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.

SourceResearch connectionFrom yearThrough year

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

At least 415 records · Page 23

All sources NASA NTRS DOE OSTI

DOE OSTI2021

Multi-objective optimization with an integrated electromagnetics and beam dynamics workflow

In particle accelerators, RF cavities are used to accelerate charged particle beams to designed high energy for physical applications. In a typical accelerator design, the optimization of RF cavities and the optimization of beam dynamics are carried out in separate studies. For a more general and unrestricted accelerator design, a coupled optimization of the RF cavities and the beam parameters is required. For this coupled optimization problem, we have developed an integrated electromagnetics and beam dynamics workflow management system. Within this system, the geometries for a set of cavity components are first adjusted; the field modes are then computed with an electromagnetics program, and imported into a beam dynamics program for beam dynamics simulation. This workflow is encapsulated into a parallel multi-objective optimizer to achieve the integrated accelerator design optimization. A multi fidelity strategy is developed to improve the speed of the optimizer. Furthermore, this integrated global optimization capability is illustrated using a photoinjector design example and yields an improved design.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2025

Contextual subspace variational quantum eigensolver calculation of the dissociation curve of molecular nitrogen on a superconducting quantum computer

Abstract We present an experimental demonstration of the Contextual Subspace Variational Quantum Eigensolver on superconducting hardware. Calculating the potential energy curve of molecular nitrogen proves challenging for many conventional quantum chemistry techniques, since static correlation dominates in the dissociation limit. Our quantum simulations retain good agreement with the Full Configuration Interaction energy, outperforming all benchmarked single-reference wavefunction techniques in capturing the bond-breaking appropriately. Moreover, our methodology is competitive with multiconfigurational approaches but at a saving of quantum resource, meaning larger active spaces can be treated for a fixed qubit allowance. To achieve this result, we deploy an error mitigation/suppression strategy comprised of Dynamical Decoupling, Measurement-Error Mitigation and Zero-Noise Extrapolation. Circuit parallelization also provides passive noise-averaging and improves the effective shot yield to reduce the measurement overhead. Furthermore, we introduce a modified adaptive ansatz construction algorithm that incorporates hardware awareness into our variational circuits, minimizing the transpilation cost for the target qubit topology.

Physics↗

DOE OSTI2020

Making ultrastrong steel tough by grain-boundary delamination

Developing ultrahigh-strength steels that are ductile, fracture resistant, and cost effective would be attractive for a variety of structural applications. We show that improved fracture resistance in a steel with an ultrahigh yield strength of nearly 2 gigapascals can be achieved by activating delamination toughening coupled with transformation-induced plasticity. Delamination toughening associated with intensive but controlled cracking at manganese-enriched prior-austenite grain boundaries normal to the primary fracture surface dramatically improves the overall fracture resistance. As a result, fracture under plane-strain conditions is automatically transformed into a series of fracture processes in “parallel” plane-stress conditions through the thickness. The present “high-strength induced multidelamination” strategy offers a different pathway to develop engineering materials with ultrahigh strength and superior toughness at economical materials cost.

36 MATERIALS SCIENCE↗

DOE OSTI2023

RAPIDS: Reconciling Availability, Accuracy, and Performance in Managing Geo-Distributed Scientific Data

In modern science, big data plays an increasingly important role. Many scientific applications, such as running simulations on supercomputers or conducting experiments on advanced instruments, produce huge amount of data at unprecedented speed. Analyzing and understanding such big data is the key for scientists to make scientific breakthroughs. However, data might become unavailable for scientists to access when outages or maintenance of the storage system occur, which severely hinders scientific discovery. To improve the data availability, data duplication and erasure coding (EC) are often used. But as the scientific data gets larger, using these two methods can cause considerable storage and network overhead.In this paper, we propose RAPIDS, a hybrid approach that combines the multigrid-based error-bounded lossy compression with erasure coding, to significantly reduce the storage and network overhead required for maintaining high data availability. Our experiments show that RAPIDS reduces the storage overhead by up to 7.5x and network overhead by up to 3x to achieve the same level of availability compared to the regular EC method. We improve RAPIDS by building two models to optimize the fault tolerance configurations and data gathering strategy. We demonstrate that RAPIDS significantly improves performance when running on many CPU cores in parallel or on GPUs.

