Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Performance evaluation”

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
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Overview of lithium injection and flowing liquid lithium results from the US–China collaboration on EAST

A multi-institutional team consisting of national labs and universities has been conducting research to understand and control the plasma-material interface (PMI) to improve long pulse discharge control and performance in the EAST device in Hefei, China. One particular focus of the group has been to evaluate the performance of different plasma-facing component (PFC) technologies and materials, particularly lithium. Flowing liquid lithium PFC technologies are currently into their third generation of development and deployment in EAST, demonstrating recycling control and mitigation of edge-localized modes (ELMs). A lithium powder dropper has also been used to control ELMs while a lithium granule injector has demonstrated ELM triggering and pacing. To support this work, the facilities at the Center for Plasma Material Interactions such as HIDRA are being commissioned and utilized to develop/test the technologies before deployment on EAST. This paper will present a summary of the collaborative lithium PMI program on EAST.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2021

MeV-scale performance of water-based and pure liquid scintillator detectors

This paper presents studies of the performance of water-based liquid scintillators (WbLS) in both 1-kt and 50-kt detectors. In this work, performance is evaluated in comparison to both pure water Cherenkov detectors and a nominal model for pure scintillator detectors. Performance metrics include energy, vertex, and angular resolution, along with a metric for the ability to separate the Cherenkov from the scintillation signal as being representative of various particle identification capabilities that depend on the Cherenkov/scintillation ratio. We also modify the time profile of scintillation light to study the same performance metrics as a function of rise and decay time. We go on to interpret these results in terms of their impact on certain physics goals, such as solar neutrinos and the search for Majorana neutrinos. This work supports and validates previous results, and the assumptions made therein, and serves as a significant stepping stone to complete detector design studies by using a more detailed detector model and full reconstruction, with a primarily data-driven optical model, and fewer model assumptions. With this model, a high-coverage 50-kt detector achieves better than 10 (1)% precision on the flux of neutrinos from the Carbon-Nitrogen-Oxygen cycle with a water-based liquid scintillator (pure LS) target in five years of data taking. A 1-kt LS detector, with a conservative 50% fiducial volume of 0.5 kt, can achieve better than 5% detection. A liquid scintillator detector has sensitivity into the normal hierarchy region for Majorana neutrinos with half-life sensitivity of $T^{0νββ}_{1/2}$ > 1.4 × 10 28 years at 90% C.L. for 10 years of data taking with a Te-loaded target.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2021

Microwave Plasma Flash Pyrolysis for Coal-to-Product Applications

Since 2010, the US coal industry has experienced a precipitous decline, with coal production volumes dropping by as much as 35% according to EIA. This decline is hardly surprising given that 90% of domestic coal use is allocated to electricity production: While coal generation supplied about 45% of the nation's power in 2010, its share is projected to fall to only 22% in 2020. Fortunately, alternative uses for coal as a feedstock rather than as energy source are available. Coal tar is an established feedstock for chemicals and high-value carbon products including carbon fiber and battery-grade graphites. At present, growth of these uses is constrained by the low availability, purity, and yields of currently available coal liquids, and high energy- and CO2-intensity of current methods of production (i.e. coking). Previously reported work conducted in a partnership with Pacific Northwest National Laboratory demonstrated technical and economic viability of flash pyrolysis of coal in a microwave plasma environment at bench scales targeting production of fuel feedstocks (synthetic oil). H Quest’s work since then indicated promise of microwave-derived coal liquids to be uniquely suited for the lower-cost production of carbon materials, including synthetic graphite, carbon fiber, and composites. Low bulk process temperatures, subsecond residence times, and ambient pressure operation enable small-scale deployment, reduce capital and operating costs, and suppress formation of impurities, including QI material, typical in conventional coal tars. With no inherent production of CO2 or water consumption, microwave plasma flash pyrolysis presents a viable route to the sustainable production of both platform chemicals and carbon-based materials. Most recently, H Quest developed a novel, ruggedized microwave plasma reactor, which departs from the proof-of-concept approach and addresses the problems typically encountered in scaling laboratory microwave plasma apparata (e.g. eliminating fragile quartz enclosures prone to coking and fouling). This reactor has been integrated with a 15kW microwave power supply, a pulverized coal injection system, and a product recovery system. A series of screening tests across a range of coals (including hv-, lv-bituminous and subbituminous PRB), entrainment gas flows and compositions and microwave energy inputs are performed to evaluate conversion rates within the novel, scaled-up system, confirm yields of liquid products, and to perform compositional (GC/MS and elemental) analyses of the products. Results of these tests will be presented and discussed.

