Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Overhead Image”

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 127 records · Page 7

Toward Rapid Unattended X-ray Tomography of Large Planar Samples at 50-nm Resolution

X-ray tomography at sub-50 nm resolution of small areas ({approx}15 {mu}mx15 {mu}m) are routinely performed with both laboratory and synchrotron sources. Optics and detectors for laboratory systems have been optimized to approach the theoretical efficiency limit. Limited by the availability of relatively low-brightness laboratory X-ray sources, exposure times for 3-D data sets at 50 nm resolution are still many hours up to a full day. However, for bright synchrotron sources, the use of these optimized imaging systems results in extremely short exposure times, approaching live-camera speeds at the Advanced Photon Source at Argonne National Laboratory near Chicago in the US These speeds make it possible to acquire a full tomographic dataset at 50 nm resolution in less than a minute of true X-ray exposure time. However, limits in the control and positioning system lead to large overhead that results in typical exposure times of {approx}15 min currently.We present our work on the reduction and elimination of system overhead and toward complete automation of the data acquisition process. The enhancements underway are primarily to boost the scanning rate, sample positioning speed, and illumination homogeneity to performance levels necessary for unattended tomography of large areas (many mm{sup 2} in size). We present first results on this ongoing project.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

Propagation of a westward traveling surge and the development of persistent auroral features

Imaging instrumentation on board the spacecraft Dynamics Explorer 1 (DE 1) is used to observe the large-scale motion of a surge over 7000 km along the auroral oval from near local midnight. Average speed of the surge is 2.2 km/s. Ground-based observations at Fort Yukon, Alaska, show the classical looped, multiple-arc structure of a westward traveling surge as it passes overhead. Within the 6-min temporal resolution provided with DE 1, the surge advances initially at a speed of about 8 km/s followed by a steady decline to about 1 km/s over a period of 17 min. This sequence is then repeated a second time, beginning with a significant intensification of the surge form. This intense surge activity is not accompanied by significant auroral activity near magnetic midnight. Following passage of the surge, persistent and localized bright emission regions remain along the auroral oval for several tens of minutes. Average separation distances are approximately 700 km. If these persistent features identify the sites of individual stepwise advances of the surge, the average time per advance is about 5 min.

Craven, J. D.↗

Evaluation of the Performance Characteristics of the CGLSS and NLDN Systems Based on Two Years of Ground-Truth Data from Launch Complex 39B, Kennedy Space Center, Florida

From May 2011 through July 2013, the lightning instrumentation at Launch Complex 39B (LC39B) at the Kennedy Space Center, Florida, has obtained high-speed video records and field change waveforms (dE/dt and three-axis dH/dt) for 54 negative polarity return strokes whose strike termination locations and times are known with accuracy of the order of 10 m or less and 1 μs, respectively. A total of 18 strokes terminated directly to the LC39B lighting protection system (LPS), which contains three 181 m towers in a triangular configuration, an overhead catenary wire system on insulating masts, and nine down conductors. An additional 9 strokes terminated on the 106 m lightning protection mast of Launch Complex 39A (LC39A), which is located about 2.7 km southeast of LC39B. The remaining 27 return strokes struck either on the ground or attached to low-elevation grounded objects within about 500 m of the LC39B LPS. Leader/return stroke sequences were imaged at 3200 frames/sec by a network of six Phantom V310 high-speed video cameras. Each of the three towers on LC39B had two high-speed cameras installed at the 147 m level with overlapping fields of view of the center of the pad. The locations of the strike points of 54 return strokes have been compared to time-correlated reports of the Cloud-to-Ground Lightning Surveillance System (CGLSS) and the National Lightning Detection Network (NLDN), and the results of this comparison will be presented and discussed.

Waveforms↗

Fast Algorithms for Scientific Data Compression

Many scientific simulations and experiments generate terabytes to petabytes of data daily, necessitating data compression techniques. Unlike video and image compression, scientists require methods that accurately preserve primary data (PD) and derived quantities of interest (QoIs). In our previous work, we demonstrated the effectiveness of hybrid compression techniques that combine machine learning with traditional approaches. This paper presents innovative computational techniques aimed at expediting the compression pipeline. Our experiments, conducted on two distinct platforms with a large-scale XGC-based fusion simulation, demonstrate that the overhead incurred by these new approaches is less than one percent of the computational resources needed for the simulation.