Wan, Lipeng↗

NASA NTRS1986

Increasing processor utilization during parallel computation rundown

Some parallel processing environments provide for asynchronous execution and completion of general purpose parallel computations from a single computational phase. When all the computations from such a phase are complete, a new parallel computational phase is begun. Depending upon the granularity of the parallel computations to be performed, there may be a shortage of available work as a particular computational phase draws to a close (computational rundown). This can result in the waste of computing resources and the delay of the overall problem. In many practical instances, strict sequential ordering of phases of parallel computation is not totally required. In such cases, the beginning of one phase can be correctly computed before the end of a previous phase is completed. This allows additional work to be generated somewhat earlier to keep computing resources busy during each computational rundown. The conditions under which this can occur are identified and the frequency of occurrence of such overlapping in an actual parallel Navier-Stokes code is reported. A language construct is suggested and possible control strategies for the management of such computational phase overlapping are discussed.

Jones, W. H.↗

NASA NTRS1986

Increasing processor utilization during parallel computation rundown

Jones, William H.↗

NASA NTRS1991

Compile-time estimation of communication costs in multicomputers

An important problem facing numerous research projects on parallelizing compilers for distributed memory machines is that of automatically determining a suitable data partitioning scheme for a program. Any strategy for automatic data partitioning needs a mechanism for estimating the performance of a program under a given partitioning scheme, the most crucial part of which involves determining the communication costs incurred by the program. A methodology is described for estimating the communication costs at compile-time as functions of the numbers of processors over which various arrays are distributed. A strategy is described along with its theoretical basis, for making program transformations that expose opportunities for combining of messages, leading to considerable savings in the communication costs. For certain loops with regular dependences, the compiler can detect the possibility of pipelining, and thus estimate communication costs more accurately than it could otherwise. These results are of great significance to any parallelization system supporting numeric applications on multicomputers. In particular, they lay down a framework for effective synthesis of communication on multicomputers from sequential program references.

Gupta, Manish↗

NASA NTRS2011

High-Payoff Space Transportation Design Approach with a Technology Integration Strategy

A general architectural design sequence is described to create a highly efficient, operable, and supportable design that achieves an affordable, repeatable, and sustainable transportation function. The paper covers the following aspects of this approach in more detail: (1) vehicle architectural concept considerations (including important strategies for greater reusability); (2) vehicle element propulsion system packaging considerations; (3) vehicle element functional definition; (4) external ground servicing and access considerations; and, (5) simplified guidance, navigation, flight control and avionics communications considerations. Additionally, a technology integration strategy is forwarded that includes: (a) ground and flight test prior to production commitments; (b) parallel stage propellant storage, such as concentric-nested tanks; (c) high thrust, LOX-rich, LOX-cooled first stage earth-to-orbit main engine; (d) non-toxic, day-of-launch-loaded propellants for upper stages and in-space propulsion; (e) electric propulsion and aero stage control.

McCleskey, C. M.↗

DOE OSTI2016

Lignases and aldo-keto reductases for conversion of lignin-containing materials to fermentable products

Termites have specialized digestive systems that overcome the lignin barrier in wood to release fermentable simple sugars. Using the termite Reticulitermes flavipes and its gut symbionts, high-throughput titanium pyrosequencing and proteomics approaches experimentally compared the effects of lignin-containing diets on host-symbiont digestome composition. Proteomic investigations and functional digestive studies with recombinant lignocellulases conducted in parallel provided strong evidence of congruence at the transcription and translational levels and provide enzymatic strategies for overcoming recalcitrant lignin barriers in biofuel feedstocks. Briefly described, therefore, the disclosure provides a system for generating a fermentable product from a lignified plant material, the system comprising a cooperating series of at least two catalytically active polypeptides, where said catalytically active polypeptides are selected from the group consisting of: cellulase Cell-1, $\beta$-glu cellulase, an aldo-keto-reductase, a catalase, a laccase, and an endo-xylanase.