Skoptsov, George↗
DOE OSTI2023

Process-based cost estimation framework for assessing economic viability of environmentally and socially sustainable rare earth element feedstocks

Rare Earth Elements (REEs) are critical raw material inputs that are required to enable the energy transition. Constraints on the U.S. supply of conventional REE feedstocks and the desire to diversify U.S. reliance on foreign supply chains have motivated the pursuit of REE recovery from unconventional feedstocks including waste products. Recovery of REEs from unconventional sources is also motivated by environmental and social considerations, including the potential to offset new mining activity and subsidize the remediation of REE-containing wastes from coal, natural gas, and other mineral extraction activities. Sourcing REEs from unconventional feedstocks would reduce supply chain constraints and minimize the environmental impacts of REE extraction, but current processes to exploit these feedstocks have uncertain technical and economic performance and the processes have not been demonstrated at scale. This work aims to reduce uncertainty in evaluating the performance of new REE production pathways. A database of operating and capital expenses for REE recovery and establish consistent process unit costs for common stages in the conventional supply chain are developed. Market prices are used to develop well-defined methods for determining product values for a diverse set of products that can be generated (e.g. oxalates, alloys, end-use products such as magnets) but are not typically considered in technoeconomic analyses. Finally, to provide insight on the effects of economies of scale and feedstock purity on capital and operating costs, process unit level cost functions relating total costs per kilogram of REE produced are reported. These tools can be applied to assess and facilitate scale-up of REE production from diverse sources of REEs.

Fritz, Alison↗
DOE OSTI2022

Testing fast reactor fuels in a thermal reactor: Comparison of transmutation metallic fuel alloys behavior by scanning electron microscopy

To optimize nuclear waste repository performance, the destruction of minor actinide elements, particularly Np and Am, in a neutron fast spectrum reactor is possible by incorporating these elements into nuclear fuel. Evaluating the performance of minor actinide containing fuel is of paramount importance to enabling this technology. However, such a task is challenging without an available domestic fast spectrum test reactor. A comparison of fuel performance tested in an available domestic thermal reactor at the Idaho National Laboratory, the Advanced Test Reactor, and in a fast spectrum reactor in France (Phénix) is presented here in this study. This study evaluates the capability of using a cadmium shrouded test position to mimic the power profile along the fuel radius present in fast spectrum reactors so that thermally driven phenomenon (e.g., constituent redistribution) can be evaluated in a thermal reactor and determined to be prototypical of a fast reactor. Thus, optical microscopy and scanning electron microscopy has been performed on irradiated 35U-29Pu-4Am-2Np-30Zr fuel samples (where the number preceding the element is the weight percent concentration) from the two mentioned reactors that present similar irradiation temperatures and power conditions. The results indicate that fuel performance phenomena are reproducible in the two irradiation conditions. The redistribution of Zr occurred in the same manner for the two samples. Similar partitioning of U-Pu-Zr phases was observed, and the behavior of Am was similar in the analyzed specimens. Finally, the overall microstructure evolution seems not to be affected by minor actinides addition compared to expected behavior of conventional U-19Pu-10Zr ternary metal fuels for both specimens. Slight differences in fuel cladding chemical interaction were, however, observed. This difference is likely driven by difference in cladding composition rather than irradiation conditions.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2020

Interfacing HDF5 with a scalable object‐centric storage system on hierarchical storage

Summary Object storage technologies that take advantage of multitier storage on HPC systems are emerging. However, to use these technologies at present, applications have to be modified significantly from current I/O libraries. HDF5, a widely used I/O middleware on HPC systems, provides a virtual object layer (VOL) that allows applications to connect to different storage mechanisms transparently without requiring significant code modifications. We recently designed the proactive data containers (PDC) object‐centric storage system that provides the capabilities of transparent, asynchronous, and autonomous data movement taking advantage of multiple storage tiers—a decision that has so far been left upon the user on most current systems. To enable PDC's features through HDF5 without modifying application codes, we have developed an HDF5 VOL connector that interfaces with PDC. We present in this article the connector interface and evaluate its performance on Cori, a Cray XC40 supercomputer located at the National Energy Research Scientific Computing Center (NERSC). Our evaluation demonstrates up to an 8× improvement compared with HDF5 that has the most recent optimizations.