Banerjee, Tania↗

The HOLO Series: Critical Ground-Based Demonstrations of Holographic Scanning Lidars

Results of two lidar measurement campaigns are presented, HOLO-1 (Utah, March 1999) and HOLO-2 (New Hampshire, June 1999). These tests demonstrate the ability of lidars utilizing holographic optical elements (HOEs) to determine tropospheric wind velocity and direction at cloud altitude. Several instruments were employed. HOLO-1 used the 1,064 mm transmission-HOE lidar (HARLIE, Goddard Space Flight Center), a zenith-staring 532 nm lidar (AROL-2, Utah State University), and a wide-field video camera (SkyCam) for imagery of clouds overhead. HOLO-2 included these instruments plus the 532 nm reflection-HOE lidar (PHASERS, St. Anselm College). HARLIE and PHASERS scan the sky at constant cone angles of 45 deg. and 42 deg. from normal, respectively. The progress of clouds and entire cloud fields across the sky is tracked by the repetitive conical scans of the HOE lidars. AROL-2 provides the attitude information enabling the SkyCam cloud images to be analyzed for independent data on cloud motion. Data from the HOE lidars are reduced by means of correlations, visualization by animation techniques, and kinematic diagrams of cloud feature motion. Excellent agreement is observed between the HOE lidar results and those obtained with video imagery and lidar ranging.

Wilkerson, Thomas D.↗

A visual display system approach for an advanced spaceflight simulator.

Future training and procedures simulators for advanced spacecraft will require that visual attachments provide forward, side and overhead window field of view capability. This paper presents an approach to high resolution, full color simulation of visual cues over the field of view offered by these crew station windows. Baseline requirements along with supporting rationale are presented. Candidate visual display hardware is identified and trade study results discussed. The display system approach presented is the result of studies which investigated the applicability of Apollo simulator visual display and image generation hardware to advanced spaceflight simulation.

Hock, E. A.↗

Real-Time Considerations for A Source-Time Dominant Auralization Scheme

A well-designed recording system can capture a moving source without risk of distortions, knowledge of the source or path, or transmission of information back to the source (i.e., a smartphone can reasonably record a plane flying overhead). This necessarily happens in real time. It would be good if signal processing schemes for auralization possessed these properties. Recent work on the NoTAP method of auralization proposed an asynchronous sample rate conversion scheme that keeps track of the (nonuniform) rate of incoming samples to formulate an effective incoming sampling frequency. This value allows the method to predict what frequency regions at the receiver are vulnerable to aliasing or imaging artifacts. Strategies of oversampling and filtering can be used to eliminate these problem regions while preserving as much of the original content as possible given the desired receiver sampling frequency. This approach creates a situation where the receiver processing can run independently of the source/path processing making it attractive for real-time implementation. This presentation discusses the challenges associated with producing a truly real-time scheme. A three-way tradeoff emerges between an interpolation mechanism that generates decorrelated noise, the computational burden, and the nearness to absolute real-time with which one wants the scheme to run.

Auralization↗

A spatially adaptive spectral re-ordering technique for lossless coding of hyper-spectral images

In this paper, we propose a new approach, applicable to lossless compression of hyper-spectral images, that alleviates some limitations of linear prediction as applied to this problem. According to this approach, an adaptive re-ordering of the spectral components of each pixel is performed prior to prediction and encoding. This re-ordering adaptively exploits, on a pixel-by pixel basis, the presence of inter-band correlations for prediction. Furthermore, the proposed approach takes advantage of spatial correlations, and does not introduce any coding overhead to transmit the order of the spectral bands. This is accomplished by using the assumption that two spatially adjacent pixels are expected to have similar spectral relationships. We thus have a simple technique to exploit spectral and spatial correlations in hyper-spectral data sets, leading to compression performance improvements as compared to our previously reported techniques for lossless compression. We also look at some simple error modeling techniques for further exploiting any structure that remains in the prediction residuals prior to entropy coding.