09 BIOMASS FUELS↗

NASA NTRS1994

A Navier-Strokes Chimera Code on the Connection Machine CM-5: Design and Performance

We have implemented a three-dimensional compressible Navier-Stokes code on the Connection Machine CM-5. The code is set up for implicit time-stepping on single or multiple structured grids. For multiple grids and geometrically complex problems, we follow the 'chimera' approach, where flow data on one zone is interpolated onto another in the region of overlap. We will describe our design philosophy and give some timing results for the current code. A parallel machine like the CM-5 is well-suited for finite-difference methods on structured grids. The regular pattern of connections of a structured mesh maps well onto the architecture of the machine. So the first design choice, finite differences on a structured mesh, is natural. We use centered differences in space, with added artificial dissipation terms. When numerically solving the Navier-Stokes equations, there are liable to be some mesh cells near a solid body that are small in at least one direction. This mesh cell geometry can impose a very severe CFL (Courant-Friedrichs-Lewy) condition on the time step for explicit time-stepping methods. Thus, though explicit time-stepping is well-suited to the architecture of the machine, we have adopted implicit time-stepping. We have further taken the approximate factorization approach. This creates the need to solve large banded linear systems and creates the first possible barrier to an efficient algorithm. To overcome this first possible barrier we have considered two options. The first is just to solve the banded linear systems with data spread over the whole machine, using whatever fast method is available. This option is adequate for solving scalar tridiagonal systems, but for scalar pentadiagonal or block tridiagonal systems it is somewhat slower than desired. The second option is to 'transpose' the flow and geometry variables as part of the time-stepping process: Start with x-lines of data in-processor. Form explicit terms in x, then transpose so y-lines of data are in-processor. Form explicit terms in y, then transpose so z-lines are in processor. Form explicit terms in z, then solve linear systems in the z-direction. Transpose to the y-direction, then solve linear systems in the y-direction. Finally transpose to the x direction and solve linear systems in the x-direction. This strategy avoids inter-processor communication when differencing and solving linear systems, but requires a large amount of communication when doing the transposes. The transpose method is more efficient than the non-transpose strategy when dealing with scalar pentadiagonal or block tridiagonal systems. For handling geometrically complex problems the chimera strategy was adopted. For multiple zone cases we compute on each zone sequentially (using the whole parallel machine), then send the chimera interpolation data to a distributed data structure (array) laid out over the whole machine. This information transfer implies an irregular communication pattern, and is the second possible barrier to an efficient algorithm. We have implemented these ideas on the CM-5 using CMF (Connection Machine Fortran), a data parallel language which combines elements of Fortran 90 and certain extensions, and which bears a strong similarity to High Performance Fortran. We make use of the Connection Machine Scientific Software Library (CMSSL) for the linear solver and array transpose operations.

Jespersen, Dennis C.↗

DOE OSTI2014

BurstMem: A high-performance burst buffer system for scientific applications

The growth of computing power on large-scale systems requires commensurate high-bandwidth I/O systems. Many parallel file systems are designed to provide fast sustainable I/O in response to applications' soaring requirements. To meet this need, a novel system is imperative to temporarily buffer the bursty I/O and gradually flush datasets to long-term parallel file systems. In this paper, we introduce the design of BurstMem, a high-performance burst buffer system. BurstMem provides a storage framework with efficient storage and communication management strategies. Our experiments demonstrate that BurstMem is able to speed up the I/O performance of scientific applications by up to 8.5× on leadership computer systems.

