Search NASA⌕ Search

SEARCH · Search NASA

Results for “Incremental Computing”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

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

At least 775 records · Page 43

Constructing Space-Time Views from Fixed Size Statistical Data: Getting the Best of Both Worlds

Many performance monitoring tools are currently available to the super-computing community. The performance data gathered and analyzed by these tools fall under two categories: statistics and event traces. Statistical data is much more compact but lacks the probative power event traces offer. Event traces, on the other hand, can easily fill up the entire file system during execution such that the instrumented execution may have to be terminated half way through. In this paper, we propose an innovative methodology for performance data gathering and representation that offers a middle ground. The user can trade-off tracing overhead, trace data size vs. data quality incrementally. In other words, the user will be able to limit the amount of trace collected and, at the same time, carry out some of the analysis event traces offer using spacetime views for the entire execution. Two basic ideas are employed: the use of averages to replace recording data for each instance and "formulae" to represent sequences associated with communication and control flow. With the help of a few simple examples, we illustrate the use of these techniques in performance tuning and compare the quality of the traces we collected vs. event traces. We found that the trace files thus obtained are, in deed, small, bounded and predictable before program execution and that the quality of the space time views generated from these statistical data are excellent. Furthermore, experimental results showed that the formulae proposed were able to capture 100% of all the sequences associated with 11 of the 15 applications tested. The performance of the formulae can be incrementally improved by allocating more memory at run-time to learn longer sequences.

Schmidt, Melisa↗

Constructing Space-Time Views from Fixed Size Statistical Data: Getting the Best of both Worlds

Many performance monitoring tools are currently available to the super-computing community. The performance data gathered and analyzed by these tools fall under two categories: statistics and event traces. Statistical data is much more compact but lacks the probative power event traces offer. Event traces, on the other hand, can easily fill up the entire file system during execution such that the instrumented execution may have to be terminated half way through. In this paper, we propose an innovative methodology for performance data gathering and representation that offers a middle ground. The user can trade-off tracing overhead, trace data size vs. data quality incrementally. In other words, the user will be able to limit the amount of trace collected and, at the same time, carry out some of the analysis event traces offer using space-time views for the entire execution. Two basic ideas arc employed: the use of averages to replace recording data for each instance and formulae to represent sequences associated with communication and control flow. With the help of a few simple examples, we illustrate the use of these techniques in performance tuning and compare the quality of the traces we collected vs. event traces. We found that the trace files thus obtained are, in deed, small, bounded and predictable before program execution and that the quality of the space time views generated from these statistical data are excellent. Furthermore, experimental results showed that the formulae proposed were able to capture 100% of all the sequences associated with 11 of the 15 applications tested. The performance of the formulae can be incrementally improved by allocating more memory at run-time to learn longer sequences.

Schmidt, Melisa↗

Decreased Lung Perfusion After Breast/Chest Wall Irradiation: Quantitative Results From a Prospective Clinical Trial

Purpose: To quantify lung perfusion changes after breast/chest wall radiation therapy (RT) using pre- and post-RT single photon emission computed tomography/computed tomography (SPECT/CT) attenuation-corrected perfusion scans; and correlate decreased perfusion with adjuvant RT dose for breast cancer in a prospective clinical trial. Methods and Materials: As part of an institutional review board–approved trial studying the impact of RT technique on lung function in node-positive breast cancer, patients received breast/chest wall and regional nodal irradiation including superior internal mammary node RT to 50 to 52.2 Gy with a boost to the tumor bed/mastectomy scar. All patients underwent quantitative SPECT/CT lung perfusion scanning before RT and 1 year after RT. The SPECT/CT scans were co-registered, and the ratio of decreased perfusion after RT relative to the pre-RT perfusion scan was calculated to allow for direct comparison of SPECT/CT perfusion changes with delivered RT dose. The average ratio of decreased perfusion was calculated in 10-Gy dose increments from 0 to 60 Gy. Results: Fifty patients had complete lung SPECT/CT perfusion data available. No patient developed symptoms consistent with pulmonary toxicity. Nearly all patients demonstrated decreased perfusion in the left lung according to voxel-based analyses. The average ratio of lung perfusion deficits increased for each 10-Gy increment in radiation dose to the lung, with the largest changes in regions of lung that received 50 to 60 Gy (ratio 0.72 [95% confidence interval 0.64-0.79], P<.001) compared with the 0- to 10-Gy region. For each increase in 10 Gy to the left lung, the lung perfusion ratio decreased by 0.06 (P<.001). Conclusions: In the assessment of 50 patients with node-positive breast cancer treated with RT in a prospective clinical trial, decreased lung perfusion by SPECT/CT was demonstrated. Our study allowed for quantification of lung perfusion defects in a prospective cohort of breast cancer patients for whom attenuation-corrected SPECT/CT scans could be registered directly to RT treatment fields for precise dose estimates.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Robust fault diagnosis of physical systems in operation