Memon, Nasir D.↗

Vision Based Localization in Urban Environments

As part of DARPA's MARS2020 program, the Jet Propulsion Laboratory developed a vision-based system for localization in urban environments that requires neither GPS nor active sensors. System hardware consists of a pair of small FireWire cameras and a standard Pentium-based computer. The inputs to the software system consist of: 1) a crude grid-based map describing the positions of buildings, 2) an initial estimate of robot location and 3) the video streams produced by each camera. At each step during the traverse the system: captures new image data, finds image features hypothesized to lie on the outside of a building, computes the range to those features, determines an estimate of the robot's motion since the previous step and combines that data with the map to update a probabilistic representation of the robot's location. This probabilistic representation allows the system to simultaneously represent multiple possible locations, For our testing, we have derived the a priori map manually using non-orthorectified overhead imagery, although this process could be automated. The software system consists of two primary components. The first is the vision system which uses binocular stereo ranging together with a set of heuristics to identify features likely to be part of building exteriors and to compute an estimate of the robot's motion since the previous step. The resulting visual features and the associated range measurements are software component, a particle-filter based localization system. This system uses the map and the then fed to the second primary most recent results from the vision system to update the estimate of the robot's location. This report summarizes the design of both the hardware and software and will include the results of applying the system to the global localization of a robot over an approximately half-kilometer traverse across JPL'S Pasadena campus.

unmanned-ground vehicles↗

2.5D Super-Resolution Approaches for X-Ray Computed Tomography-Based Inspection of Additively Manufactured Parts

X-ray computed tomography (XCT) is a key tool in non-destructive evaluation of additively manufactured (AM) parts, allowing for internal inspection and defect detection. Despite its widespread use, obtaining high-resolution CT scans can be extremely time consuming. This issue can be mitigated by performing scans at lower resolutions; however, reducing the resolution compromises spatial detail, limiting the accuracy of defect detection. Super-resolution algorithms offer a promising solution for overcoming resolution limitations in XCT reconstructions of AM parts, enabling more accurate detection of defects. While 2D super-resolution methods have demonstrated state-of-the-art performance on natural images, they tend to under-perform when directly applied to XCT slices. On the other hand, 3D super-resolution methods are computationally expensive, making them infeasible for large-scale applications. To address these challenges, we propose a 2.5D super-resolution approach tailored for XCT of AM parts. Our method enhances the resolution of individual slices by leveraging multi-slice information from neighboring 2D slices without the significant computational overhead of full 3D methods. Specifically, we use neighboring low-resolution slices to super-resolve the center slice, exploiting inter-slice spatial context while maintaining computational efficiency. This approach bridges the gap between 2D and 3D methods, offering a practical solution for high-throughput defect detection in AM parts.

Sullivan, Haley↗

Technical Performance and Cost Optimization of Unobtrusive Multi-static Serial LiDAR Imager (UMSLI) for Wide-area Surveillance and Identification of Marine Life at Marine Energy Installations

Florida Atlantic University developed an underwater optical monitoring system prototype - Unobtrusive Multi-static Serial LiDAR Imager (UMSLI) - suitable for marine energy full project lifecycle observation (baseline, commissioning, and decommissioning), with an automated real-time classification of marine animals. With precursor 2014 DOE funding (award DE-EE0006787), a prototype UMSLI was demonstrated in a controlled laboratory environment and achieved a TRL 6. This EERE DE-EE0007828 award aimed to both increase the TRL of the UMSLI by improving the technology performance (e.g., increase distance of marine animal target detection capability, add additional species classification capabilities, improve the system performance during the day, etc.) and reduce the cost. The UMSLI presents a novel application of underwater distributed Light Detection And Ranging (LiDAR), an advanced remote sensing method that uses light in the form of laser pulses for an application, this paired with an algorithm provides 360 degrees underwater detection, imaging, and classification of marine life. This solution for underwater monitoring of biota preserves the advantages of traditional optical and acoustic solutions while overcoming many associated disadvantages for marine energy site environmental monitoring, such as difficulties in species detection and classification in low light or night and turbid environments. This new approach is a purposefully designed, reconfigurable adaptation of an existing class 3B laser technology into one UMSLI instrument that can be easily mounted on or around different classes of marine energy equipment, such as devices to capture ocean current energy or wave energy. The system uses relatively low average power, utilizes far-red (> 635nm) laser illumination to be invisible and eye-safe to marine animals, is compact, and cost-effective. The equipment is designed for long-term, maintenance-free operations (i.e., current design is targeting more than 7 days continuous operation), to inherently generate a sparse primary dataset that only includes detected anomalies, and to allow robust real-time automated animal classification and identification with a low data bandwidth requirement. The technology’s overarching goal for application, is a system that can be deployed to collect pre-installation baseline species observations at a proposed marine energy deployment site with minimal post-processing overhead. The envisioned system will also produce high-resolution imagery of marine animals through a wide range of conditions and support automated tracking and notification of the presence of managed animals within established perimeters of marine energy equipment to satisfy deployed marine energy projects’ endangered and threatened species monitoring requirements. Through the current project, we demonstrated the UMSLI prototype in an operational environment and increased the UMSLI’s Technology Readiness Level from 6 to 7. The project resulted in many novel technologies, including: 1) an eye-safe red laser based, low-cost LiDAR system that can detect targets up to 10 meters distance; 2) the GAN-based machine learning underwater LiDAR image enhancement technique (this is the first known application of GAN technique in underwater LiDAR); 3) LiDAR-based real-time automated detection capabilities; and 4) a template matching based automated classification tool. These technologies build a solid foundation for future efforts to develop an extended range electro-optical monitoring system suitable for marine energy deployments. In addition to technology development, a key lesson learned is that addressing regulatory and safety requirements must be front and center in any marine energy monitoring applications.