Wang, Teng↗

NASA NTRS2015

Human System Risk Management for Space Flight

This brief abstract reviews the development of the current day approach to human system risk management for space flight and the development of the critical components of this process over the past few years. The human system risk management process now provides a comprehensive assessment of each human system risk by design reference mission (DRM) and is evaluated not only for mission success but also for long‐term health impacts for the astronauts. The discipline of bioastronautics is the study of the biological and medical effects of space flight on humans. In 1997, the Space Life Sciences Directorate (SLSD) initiated the Bioastronautics Roadmap (Roadmap) as the "Critical Path Roadmap", and in 1998 participation in the roadmap was expanded to include the National Space Biomedical Research Institute (NSBRI) and the external community. A total of 55 risks and 250 questions were identified and prioritized and in 2000, the Roadmap was base-lined and put under configuration control. The Roadmap took into account several major advisory committee reviews including the Institute of Medicine (IOM) "Safe Passage: Astronaut care for Exploration Missions", 2001. Subsequently, three collaborating organizations at NASA HQ (Chief Health and Medical Officer, Office of Space Flight and Office of Biological & Physical Research), published the Bioastronautics Strategy in 2003, that identified the human as a "critical subsystem of space flight" and noted that "tolerance limits and safe operating bands must be established" to enable human space flight. These offices also requested a review by the IOM of the Roadmap and that review was published in October 2005 as "A Risk Reduction Strategy for Human Exploration of Space: A Review of NASA's Bioastronautics Roadmap", that noted several strengths and weaknesses of the Roadmap and made several recommendations. In parallel with the development of the Roadmap, the Office of the Chief Health and Medical Officer (OCHMO) began a process in 2004 of evaluating the tolerance limits and safe operating bands called for in the Bioastronautics Strategy. Over the next several years, the concept of the "operating bands" were turned into Space Flight Human System Standards (SFHSS), developed by the technical resources of the SLSD at the NASA Johnson Space Center (JSC). These standards were developed and reviewed at the SLSD and then presented to the OCHMO for acceptance. The first set of standards was published in 2007 as the NASA‐STD‐3001, Volume 1, Crew Health that elaborated standards for several physiological areas such as cardiovascular, musculoskeletal, radiation exposure and nutrition. Volume 2, Human Factors, Habitability and Human Health was published in 2011, along with development guidance in the Human Integration Design Handbook (HIDH). Taken together, the SFHSS Volumes 1 and 2, and the HIDH replaced the NASA‐STD‐3000 with new standards and revisions of the older document. Three other changes were also taking place that facilitated the development of the human system risk management approach. In 2005, the life sciences research and development portfolio underwent a comprehensive review through the Exploration Systems Architecture Study (ESAS) that resulted in the reformulation of the Bioastronautics Program into Human Research Program (HRP) that was focused on appropriate mitigation results for high priority human health risks. The baseline HRP budget was established in August 2005. In addition, the OCHMO formulated the Health and Medical Technical Authority (HMTA) in 2006 that established the position of the Chief Medical Officer (CMO) at the NASA JSC along with other key technical disciplines, and the OCHMO became the responsible office for the SFHSS as noted above. The final change was the establishment in 2008 of the Human System Risk Board (HSRB), chaired by the CMO with representation from the HRP, SLSD management and technical experts. The HSRB then began to review all human system risks, established a comprehensive risk management and configuration management plan and data sharing policy. These major developments of standards, the HRP, the HMTA and a forum for review of human system risks (HSRB) facilitated the integration of human research, medical operations, systems engineering and many other disciplines in the comprehensive review of human system risks. The HSRB began a comprehensive review of all potential inflight medical conditions and events and over the course of several reviews consolidated the number of human system risks to 30 where the greatest emphasis is placed for investing program dollars for risk mitigation. The HSRB considers all available evidence from human research, medical operations and occupational surveillance in assessing the risks for appropriate mitigation and future work. All applicable DRMs (low earth orbit 6 and 12 months, deep space sortie for 30 days and 1 year, a one year lunar mission, and a planetary mission for 3 years) are considered as human system risks are modified by the hazards associated with space flight such as microgravity, exposure to radiation, distance from the earth, isolation and a closed environment. Each risk has a summary assessment representing the state of knowledge/evidence base for that risk, the available risk mitigations, traceability to the SFHSS and program requirements, and future work required. These data then can drive coordinated budgets across the HRP, the International Space Station, Crew Health and Safety and Advanced Exploration System budgets. These risk assessments were completed for 6 DRMs in December of 2014 and serve as the baseline for which subsequent research and technology development and crew health care portfolios can be assessed. The HSRB will review each risk at least annually and especially when new information is available that must be considered for effective risk mitigation. The current status of each risk can be reported to program management for operations, budget reviews and general oversight of the human system risk management program.