Mu, Jingqing↗
DOE OSTI2022

DFSynthesizer: Dataflow-based Synthesis of Spiking Neural Networks to Neuromorphic Hardware

Spiking Neural Networks (SNNs) are an emerging computation model that uses event-driven activation and bio-inspired learning algorithms. SNN-based machine learning programs are typically executed on tile-based neuromorphic hardware platforms, where each tile consists of a computation unit called a crossbar, which maps neurons and synapses of the program. However, synthesizing such programs on an off-the-shelf neuromorphic hardware is challenging. This is because of the inherent resource and latency limitations of the hardware, which impact both model performance, e.g., accuracy, and hardware performance, e.g., throughput. We propose DFSynthesizer, an end-to-end framework for synthesizing SNN-based machine learning programs to neuromorphic hardware. The proposed framework works in four steps. First, it analyzes a machine learning program and generates SNN workload using representative data. Second, it partitions the SNN workload and generates clusters that fit on crossbars of the target neuromorphic hardware. Third, it exploits the rich semantics of the Synchronous Dataflow Graph (SDFG) to represent a clustered SNN program, allowing for performance analysis in terms of key hardware constraints such as number of crossbars, dimension of each crossbar, buffer space on tiles, and tile communication bandwidth. Finally, it uses a novel scheduling algorithm to execute clusters on crossbars of the hardware, guaranteeing hardware performance. We evaluate DFSynthesizer with 10 commonly used machine learning programs. Our results demonstrate that DFSynthesizer provides a much tighter performance guarantee compared to current mapping approaches.

Computer Science↗
DOE OSTI2021

Assembling a Cyber Range to Evaluate Artificial Intelligence / Machine Learning (AI/ML) Security Tools

In this case study, we will describe the design and assembly of a cyber security test range we have built at Oak Ridge National Laboratory in Oak Ridge, TN, USA. The range is designed to provide a flexible environment to evaluate cyber security tools—particularly those involving AI/ML—in a way that provides realistic environments and where we can control the experiments to determine the strengths and weaknesses of the tools. We have designed in the ability to repeat the evaluations, so additional tools can be evaluated and compared at a later time. The system is one that can be scaled up or down for experiment sizes. At the time of the conference we will have completed two full-scale, national, government challenges on this range. These challenges are evaluating the performance and operating costs for AI/ML-based cyber security tools for application into large, government-sized environments. These evaluations will be described, in order to provide motivation and context for various design decisions and adaptations we have made. The first challenge measured end-point security tools against 100K malware samples chosen across a range of types. The second is network detection of attempted penetration and exploitations with varying levels of covertness in a high-volume, business network. The scale of each of these challenges is requiring us to create automation systems to repeat the experiments identically for each tool. Preventing there being easy signs of malicious activity for the AI/ML tools to focus on has been a particularly interesting and challenging aspect of designing and executing these challenge events. After the events, the range continues to be used for other research such as adversarial machine learning where the repeatability, scale, and automation required for the national challenge events become essential elements for research.