16 TIDAL AND WAVE POWER↗

MO-FG-CAMPUS-TeP1-05: Rapid and Efficient 3D Dosimetry for End-To-End Patient-Specific QA of Rotational SBRT Deliveries Using a High-Resolution EPID

Purpose: EPID-based patient-specific quality assurance provides verification of the planning setup and delivery process that phantomless QA and log-file based virtual dosimetry methods cannot achieve. We present a method for EPID-based QA utilizing spatially-variant EPID response kernels that allows for direct calculation of the entrance fluence and 3D phantom dose. Methods: An EPID dosimetry system was utilized for 3D dose reconstruction in a cylindrical phantom for the purposes of end-to-end QA. Monte Carlo (MC) methods were used to generate pixel-specific point-spread functions (PSFs) characterizing the spatially non-uniform EPID portal response in the presence of phantom scatter. The spatially-variant PSFs were decomposed into spatially-invariant basis PSFs with the symmetric central-axis kernel as the primary basis kernel and off-axis representing orthogonal perturbations in pixel-space. This compact and accurate characterization enables the use of a modified Richardson-Lucy deconvolution algorithm to directly reconstruct entrance fluence from EPID images without iterative scatter subtraction. High-resolution phantom dose kernels were cogenerated in MC with the PSFs enabling direct recalculation of the resulting phantom dose by rapid forward convolution once the entrance fluence was calculated. A Delta4 QA phantom was used to validate the dose reconstructed in this approach. Results: The spatially-invariant representation of the EPID response accurately reproduced the entrance fluence with >99.5% fidelity with a simultaneous reduction of >60% in computational overhead. 3D dose for 10{sub 6} voxels was reconstructed for the entire phantom geometry. A 3D global gamma analysis demonstrated a >95% pass rate at 3%/3mm. Conclusion: Our approach demonstrates the capabilities of an EPID-based end-to-end QA methodology that is more efficient than traditional EPID dosimetry methods. Displacing the point of measurement external to the QA phantom reduces the necessary complexity of the phantom itself while offering a method that is highly scalable and inherently generalizable to rotational and trajectory based deliveries. This research was partially supported by Varian.

60 APPLIED LIFE SCIENCES↗

3D Printing of Advanced Biocomposites on Earth and Beyond

Human exploration off planet is severely limited by the cost of launching materials into space and re-supply. Thus materials brought from earth must be light, stable and reliable at destination. Using traditional approaches a lunar or Mars base would require either transporting a hefty store of metals or heavy manufacturing equipment and construction materials for in situ extraction; both would severely limit any other mission objectives. Long-term human space presence requires periodic replenishment, adding a massive cost overhead. Even robotic missions often sacrifice science goals for heavy radiation and thermal protection. Biology has the potential to solve these problems because it can replicate and repair itself, and do a wide variety of chemical reactions including making food, fuel and materials. Synthetic biology enhances and expands life's evolved repertoire. Using organisms as feedstock, additive manufacturing could make possible the dream of producing bespoke tools, food, smart fabrics and even replacement organs on demand. Image what new products can be enabled by such a technology, on earth or beyond!

Manufacturing↗

Field Exploration and Life Detection Sampling Through Planetary Analogue Sampling (FELDSPAR).