Davis, Jeffrey↗

DOE OSTI2022

Towards Low-Overhead Resilience for Data Parallel Deep Learning

Data parallel techniques have been widely adopted both in academia and industry as a tool to enable scalable training of deep learning models. At scale, DL training jobs can fail due to software or hardware bugs, may need to be preempted or terminated due to unexpected events, or may perform suboptimally because they were misconfigured. Under such circumstances, there is a need to recover and/or reconfigure data-parallel DL training jobs on-the-fly, while minimizing the impact on the accuracy of the DNN model and the runtime overhead. In this regard, state-of-art techniques adopted by the HPC community mostly rely on checkpoint-restart, which inevitably leads to loss of progress, thus increasing the runtime overhead. In this paper we explore alternative techniques that exploit the properties of modern deep learning frameworks (overlapping of gradient averaging and weight updates with local gradient computations through pipeline parallelism) to reduce the overhead of resilience/elasticity. To this end we introduce a failure simulation framework and two resilience strategies (immediate mini-batch rollback and lossy forward recovery), which we study compared with checkpoint-restart approaches in a variety of settings in order to understand the trade-offs between the accuracy loss of the DNN model and the runtime overhead.

data-parallel training↗

DOE OSTI2021

Local time stepping for the shallow water equations in MPAS

In this work we assess the performance of a set of local time-stepping (LTS) schemes for the shallow water equations implemented in the Model for Prediction Across Scales (MPAS). The goal of LTS is to speed up the simulation by allowing different time-steps on different regions of the computational grid. The LTS schemes considered here were originally introduced by Hoang et al. (2019) [26], who laid out the mathematical foundation of the methods. Here, the authors take on the task of presenting a fast, efficient and scalable parallel implementation of these LTS methods on high performance computing machines, with the aim to provide a recipe for other climate modeling groups that may be interested in employing LTS algorithms in their codes. As a matter of fact, even if MPAS is our framework of choice, our approach is general enough and could be of interest to other groups beyond the MPAS community. Due to their nature, LTS methods possess an inherent load imbalance that needs to be carefully addressed in order to obtain efficient scalability. Even more important is the far from trivial task of computing the right-hand side terms only on specific LTS regions during the time-stepping procedure. An inefficient handling of this task causes a drastic decay of the CPU time performance, making the LTS algorithms practically of no use. The emphasis of the present work is therefore on the computational and parallel aspects of the LTS methods, whose proper treatment is crucial to make the methods run faster against existing strategies, such as for instance high-order explicit global time-stepping schemes. This is in fact the ultimate goal of using an LTS procedure and it is the one to which we direct all our optimization efforts.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2017

Systems Level Regulation of Rhythmic Growth Rate and Biomass Accumulation in Grasses