Ideas are presented and demonstrated for improved robustness in diagnostic problem solving of complex physical systems in operation, or operative diagnosis. The first idea is that graceful degradation can be viewed as reasoning at higher levels of abstraction whenever the more detailed levels proved to be incomplete or inadequate. A form of abstraction is defined that applies this view to the problem of diagnosis. In this form of abstraction, named status abstraction, two levels are defined. The lower level of abstraction corresponds to the level of detail at which most current knowledge-based diagnosis systems reason. At the higher level, a graph representation is presented that describes the real-world physical system. An incremental, constructive approach to manipulating this graph representation is demonstrated that supports certain characteristics of operative diagnosis. The suitability of this constructive approach is shown for diagnosing fault propagation behavior over time, and for sometimes diagnosing systems with feedback. A way is shown to represent different semantics in the same type of graph representation to characterize different types of fault propagation behavior. An approach is demonstrated that threats these different behaviors as different fault classes, and the approach moves to other classes when previous classes fail to generate suitable hypotheses. These ideas are implemented in a computer program named Draphys (Diagnostic Reasoning About Physical Systems) and demonstrated for the domain of inflight aircraft subsystems, specifically a propulsion system (containing two turbofan systems and a fuel system) and hydraulic subsystem.

Abbott, Kathy Hamilton↗

Improving Control System Cyber-State Awareness using Known Secure Sensor Measurements

Abstract—This paper presents design and simulation of a low cost and low false alarm rate method for improved cyber-state awareness of critical control systems - the Known Secure Sensor Measurements (KSSM) method. The KSSM concept relies on physical measurements to detect malicious falsification of the control systems state. The KSSM method can be incrementally integrated with already installed control systems for enhanced resilience. This paper reviews the previously developed theoretical KSSM concept and then describes a simulation of the KSSM system. A simulated control system network is integrated with the KSSM components. The effectiveness of detection of various intrusion scenarios is demonstrated on several control system network topologies.

97 MATHEMATICS AND COMPUTING↗

Geometrically nonlinear analysis of layered composite plates and shells

A degenerated three dimensional finite element, based on the incremental total Lagrangian formulation of a three dimensional layered anisotropic medium was developed. Its use in the geometrically nonlinear, static and dynamic, analysis of layered composite plates and shells is demonstrated. A two dimenisonal finite element based on the Sanders shell theory with the von Karman (nonlinear) strains was developed. It is shown that the deflections obtained by the 2D shell element deviate from those obtained by the more accurate 3D element for deep shells. The 3D degenerated element can be used to model general shells that are not necessarily doubly curved. The 3D degenerated element is computationally more demanding than the 2D shell theory element for a given problem. It is found that the 3D element is an efficient element for the analysis of layered composite plates and shells undergoing large displacements and transient motion.

Chao, W. C.↗

Strain Rate Dependent Deformation and Strength Modeling of a Polymer Matrix Composite Utilizing a Micromechanics Approach