Exploration missions to Mars rely on rovers to perform analyses over small sampling areas; however, landing sites for these missions are selected based on large-scale, low-resolution remote data. The use of Earth analogue environments to estimate the multi-scale spatial distributions of key signatures of habitability can help ensure mission science goals are met. A main goal of FELDSPAR is to conduct field operations analogous to Mars sample return in its science, operations, and technology from landing site selection, to in-field sampling location selection, remote or stand-off analysis, in situ analysis, and home laboratory analysis. Lava fields and volcanic regions are relevant analogues to Martian landscapes due to desiccation, low nutrient availability, and temperature extremes. Operationally, many Icelandic lava fields are remote enough to require that field expeditions address several sampling constraints that are experienced in robotic exploration, including in situ and sample return missions. The Fimmvruhls lava field was formed by a basaltic effusive eruption associated with the 2010 Eyjafjallajkull eruption. Mlifellssandur is a recently deglaciated plain to the north of the Myrdalsjkull glacier. Holuhraun was formed by a 2014 fissure eruptions just north of the large Vatnajkull glacier. Dyngjusandur is an alluvial plain apparently kept barren by repeated mechanical weathering. Informed by our 2013 expedition, we collected samples in nested triangular grids every decade from the 10 cm scale to the 1 km scale (as permitted by the size of the site). Satellite imagery is available for older sites, and for Mlifellssandur, Holuhraun, and Dyngjusandur we obtained overhead imagery at 1 m to 200 m elevation. PanCam-style photographs were taken in the field by sampling personnel. In-field reflectance spectroscopy was also obtained with an ASD spectrometer in Dyngjusandur. All sites chosen were 'homogeneous' in apparent color, morphology, moisture, grain size, and reflectance spectra at all scales greater than 10 cm. Field lab assays were conducted to monitor microbial habitation, including ATP quantification, qPCR for fungal, bacterial, and archaeal DNA, and direct cell imaging using fluorescence microscopy. Home laboratory analyses include Raman spectroscopy and community sequencing. ATP appeared to be significantly more sensitive to small changes in sampling location than qPCR or fluorescence microscopy. Bacterial and archaeal DNA content were more consistent at the smaller scales, but similarly variable across more distant sites. Conversely, cell counts and fungal DNA content have significant local variation but appear relatively homogeneous over scales of 1 km. ATP, bacterial DNA, and archaeal DNA content were relatively well correlated at many spatial scales. While we have observed spatial variation at various scales and are beginning to observe how that variation fluctuates over time as biodiversity recovers after an eruption, we do not yet fully understand what parameters lead to the observed spatial variation. Home laboratory analyses will help us further understand the elemental and structural composition of the basaltic matrices, but further field analyses are vital for the understanding how temperature, moisture, incident radiation, and so forth influence the habitability of a microclimate.

Field↗

Multi-static Serial LiDAR for Surveillance and Identification of Marine Life at MHK Installations

Final Report for project DE-EE0006787: Multi-static Serial LiDAR for Surveillance and Identification of Marine Life at MHK Installations. This project developed and tested an optical monitoring system prototype that will be suitable for marine and hydrokinetic (MHK) full project lifecycle observation (baseline, commissioning, and decommissioning), with automated real-time classification of marine animals. This system can be deployed to collect pre-installation baseline species observations at a proposed deployment site with minimal post-processing overhead. To satisfy deployed MHK project species of concern (e.g. Endangered Species Act-listed) monitoring requirements, the system provides automated tracking and notification of the presence of managed animals within established perimeters of MHK equipment and provides high resolution imagery of their behavior through a wide range of conditions. During a project’s decommissioning stage, the system can remain installed to provide resource managers with post-installation data. Our technology, known as an Unobtrusive Multi-static Serial LiDAR Imager (UMSLI), is a technology transfer of underwater distributed LiDAR imaging technology that preserves the advantages of traditional optical and acoustic solutions while overcoming associated disadvantages for MHK environmental monitoring applications. This new approach is a purposefully-designed, reconfigurable adaptation of an existing technology that can be easily mounted on or around different classes of MHK equipment. The system uses low average power red (638nm) laser illumination to be invisible and eye-safe to marine animals and is compact and cost effective. The equipment is designed for long term, maintenance-free operations, to inherently generate a sparse primary dataset that only includes detected anomalies (animal presence information), and to allow robust real-time automated animal classification/identification with a low data bandwidth requirement. Advantages of the technology over others currently being used or being considered for MHK monitoring include: Unlike a conventional camera, the depth of field is near-infinite and limited by attenuation (approximately 5-8 m) rather than focal properties of a lens; Operation in an adaptive mode which can project a sparse grid of pulses with higher peak power for longer range detection (>10 meters) and track animals within a zone of interest with high resolution imagery for identification of marine life at closer range (<5m); System detection limit and Signal-to-Noise-Ratio is superior to a camera, due to rejection of both backscattering component and ambient solar background; Multiple wide-angle pulsed laser illuminators and bucket detectors can be flexibly configured to cover a 4pi steradian (i.e. omnidirectional) scene volume, while also retrieving 3D features of animal targets from timing information; Process and classification framework centered around a novel active learning and incremental classification classifier that enables accurate identification of a variety of marine animals automatically; A two-tiered monitoring architecture and invisible watermarking-based data archiving and retrieving approach ensures significant data reduction while preserving high fidelity monitoring. A methodology to train and optimize the classifier for target species of concern to optimize site monitoring effectiveness. This technological innovation addresses a high priority regulatory requirement to observe marine life interaction near MHK projects. Our solution improves resource manager confidence that any interactions between marine animals and equipment are observed in a cost-effective and automated manner. Without EERE funding, this novel application of multi-static LiDAR would not have been available to the MHK community for environmental monitoring.