Objectives: Several breakthroughs have been recently made in our understanding of plant growth and biomass accumulation. It was found that plant growth is rhythmically controlled throughout the day by the circadian clock through a complex interplay of light and phytohormone signaling pathways. While plants such as the C4 energy crop sorghum (Sorghum bicolor (L.) Moench) and possibly the C3 grass Brachypodium distachyon also exhibit daily rhythms in growth rate, the molecular details of its regulation remain to be explored. A better understanding of diurnally regulated growth behavior in grasses may lead to species-specific mechanisms highly relevant to future strategies to optimize energy crop biomass yield. Here we propose to devise a systems approach to identify, in parallel, regulatory hubs associated with rhythmic growth in C3 and C4 plants. We propose to use rhythmicity in daily growth patterns to drive the discovery of regulatory network modules controlling biomass accumulation. Description: The project is divided in three main parts: 1) Performing time-lapse imaging and growth measurement in B. distachyon and S. bicolor to determine growth rate dynamic during the day/night cycle. Identifying growth-associated genes whose expression patterns follow the observed growth dynamics using deep sequencing technology, 2) identifying regulators of these genes by screening for DNA-binding proteins interacting with the growth-associated gene promoters identified in Aim 1. Screens will be performed using a validated yeast-one hybrid strategy paired with a specifically designed B. distachyon and S. bicolor transcription factor libraries (1000 clones each), and 3) Selecting 50 potential growth regulators from the screen for downstream characterization. The selection will be made by using a sytems biology approach by calculating the connectivity between growth rate, rhythmic gene expression profiles and TF expression profile and determine which TF is likely part of a hub and a potential regulator of plant growth. We will confirme the 50 DNA-protein interactions using transient transcriptional assays in B. distachyon and S. bicolor, and perform further testing in vivo by measuring growth parameters in transgenic loss- or gain-of-function lines for the selected transcription factors (25 B. distachyon and 3 S. bicolor lines). In 2016 the Principal Investigator relocated the laboratory to The Scripps Research Institute (TSRI) where the project has continued with an end date of 08/31/17. Accomplishments: We successfully collected large datasets of gene expression form the model grass Brachypodium. We used and developed bioinformatics analysis tools to investigate the structure, dynamics and robustness of circadian regulated gene expression in Brachypodium. Relevant Discoveries: We were able to determine that the endogenous circadian clock appears to play a much more subdued role in growth regulation in Brachypodium, that has been demonstrated in either Arabidopsis, or crop plants like Rice, Corn and Soybean. This led to our conclusion that Brachypodium unfortunately is unlikely to serve as an informative model for understanding how growth regulation in plants is under the control of circadian network circuitry. However, we were able to leverage our datasets in Brachypodium to inform us and reinforce a large collaborative study on gene networks governing cell wall deposition in Arabidopsis. This led to a major and highly cited publication relevant to improving biomass production.

09 BIOMASS FUELS↗

NASA NTRS1989

Predeployment validation of fault-tolerant systems through software-implemented fault insertion

Fault injection-based automated testing (FIAT) environment, which can be used to experimentally characterize and evaluate distributed realtime systems under fault-free and faulted conditions is described. A survey is presented of validation methodologies. The need for fault insertion based on validation methodologies is demonstrated. The origins and models of faults, and motivation for the FIAT concept are reviewed. FIAT employs a validation methodology which builds confidence in the system through first providing a baseline of fault-free performance data and then characterizing the behavior of the system with faults present. Fault insertion is accomplished through software and allows faults or the manifestation of faults to be inserted by either seeding faults into memory or triggering error detection mechanisms. FIAT is capable of emulating a variety of fault-tolerant strategies and architectures, can monitor system activity, and can automatically orchestrate experiments involving insertion of faults. There is a common system interface which allows ease of use to decrease experiment development and run time. Fault models chosen for experiments on FIAT have generated system responses which parallel those observed in real systems under faulty conditions. These capabilities are shown by two example experiments each using a different fault-tolerance strategy.

Czeck, Edward W.↗

DOE OSTI2025

DeepHyper: A Python Package for Massively Parallel Hyperparameter Optimization in Machine Learning

Machine learning models are increasingly applied across scientific disciplines, yet their effectiveness often hinges on heuristic decisions—such as data transformations, training strategies, and model architectures—that are not learned by the models themselves. Automating the selection of these heuristics and analyzing their sensitivity is crucial for building robust and efficient learning workflows. DeepHyper addresses this challenge by democratizing hyperparameter optimization, providing accessible tools to streamline and enhance machine learning workflows from a laptop to the largest supercomputer in the world. Building on top of hyperparameter optimization, it unlocks new capabilities around ensembles of models for improved accuracy and uncertainty quantification. All of these organized around efficient parallel computing.

ensemble↗

DOE OSTI2020

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↗