Potential gas turbine applications will expose polymer matrix composites to very high strain rate loading conditions, requiring an ability to understand and predict the material behavior under extreme conditions. Specifically, analytical methods designed for these applications must have the capability of properly capturing the strain rate sensitivities and nonlinearities that are present in the material response. The Ramaswamy-Stouffer constitutive equations, originally developed to analyze the viscoplastic deformation of metals, have been modified to simulate the nonlinear deformation response of ductile, crystalline polymers. The constitutive model is characterized and correlated for two representative ductile polymers. Fiberite 977-2 and PEEK, and the computed results correlate well with experimental values. The polymer constitutive equations are implemented in a mechanics of materials based composite micromechanics model to predict the nonlinear, rate dependent deformation response of a composite ply. Uniform stress and uniform strain assumptions are applied to compute the effective stresses of a composite unit cell from the applied strains. The micromechanics equations are successfully verified for two polymer matrix composites. IM7/977-2 and AS4/PEEK. The ultimate strength of a composite ply is predicted with the Hashin failure criteria that were implemented in the composite micromechanics model. The failure stresses of the two composite material systems are accurately predicted for a variety of fiber orientations and strain rates. The composite deformation model is implemented in LS-DYNA, a commercially available transient dynamic explicit finite element code. The matrix constitutive equations are converted into an incremental form, and the model is implemented into LS-DYNA through the use of a user defined material subroutine. The deformation response of a bulk polymer and a polymer matrix composite are predicted by finite element analyses. The results compare reasonably well to experimental values, with some discrepancies. The discrepancies are at least partially caused by the method used to integrate the rate equations in the polymer constitutive model.

Goldberg, Robert K.↗

Version 4 of the SMAP Level-4 Soil Moisture Algorithm and Data Product

The NASA Soil Moisture Active Passive (SMAP) mission Level-4 Soil Moisture (L4_SM) product provides global, 3-hourly, 9-km resolution estimates of surface (0-5 cm) and root-zone (0-100 cm) soil moisture with a mean latency of ~2.5 days. The underlying L4_SM algorithm assimilates SMAP radiometer brightness temperature (Tb) observations into the NASA Catchment land surface model using a spatially-distributed ensemble Kalman filter. Version 4 of the L4_SM modeling system includes a reduction in the upward recharge of surface soil moisture from below under non-equilibrium conditions, resulting in reduced bias and improved dynamic range of L4_SM surface soil moisture compared to earlier versions. This change and additional technical modifications to the system reduce the mean and standard deviation of the observation-minus-forecast Tb residuals and overall soil moisture analysis increments while maintaining the skill of the L4_SM soil moisture estimates versus independent in situ measurements; the average, bias-adjusted RMSE in Version 4 is 0.039 m(exp 3) m(exp -3) for surface and 0.026 m(exp 3) m(exp -3) for root-zone soil moisture. Moreover, the coverage of assimilated SMAP observations in Version 4 is near-global owing to the use of additional satellite Tb records for algorithm calibration. L4_SM soil moisture uncertainty estimates are biased low (by 0.01-0.02 m(exp 3) m(exp -3)) against actual errors (computed versus in situ measurements). L4_SM runoff estimates, an additional product of the L4_SM algorithm, are biased low (by 35 mm year (exp -1)) against streamflow measurements. Compared to Version 3, bias in Version 4 is reduced by 46% for surface soil moisture uncertainty estimates and by 33% for runoff estimates.

RMSE↗

Viper Science Operations: Science Traverse Planning Perspectives, Processes, and Tools.

Introduction: The NASA Volatiles Investigating Polar Exploration Rover (VIPER) launchesin late 2024 towards a landing site on Mons Mouton in the Nobile region of the Lunar South Pole (LSP). As described in other LPSC presentations [1,2,3], the primary objective of the VIPER mission is to study the composition and distribution of hydrogen-bearing and other volatiles by way of a complementary suite of payloads that the rover will carry to the Moon: three “prospecting” instruments which operate continuously while roving - the Neutron Spectrometer System (NSS), the Near InfraRed Volatiles Spectrometer System (NIRVSS), and the Mass Spectrometer observing lunar operations (MSolo). A 1-meter auguring/percussive drill called the The Regolith and Ice Drill for Exploration of New Terrains (TRIDENT) is used to bring subsurface cuttings to the surface in 10-cm increments where they are interrogated by NIRVSS and MSolo. The Visible Imaging System is comprised of eight cameras (the NavCam stereo pair mounted on the mast gimbal, the AftCam stereo pair mounted on the aft panel, and four HazCams mounted in the wheel wells) that capture grayscale visible wavelength images of the lunar environment and the rover’s upper deck. The VIPER mobility system is a four-wheel design that includes the following motorized modules: Suspension, Steering, and Drive/Propulsion. Wheel odometry and applied torque — combined with position data from rover imagery — can be used to compute assessments of ‘slip and sinkage’. Slope estimates can be computed from stereo rover imagery, as well as pitch and roll data from the IMU (Inertial Measurement Unit) and Star Tracker telemetry.