16 TIDAL AND WAVE POWER↗

Scattering-type scanning near-field optical microscopy with Akiyama piezo-probes

Recent developments of the scattering-type scanning near-field optical microscope at cryogenic temperatures (cryogenic s-SNOM or cryo-SNOM) have led to many breakthroughs in the studies of low energy excitations in quantum materials. However, the simultaneous demands on vibration isolation, low base temperature, precise nano-positioning, and optical access make the construction of a cryo-SNOM a daunting task. Adding to the overhead space required for a cryo-SNOM is the atomic force microscopy control, which predominantly utilizes a laser-based detection scheme for determining the cantilever tapping motion. In this work, we provide an alternative and straightforward route to performing s-SNOM using metal-coated Akiyama probes, where the cantilever tapping motion is detected through a piezoelectric signal. We show that the Akiyama-based cryo-SNOM attains high spatial resolution, good near-field contrast, and is able to perform imaging with a significantly more compact system compared to other cryo-SNOM implementations. Our results firmly establish the potential of s-SNOM based on self-sensing piezo-probes, which can easily accommodate far-infrared wavelengths and high magnetic fields in the future.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Use of residual redundancy in the design of joint source/channel coders

A technique for providing error protection without the additional overhead required for channel coding is developed, starting from the premise that, during source coder design, for the sake of simplicity or due to imperfect knowledge, assumptions have to be made about the source which are often incorrect. This results in residual redundancy at the output of the source coder. The residual redundancy can then be used to provide error protection in much the same way as the insertion of redundancy in convolutional coding provides error protection. To show the validity of this approach, it is applied to image coding using differential pulse code modulation (DPCM). Substantial performance gains are obtained, in terms of both objective and subjective measures.

Sayood, Khalid↗

Parallelization and checkpointing of GPU applications through program transformation

GPUs have emerged as a powerful tool for accelerating general-purpose applications. The availability of programming languages that makes writing general-purpose applications for running on GPUs tractable have consolidated GPUs as an alternative for accelerating general purpose applications. Among the areas that have benefited from GPU acceleration are: signal and image processing, computational fluid dynamics, quantum chemistry, and, in general, the High Performance Computing (HPC) Industry. In order to continue to exploit higher levels of parallelism with GPUs, multi-GPU systems are gaining popularity. In this context, single-GPU applications are parallelized for running in multi-GPU systems. Furthermore, multi-GPU systems help to solve the GPU memory limitation for applications with large application memory footprint. Parallelizing single-GPU applications has been approached by libraries that distribute the workload at runtime, however, they impose execution overhead and are not portable. On the other hand, on traditional CPU systems, parallelization has been approached through application transformation at pre-compile time, which enhances the application to distribute the workload at application level and does not have the issues of library-based approaches. Hence, a parallelization scheme for GPU systems based on application transformation is needed. Like any computing engine of today, reliability is also a concern in GPUs. GPUs are vulnerable to transient and permanent failures. Current checkpoint/restart techniques are not suitable for systems with GPUs. Checkpointing for GPU systems present new and interesting challenges, primarily due to the natural differences imposed by the hardware design, the memory subsystem architecture, the massive number of threads, and the limited amount of synchronization among threads. Therefore, a checkpoint/restart technique suitable for GPU systems is needed. The goal of this work is to exploit higher levels of parallelism and to develop support for application-level fault tolerance in applications using multiple GPUs. Our techniques reduce the burden of enhancing single-GPU applications to support these features. To achieve our goal, this work designs and implements a framework for enhancing a single-GPU OpenCL application through application transformation.

97 MATHEMATICS AND COMPUTING↗