Nichols, Jeff↗
DOE OSTI2021

An evaluation of tri-valent oxide (Cr 2 O 3 ) as a grain enlarging dopant for UO 2 nuclear fuels fabricated under reducing environment

A study was performed to evaluate the microstructure and crystallography of nominally 500–2000 Cr 2 O 3 -doped UO 2 fabricated in a temperature range of 1150–1750°C under reducing experimental conditions. We observed an increase in grain size of the samples with the increase in heat treating temperature as expected. For a given sintering temperature (1700–1750°C), an increase in the grain size was also observed with the increase in Cr 2 O 3 concentration up to a value of ~1000–1200 wppm. A decrease in fission gas release as a function of grain size was estimated for the Cr 2 O 3 -doped UO 2 samples assuming specified post-irradiation annealing conditions. A nearly linear decrease was obtained in the lattice parameter of the Cr 2 O 3 -doped UO 2 fcc phase with the increase in Cr 2 O 3 concentration, especially up to a nominal value of 1000 wppm. The lattice parameter decrease was also persistent with the increase in the average grain size as a result of addition of Cr 2 O 3 into the UO 2 lattice. An increase in the crystallite size and a decrease in the microstrain of the $fcc$ phase were observed with the increase in the average grain size of the samples, indicating a higher crystallinity of the Cr 2 O 3 -doped samples than that of the undoped UO 2 sample.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2025

Differentiable hybrid neural network approach for enhancing reactor dynamics simulations

Reactor dynamics simulations provide essential insights into the time-dependent behavior of nuclear reactors under various operating conditions. However, high-fidelity simulations can be computationally intensive, requiring significant computational resources. Here, to address this challenge, this study employs a differentiable hybrid model that utilizes neural networks as a corrector to enhance the performance of a low-fidelity simulation, aligning its predictions with those of a high-fidelity simulation. Low-fidelity and high-fidelity simulations were obtained by adjusting the mesh size in the System Dynamics Analysis Tool. The differentiable hybrid model was trained in two approaches: time-step-wise and sequence-wise. It was then applied to simulate various transients in a molten salt reactor. Its performance was evaluated by comparing its responses to transients against those of the high-fidelity simulation. An additional approach was performed using a data-driven model to correct the low-fidelity simulation. In comparison, the differentiable hybrid model showed significant improvements in transient prediction, effectively addressing the limitations of the low-fidelity simulations. The results highlighted the robustness of the differentiable hybrid model in both training approaches. It delivered simulations that were at least 3.8 times faster than high-fidelity models. In the time-step-wise approach, it achieved at least a 39% improvement in accuracy. In the sequence-wise approach, it showed at least an 81% accuracy improvement over the full transient. This approach offers a promising path for improving computational efficiency without compromising accuracy in nuclear reactor simulations, making it suitable for real-time digital twin applications.

42 - ENGINEERING↗
DOE OSTI2023

Inferring adversarial behaviour in cyber‐physical power systems using a Bayesian attack graph approach

Abstract Highly connected smart power systems are subject to increasing vulnerabilities and adversarial threats. Defenders need to proactively identify and defend new high‐risk access paths of cyber intruders that target grid resilience. However, cyber‐physical risk analysis and defense in power systems often requires making assumptions on adversary behaviour, and these assumptions can be wrong. Thus, this work examines the problem of inferring adversary behaviour in power systems to improve risk‐based defense and detection. To achieve this, a Bayesian approach for inference of the Cyber‐Adversarial Power System (Bayes‐CAPS) is proposed that uses Bayesian networks (BNs) to define and solve the inference problem of adversarial movement in the grid infrastructure towards targets of physical impact. Specifically, BNs are used to compute conditional probabilities to queries, such as the probability of observing an event given a set of alerts. Bayes‐CAPS builds initial Bayesian attack graphs for realistic power system cyber‐physical models. These models are adaptable using collected data from the system under study. Then, Bayes‐CAPS computes the posterior probabilities of the occurrence of a security breach event in power systems. Experiments are conducted that evaluate algorithms based on time complexity, accuracy and impact of evidence for different scales and densities of network. The performance is evaluated and compared for five realistic cyber‐physical power system models of increasing size and complexities ranging from 8 to 300 substations based on computation and accuracy impacts.