Darlene Sze Shien Lim↗

Advanced Computational Modeling of High-Level Waste Vitrification at the Hanford Site

The U.S. Department of Energy (DOE) has selected vitrification for stabilizing legacy tank waste at the Hanford site, where radioactive waste from plutonium production was historically stored in underground tanks. This waste will be separated into low-activity waste (LAW) and high-level waste (HLW) fractions and processed at the Waste Treatment and Immobilization Plant (WTP). At WTP, glass melters are used for the vitrification of radioactive tank waste, transforming it into a stable borosilicate glass form for safe long-term storage. The melter vessel is constructed from highly durable and heat-resistant materials, where the vitrification process occurs. The main regions that are modeled are the melt pool, plenum, cold cap, riser/discharge chamber, and surrounding structure with insulation layers. Forced convection induced by air bubblers at the base of the melter ensure uniform temperature distribution and provide heat to the cold cap layer. The cold cap is a region of reacting batch feed that floats on top of the molten glass and is where the batch-to-glass reactions occur. Joule heating provided by electrodes mounted along the vertical walls of the melter and immersed directly in the glass, generates the necessary heat for the net endothermic conversion processes that occur in the cold cap. The high temperatures, radioactivity, and opaque nature of the glass prevent direct observation inside the melters. Therefore, computational models are essential for providing insight into factors that affect melter throughput. Thermocouples in the plenum provide operators with plenum temperature measurements. Operational adjustments include bubbling rate, voltage supplied to the electrodes, feed adjustments, and glass removal rate. Different computational fluid dynamics (CFD) models have been developed, each serving a specific purpose. There are CFD models of different scale melters, as well as models that capture the two-phase flow interfaces of rising bubbles in the molten glass or models with a simplified molten glass region so that the surrounding structure and plenum can be feasibly incorporated. Pilot-scale melter models have been developed to serve as validation of the methods employed in the simulation of the full-scale WTP melters. Models incorporating resolved bubbling are used to develop momentum source terms to implement into a single phase, multi-region, steady-state flow model that is being validated by measured process parameters such as glass production rate, voltage, input power, plenum temperatures, etc. The resolved bubbling model uses the multiphase volume of fluid approach to model the system with a high-resolution interface capturing scheme to maintain sharp interfaces between the molten glass and the air phase. The suite of CFD models is continually being improved to incorporate more realistic physics and achieve faster turnaround time. For example, an incremental controller is implemented to automatically adjust electrode voltage within the simulation to a molten glass set point temperature of 1150°C. Newer models feature improved meshes to ensure conformal meshes between regions and eliminate unnecessary mesh refinement in areas that are not of interest (such as boundary layers in offgas ports). Instead of explicitly modeling the structural, refractory, and insulation layers of the melter, a thermal resistance approach is used with published correlations used for boundary conditions. The development of robust and efficient CFD models will be instrumental in enabling the WTP to successfully fulfill its mission of safely stabilizing legacy nuclear waste.

12 - MGMT OF RADIOACTIVE AND NON-RADIOACTIVE WASTE↗

An Incremental Tensor Train Decomposition Algorithm

We present a new algorithm for incrementally updating the tensor train decomposition of a stream of tensor data. This new algorithm, called the tensor train incremental core expansion (TT-ICE) improves upon the current state-of-the-art algorithms for compressing in tensor train format by developing a new adaptive approach that incurs significantly slower rank growth and guarantees compression accuracy. This capability is achieved by limiting the number of new vectors appended to the TT-cores of an existing accumulation tensor after each data increment. These vectors represent directions orthogonal to the span of existing cores and are limited to those needed to represent a newly arrived tensor to a target accuracy. We provide two versions of the algorithm: TT-ICE and TT-ICE accelerated with heuristics (TT-ICE*). Here, we provide a proof of correctness for TT-ICE and empirically demonstrate the performance of the algorithms in compressing large-scale video and scientific simulation datasets. Compared to existing approaches that also use rank adaptation, TT-ICE* achieves 57× higher compression and up to 95% reduction in computational time.

97 MATHEMATICS AND COMPUTING↗

IWA : an analysis program for isentropic wave measurements.

