

Vision Transformers (ViTs) are pivotal for foundational models in scientific imagery, including Earth science applications, due to their capability to process large sequence lengths. While transformers for text have inspired scaling sequence lengths in ViTs, adapting these for ViTs introduces unique challenges. We develop distributed sequence parallelism for ViTs, enabling them to handle up to 1M tokens. Our approach, leveraging DeepSpeed-Ulysses and Long-Sequence-Segmentation with model sharding, is the first to apply sequence parallelism in ViT training, achieving a 94% batch scaling efficiency on 2,048 AMD-MI250X GPUs. Evaluating sequence parallelism in ViTs, particularly in models up to 10B parameters, highlighted substantial bottlenecks. We countered these with hybrid sequence, pipeline, and flash attention strategies, to scale beyond single GPU memory limits. Our method significantly enhances climate modeling accuracy by 20% in temperature predictions, marking the first training of a vision transformer model to convergence with a sequence length of 188K tokens, using full self-attention.

Tsaris, Aristeidis (aris) [ORNL] (ORCID:0000000277↗
DOE OSTI2014

Simulation and Controls for Heavy Duty Dual Mode Hybrid Powertrain (Final CRADA Report)

Hybrid powertrains are of considerable interest because of potential reductions in fuel consumption, criteria pollutants and green house gas (GHG) emissions. Parallel hybrids have been applied to light and medium duty trucks, where urban driving cycles are prevalent, while series hybrids have been successfully used for other applications like transit and school buses. Unfortunately, hybridization of the Class 8, heavy-duty (HD) powertrain is inherently challenging due the expected long-haul driving requirements and limited opportunities for regenerative braking. The Participant has conceived and demonstrated a transformational Dual Mode Hybrid Powertrain (DMHP) technology developed specifically for the needs and function of Class 8 line haul trucks. The DMHP system enables a new paradigm in powertrain operational efficiency in the Class 8 truck segment. It decouples the connection between the engine operating point and the truck road load demands over a broad operating range through an innovative hybrid design. The DMHP operation choices include running in full series, full parallel and engine-off modes. The DMHP offers the opportunity for an engine to operate in a narrow range, thus providing a strategy for maximized fuel economy and minimized emissions. Further, it is expected that transient torque and power wheel demands are handled in whole or part by the electric system, thus reducing the frequency and intensity of engine transients and further improving the fuel economy and emissions. Fuel consumption and emissions have been further reduced through the elimination of overnight hoteling and idling at stops. Finally, based on the unique operating profile of an engine integrated into our hybrid powertrain, a transformational HD truck engine design concept next can emerge. Recent research activities by Oak Ridge National Laboratory (ORNL) have yielded significant data in real-life speed and load profiles of Class 8, long haul trucks. In addition, preliminary simulations of the DMHP carried out by ORNL reveal significant optimization opportunities of the DMHP by applying systematic simulation and controls approaches. An improved understanding of the complex interactions offered by the on-board engine, energy storing system, and electric machines is necessary for the development of control methodologies and practical implementation. We will continue to further this understanding through detailed experimentation and modeling, drawing on and expanding ORNL’s core competency in basic engine research and development (R&D) and advanced controls. This knowledge will be used to develop, implement, and evaluate control strategies on an actual DMHP using Meritor’s components and subsystems. Our initial focus will be on optimization of DMHP utilizing a “stock” diesel engine that is commercially available in the market place. A new DMHP-specific engine design concept will be pursued at a later phase of this Cooperative Research and Development Agreement (CRADA). Meritor will develop the experimental setup, conduct sub-system experiments at their facility, and implement and evaluate potential control strategies resulting from this CRADA partnership. Oak Ridge National Laboratory (ORNL) will develop a real-time simulation model for use in real-time control of the DMHP. The model and potential algorithms will be evaluated on a HD, hybrid powertrain dynamometer facility that will also be developed at the ORNL Vehicle Systems Integration facility.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2019

Real-Time Boundary-Layer Profiling at the Southern Great Plains Field Campaign Report

The National Research Council (NRC 2009, 2010) has argued that the nation needs to establish a boundary-layer profiling network across the conterminous United States (CONUS). These observations would benefit a wide range of applications including short-term severe weather forecasting by providing better initial conditions for weather prediction models (Wulfmeyer et al. 2015). While active remote sensors like Raman lidars (e.g., Turner et al. 2016) and differential absorption lidars (e.g., Spuler et al. 2015) are able to provide high-temporal-and-vertical-resolution profiles of water vapor and temperature from the ground, these systems are not yet available commercially. Passive remote-sensing systems, such as multi-channel microwave radiometers and infrared spectrometers, are commercially available and can provide retrieved profiles of temperature and humidity (e.g., Löhnert et al. 2009; Blumberg et al. 2015). These passive thermodynamic profiling systems are certainly a candidate for a possible boundary-layer network. Radio acoustic sounding systems (RASS), coupled with either a radar or sodar, can provide partial profiles of virtual temperature and are another option. However, the RASS signal is audible and considered by many to be annoying, and strong winds blow the acoustic signal away from the radar and greatly hamper its maximum vertical range. Three basic ground-based remote sensing technologies are used for wind profiling. All of these are active remote sensors that take advantage of a Doppler-shifted signal along different radial directions to determine the wind speed and direction. The three approaches are sodar (which uses pulses of sonic energy), radar (pulses of microwave energy), or lidar (pulses of laser energy). All three methods use either plane-parallel-indicator (PPI; constant elevation) scans that are analyzed with a velocity-azimuth display (VAD) technique or Doppler beam-swinging observing strategies, which are then analyzed to provide the horizontal wind speed and direction profiles as a function of range.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2017

AMReX

AMReX is a software framework for the development and implementation of massively parallel algorithms using block-structured adaptive mesh refinement (AMR). Block-structured AMR provides the basis for the temporal and spatial discretization strategy for a large variety of applications of interest to the DOE. AMReX provides a unified infrastructure with the functionality needed for applications to be able to effectively and efficiently utilize machines from laptops to exascale accelerator-based architectures. AMR reduces the computational cost and memory footprint compared to a uniform mesh while preserving accurate descriptions of different physical processes in complex multi-physics algorithms. AMReX supports algorithms that solve systems of partial differential equations (PDEs) in simple or complex geometries, and those that use particles and/or particle-mesh operations to represent component physical processes.

Zhang, Weiqun↗