Sahu, Abhijeet↗
DOE OSTI2021

A Multi-Site Networked Hardware-in-Loop Platform for Evaluation of Interoperability and Distributed Intelligence at Grid-Edge

Electric power systems have experienced large increases in the number of intelligent, connected and controllable devices being deployed, leading to a high degree of distributed intelligence at the grid-edge. These devices, both utility-owned and consumer-owned, include but are not limited to: renewable generation sources, energy storage, remote switches, voltage regulators, and smart controllable loads such as electric vehicles. These new devices provide significant potential for increased operational flexibility that can be leveraged to achieve system reconfiguration, resiliency improvements, power quality improvements, and distribution system automation. However, there are two significant challenges that must be addressed before these assets can be leveraged for operations: interoperability and system level validation prior to deployment. Because of the complexity of distributed control systems, and their interactions with legacy centralized controls, a purely simulations-based approach for pre-deployment validation is not sufficient. It requires hardware-in-loop testing to emulate the operational hardware devices and evaluate their performance. Additionally, securely integrating multiple test facilities at utility operators and vendors might enable rapid scale-up of evaluation platforms, and remove the need for multiple expensive standalone installations. Presented in this paper, is the development of a multi-site evaluation platform that employs Advanced Distribution Management Systems (ADMS), distributed control devices, real-time hardware-in-loop assets, secure communication links, and protocol adapters. This platform uses standards-based approaches and open-source tools, and hence can serve as a template for other researchers and institutions to implement their multi-site evaluation frameworks for pre-deployment testing.

Essakiappan, Somasundaram↗
DOE OSTI2023

Combining Sparse Approximate Factorizations with Mixed-precision Iterative Refinement

The standard LU factorization-based solution process for linear systems can be enhanced in speed or accuracy by employing mixed-precision iterative refinement. Most recent work has focused on dense systems. We investigate the potential of mixed-precision iterative refinement to enhance methods for sparse systems based on approximate sparse factorizations. In doing so, we first develop a new error analysis for LU- and GMRES-based iterative refinement under a general model of LU factorization that accounts for the approximation methods typically used by modern sparse solvers, such as low-rank approximations or relaxed pivoting strategies. We then provide a detailed performance analysis of both the execution time and memory consumption of different algorithms, based on a selected set of iterative refinement variants and approximate sparse factorizations. Our performance study uses the multifrontal solver MUMPS, which can exploit block low-rank factorization and static pivoting. We evaluate the performance of the algorithms on large, sparse problems coming from a variety of real-life and industrial applications showing that mixed-precision iterative refinement combined with approximate sparse factorization can lead to considerable reductions of both the time and memory consumption.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Ventilator Testbed

A number of open-source ventilator designs have been published online since the outbreak of the COVID-19 pandemic. These ventilator designs have been created and released to help address the limitations in manufacturing and distribution timelines, and to help drive availability of more cost efficient ventilators. Many of these ventilators have been designed to be relatively simple to build and to use components that can be easily-sourced and acquired in large numbers from local suppliers. A test bed for evaluating the performance of these open-source ventilators was developed at Los Alamos National Laboratory as part of the Technology Evaluation and Demonstration program.

42 ENGINEERING↗
DOE OSTI2023

Quantification of LEU Holdup using gamma ray imaging and inverse transport solver

Holdup is the residual amount of special nuclear material (SNM) remaining in a processing facility after the bulk materials have been cleaned out. In commercial uranium processing facilities, quantification of holdup is a major challenge because of the highly variable shapes and sizes of the deposits. Any method that attempts to generalize and calibrate deposit shapes in order to quantify holdup will be prone to high uncertainties. Uncertainties on the order of ±50% are typical in holdup results. In international safeguards applications, a ±50% uncertainty can result in a large amount of material unaccounted for (MUF) thereby increasing the difficulty of detecting material diversion and facility misuse. An imaging-based methodology has been developed with the objective of significantly reducing this uncertainty by using the true deposit shape, instead of relying on oversimplified geometric assumptions. The project is a collaboration between ORNL, Y-12, and the University of Tennessee, Knoxville, TN. Uranium sources of known masses were measured using the Germanium Gamma-ray Imager (GeGI), a high-resolution imaging spectrometer, creating a pixelated map for each spectral bin. Two different gamma imaging methods are employed in this work: coded aperture imaging and Compton imaging. A validated MonteCarlo model of the detector has been developed using the GEANT4 code for determining the intrinsic response of the detector, its enclosure, and the coded aperture mask. An inverse transport solver based on the Markov Chain Monte-Carlo approach known as Differential Evolution Adaptive Metropolis (DREAM) is employed to use the measurement data from the image pixels (coded aperture or Compton) to solve for the mass of 235 U in the deposit. A reliable method based on the DREAM solver has been developed to flag the infinite thickness condition of a uranium deposit. The project team is working towards improving the image reconstruction for Compton imaging so that a better localization of the source can be achieved. Besides treating the coded aperture and Compton imaging methods independently, the project is also evaluating a combined method that uses the Compton scatter data from a coded aperture measurement. GEANT4 simulations are being performed to evaluate the combined approach. The impact on the DREAM optimization as the source thickness progressively approaches infinite thickness is being evaluated. A number of uranium sources available at ORNL have been measured, and the DREAM results have been tested and validated for the coded aperture imaging. A similar effort will be carried out to validate the Compton based method once the development of algorithms for better localization are complete. The imaging based quantification is very amenable to unattended monitoring of holdup accumulation at key measurement points. A proof of concept measurement has been completed to demonstrate this capability The current work used the high energy resolution imager GeGI. However, the approach and methodologies are applicable to other imagers such as the cadmium zin telluride (CZT) based imager manufactured by H3D, Inc.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2023