IWA (Isentropic Wave Analysis) is a program for analyzing velocity profiles of isentropic compression experiments. IWA applies incremental impedance matching correction to measured velocity profiles to obtain in-situ particle velocity profiles for Lagrangian wave analysis. From the in-situ velocity profiles, material properties such as wave velocities, stress, strain, strain rate, and strength are calculated. The program can be run in any current version of MATLAB (2008a or later) or as a Windows XP executable.

97 MATHEMATICS AND COMPUTING↗

Computer acquired performance data from a chemically vapor-deposited-rhenium, niobium planar diode

Performance data from a chemically vapor-deposited-rhenium, niobium thermionic converter are presented. The planar converter has a guard-ringed collector and a nominal fixed spacing of 0.25 mm (10 mils). The data were obtained by using a computerized acquisition system and are available on request to one of the authors on microfiche as individual and composite parametric current, voltage curves. The parameters are the temperatures of the emitter T sub E collector T sub C, and cesium reservoir T sub R. The composite plots have constant T sub E and varying T sub C or T sub R, or both. Current, voltage envelopes having constant T sub E with and without fixed T sub C appear in the present report. The diode was tested at increments between 1600 and 2000 K for the emitter Hohlraum, 800 to 1100 K for the collector, and 540 and 650 K for the reservoir. A total of 312 current, voltage curves were obtained in the present performance evaluation. Current, voltage envelopes from three rhenium emitter converters evaluated in the present program are also given. The data are compared at commom emitter Hohlraum temperatures.

Manista, E. J.↗

Prediction of effects of wing contour modifications on low-speed maximum lift and transonic performance for the EA-6B aircraft

Computational predictions of the effects of wing contour modifications on maximum lift and transonic performance were made and verified against low speed and transonic wind tunnel data. This effort was part of a program to improve the maneuvering capability of the EA-6B electronics countermeasures aircraft, which evolved from the A-6 attack aircraft. The predictions were based on results from three computer codes which all include viscous effects: MCARF, a 2-D subsonic panel code; TAWFIVE, a transonic full potential code; and WBPPW, a transonic small disturbance potential flow code. The modifications were previously designed with the aid of these and other codes. The wing modifications consists of contour changes to the leading edge slats and trailing edge flaps and were designed for increased maximum lift with minimum effect on transonic performance. The prediction of the effects of the modifications are presented, with emphasis on verification through comparisons with wind tunnel data from the National Transonic Facility. Attention is focused on increments in low speed maximum lift and increments in transonic lift, pitching moment, and drag resulting from the contour modifications.

Allison, Dennis O.↗

Global Failure Modes in Composite Structures for High Altitudes

This report summarizes the accomplishments under the referenced grant. The work described was started under the guidance and supervision of the late Dr. James Stames as the technical contact. It was aimed at investigating the development of analysis tools to deal with the problem of rupture in reinforced structural skin of future composites-based aircraft. It was of particular interest to assess methods by which failure features reminiscent of cracks in metallic structures would develop and propagate in fiber reinforced structures in interaction with the reinforcing frame. To eventually achieve that goal it was necessary to first understand the stress or strain distribution at the front of such features so that interactions between such features and reinforcing agents could be assessed computationally. Thus the major emphasis here was on the assessment of damage front and methods on how to assess or characterize it. During the conduct of this research program Dr. Stames changed to a different NASA- internal assignment, which divorced him of the direct supervision of this grant. A student who was approximately % into the completion of his Ph.D. research needed to finish this work, and NASA funds were made available under Dr. Damodar Ambur, the successor Branch Manager for Dr. James Starnes, for the completion of this work. The current grant was the thus a new and fmal support increment for completion of the started research. Final reports for previous funding have been completed and submitted. Because of the interconnection of this last phase of the investigation with previous work it is deemed useful to make the Ph.D. thesis by Luis Gonzales the body of this report.