Design, performance, and cost savings of using GF-PC additively manufactured tooling for stamping of HSS 590 sheet metal

Conventional sheet metal forming tooling in the automotive industry is made up of hardened steel and used for mass-production. Prototype tooling made of metal is durable, but it is only used for a small number of parts despite its high cost, contributing heavily to the vehicle development cost. Additive Manufacturing (AM) offers a low-cost alternative for fabricating tooling suitable for low-volume production. Here, this study investigates the feasibility of using AM polymer composite tooling for the stamping of 1.5mm thickness HSS 590 steel sheets through a two-pronged approach – experimental and numerical analysis. Sheet metal stamping experiments were conducted with glass fiber reinforced polycarbonate (GF-PC) AM polymer tooling and their performance was evaluated based on various metrics such as tool deformation and part accuracy. Finite element simulations of the stamping process were performed, which accurately captured the tool deformations seen in experiments with the use of anisotropic material models. The effective use of simulations in optimizing process parameters to achieve the desired final part geometry is also demonstrated. The GF-PC AM polymer composite tooling was found to be well suited for low volume production applications with potential cost savings.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Leveraging Fine-Grained Occupancy Estimation Patterns for Effective HVAC Control

As occupancy sensing technologies become mature, various occupancy sensors are increasingly deployed in commercial buildings for pervasive occupancy monitoring. These sensors provide occupant-count data, which contains rich spatiotemporal information about occupancy patterns. With long-term occupant-count data collected from a commercial building, we design three different predictive models that capture the occupancy dynamics and examine how a model predictive control of the HVAC system benefits from actual occupancy count prediction. Our analysis reveals that mispredictions of occupancy states, especially false positives and false negatives, may introduce inefficient control that leads to energy waste or user discomfort. To address this issue, we take a step further to design an adaptive model predictive controller that minimizes inefficient control actions according to misprediction types and distributions. A comprehensive evaluation is performed in OpenBuild and EnergyPlus simulators to study the effectiveness of the proposed end-to-end control strategy. The evaluation shows that the proposed solution reduces energy consumption by 29.5% while improving the average weighted occupants comfort by 86.7% in Predicted Mean Vote (PMV) over the fixed schedule strategy.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Porting ATLAS Fast Calorimeter Simulation to GPUs with Performance Portable Programming Models

FastCaloSim is a parameterized simulation of the particle energy response and of the energy distribution in the ATLAS calorimeter. It is a relatively small and self-contained package with massive inherent parallelism and captures the essence of GPU offloading via important operations like data transfer, memory initialization, floating point operations, and reduction. It was identified by the High Energy Physics Center for Computational Excellence project as a good testbed for evaluating the performance and ease of portability of programming models. In this paper, we will discuss the results of our evaluation of the porting process to Kokkos, SYCL, Alpaka, OpenMP and std::par (nvc++), and compare performance on NVIDIA, AMD and Intel GPUs, as well as multicore CPUs.

97 MATHEMATICS AND COMPUTING↗