Knauss, W. G.↗

SU-F-T-662: Feasibility Study of Fe3O4/TaOx Nano Particles as a Radiosensitizer for Radiation Therapy

Purpose: To investigate the feasibility of using multifunctional Fe{sub 3}O{sub 4}/TaOx(core / shell) nano particles developed for CT and MRI contrast agent as dose enhancing radiosensitizers. Methods: Firstly, to verify the imaging detectability of Fe{sub 3}O{sub 4}/TaOx nano particles, in-vivo tests were conducted. Approximately 600 mg/kg of 19 nm diameter Fe{sub 3}O{sub 4}/TaOx nano particles dispersed in phosphate buffered saline(PBS) were injected to ten nude Balb/c mice through the tail vein. Difference between pre- and post-injection images was analyzed by computing the pixel histogram and correlation coefficient factor using MATLAB in the user defined ROI. Secondly, to quantify the potential therapeutic enhancement with nano materials, DER (Dose Enhancement Ratio) and number of SER (Secondary Electron Ratio) were computed using TOPAS(ver.2.0 P-03) MC simulation. Results: In CT, MRI imaging, the aorta, the blood vessel, and the liver were clearly visualized after intravenous injection of Fe{sub 3}O{sub 4}/TaOx nano particles. There was large different between pre and post-injection images of Histogram data and Coefficients of correlation factor in CT and MR are 0.006, 0.060, respectively. When 70 MeV protons were irradiated for a Gold, Tantalum, TaOx, Fe{sub 3}O{sub 4}/TaOx, Fe{sub 3}O{sub 4} nano particle, DER was 2.59, 2.41, 1.68, 1.54 and 1.36 respectively. Similarly, SER increment was 2.31, 2.15, 1.56, 1.46, and 1.27 for Gold, Tantalum, TaOx, Fe{sub 3}O{sub 4}/TaOx, Fe{sub 3}O{sub 4} nano particle, respectively. Conclusion: Fe{sub 3}O{sub 4}/TaOx nano particles have potential as a multifunctional agent which enhances the accuracy in cancer detection through visualization of developed tumor lesion and increases the therapeutic effect in proton therapy. The dose enhancement with Fe{sub 3}O{sub 4}/TaOx was estimated as half of the Gold. However, tumor targeting such as combined with magnetic field may overcome the low DER. This research was supported by the NRF funded by the Ministry of Science, ICT & Future Planning (2012M3A9B6055201 and 2012R1A1A2042414), Samsung Medical Center grant[GFO1130081].

60 APPLIED LIFE SCIENCES↗

Primal interface debonding formulation for finite strain isotropic plasticity

In this work, a framework is developed for modeling ductile damage of nonlinear materials whose plastic deformation is characterized using rate independent classical plasticity. This method relies on the assumption that the free energy can be decomposed into elastic, plastic and damage parts. A thermodynamically consistent method is derived which satisfies the second law of thermodynamics in the Clausius–Duhem inequality form. The dissipation associated with plasticity takes place in the domain only, while damage dissipation is localized to the interface. The method is developed using Variational Multiscale ideas to obtain definitions of the interface fluxes within a primal formulation analogous to the Discontinuous Galerkin method, which ensures weakly vanishing interface gap prior to reaching a damage initiation criterion. The local nonlinear problem to calculate both plastic deformation gradient and damage variable follows an incremental approach similar to classical plasticity return mapping algorithm. This elastoplastic damage formulation is developed for material undergoing finite strain, and it naturally accommodates a trapezoidal traction separation law (TSL) whose shape can be varied to model either ductile interface behavior or brittle interface behavior. The formulation's performance is assessed through modeling a patch test and a compact tension specimen.

42 ENGINEERING↗

DARMA-MPI Interoperability WBS 2.3.1.04 Milestone 15 (ECP Milestone Report)

DARMA (Distributed Asynchronous Resilient Models for Applications) is a runtime library developed as part of the the Sandia ATDM (Advanced Technology Development and Mitigation) program. DARMA supports applications within 2.2.5.03 ADNN03-ASC ATDM SNL Application, which includes a number of applications featuring dynamic physics which requires load balancing and asynchronous communication for high performance. We have implemented a modern C++ programming model that can enable dynamic, asynchronous communication on top of existing data structures from a serial or MPI code. DARMA development has occurred in parallel with a verification milestone for ATDM in FY18. For FY19, DARMA will impact ATDM by enabling the performance benefits of a dynamic runtime through only incremental changes to existing verified MPI codes. The results presented here demonstrate the DARMA development process for an MPI mini-app, showing a 3-4x improvement in performance for a challenging problem relative to the parent MPI code and coming within 25 percent of the theoretically optimal performance achievable from a perfect, fine-grained load balancer for most cases.

97 MATHEMATICS AND COMPUTING↗