Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “computer Codes Computer Programs”

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 1,621 records · Page 90
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

WHONDRS-GUI: a web application for global survey of surface water metabolites

Background The Worldwide Hydrobiogeochemistry Observation Network for Dynamic River Systems (WHONDRS) is a consortium that aims to understand complex hydrologic, biogeochemical, and microbial connections within river corridors experiencing perturbations such as dam operations, floods, and droughts. For one ongoing WHONDRS sampling campaign, surface water metabolite and microbiome samples are collected through a global survey to generate knowledge across diverse river corridors. Metabolomics analysis and a suite of geochemical analyses have been performed for collected samples through the Environmental Molecular Sciences Laboratory (EMSL). The obtained knowledge and data package inform mechanistic and data-driven models to enhance predictions of outcomes of hydrologic perturbations and watershed function, one of the most critical components in model-data integration. To support efforts of the multi-domain integration and make the ever-growing data package more accessible for researchers across the world, a Shiny/R Graphical User Interface (GUI) called WHONDRS-GUI was created. Results The web application can be run on any modern web browser without any programming or operational system requirements, thus providing an open, well-structured, discoverable dataset for WHONDRS. Together with a context-aware dynamic user interface, the WHONDRS-GUI has functionality for searching, compiling, integrating, visualizing and exporting different data types that can easily be used by the community. The web application and data package are available at https://data.ess-dive.lbl.gov/view/doi:10.15485/1484811 , which enables users to simultaneously obtain access to the data and code and to subsequently run the web app locally. The WHONDRS-GUI is also available for online use at Shiny Server ( https://xmlin.shinyapps.io/whondrs/ ).

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2012

Dose calculation software for helical tomotherapy, utilizing patient CT data to calculate an independent three-dimensional dose cube

Purpose: Treatment plans for the TomoTherapy unit are produced with a planning system that is integral to the unit. The authors have produced an independent dose calculation system, to enable plans to be recalculated in three dimensions, using the patient's CT data. Methods: Software has been written using MATLAB. The DICOM-RT plan object is used to determine the treatment parameters used, including the treatment sinogram. Each projection of the sinogram is segmented and used to calculate dose at multiple calculation points in a three-dimensional grid using tables of measured beam data. A fast ray-trace algorithm is used to determine effective depth for each projection angle at each calculation point. Calculations were performed on a standard desktop personal computer, with a 2.6 GHz Pentium, running Windows XP. Results: The time to perform a calculation, for 3375 points averaged 1 min 23 s for prostate plans and 3 min 40 s for head and neck plans. The mean dose within the 50% isodose was calculated and compared with the predictions of the TomoTherapy planning system. When the modified CT (which includes the TomoTherapy couch) was used, the mean difference for ten prostate patients, was -0.4% (range -0.9% to +0.3%). With the original CT (which included the CT couch), the mean difference was -1.0% (range -1.7% to 0.0%). The number of points agreeing with a gamma 3%/3 mm averaged 99.2% with the modified CT, 96.3% with the original CT. For ten head and neck patients, for the modified and original CT, respectively, the mean difference was +1.1% (range -0.4% to +3.1%) and 1.1% (range -0.4% to +3.0%) with 94.4% and 95.4% passing a gamma 4%/4 mm. The ability of the program to detect a variety of simulated errors has been tested. Conclusions: By using the patient's CT data, the independent dose calculation performs checks that are not performed by a measurement in a cylindrical phantom. This enables it to be used either as an additional check or to replace phantom measurements for some patients. The software has potential to be used in any application where one wishes to model changes to patient conditions.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2021

Uncertainty Analysis for a Criticality Benchmark [Slides]

This presentation discusses how KRUSTY was analyzed with LANL’s MCNP neutron transport code, Monte Carlo N-Particle. In the Monte Carlo method, neutrons are born in fission events and fly around the geometry having events (e.g., scatter, capture, fission, escape) according to the natural probabilities. “Natural probabilities” are given as neutron cross sections , which are measured and evaluated nuclear data. One of the goals of the benchmark project is to improve the nuclear cross sections (and the codes that use them). In this work, the response of interest is called k eff , which describes the neutron multiplication in a near critical system. Like most quantities, k eff is not measured directly; it is only inferred. In summation, MCNP results for k eff matched the measurements extremely well. The evaluated uncertainties are asymmetric and smaller than 0.00100 (e.g., + 0.00080 / – 0.00062). The largest uncertainties are due to alignment.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2015

Creation of a Benchmark Suite for SNL Nuclear Criticality Safety Code Validation (Volume 1 - MCNP)

The Nuclear Criticality Safety Program at Sandia National Laboratories has developed a suite of benchmark problems to be utilized for validation of MCNP6 Version 1.0 with ENDF/B-VII Release 1 cross sections. The benchmark suite covers a broad range of fissile material types, material forms, moderators, reflectors, and neutron energy spectra. It is anticipated that this benchmark suite will cover the vast majority of critical safety applications at SNL. The benchmark suite establishes a Bias and Bias Uncertainty for use in criticality safety analyses. In addition, the Bias and Bias Uncertainty value derived from the benchmark suite using the traditional SNL NCS methodology is demonstrated to be equivalent to the more robust statistical techniques used at many DOE sites.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2019

Chapter Six - Derivation, cross-validation, and comparison of analytic formulas for electrostatic deflector aberrations

Here, we derived first and second order analytic aberration formulas in the horizontal transverse plane for an electrostatic deflector specified by the reference orbit radius, the central angle spanning the deflector, and its inhomogeneity coefficients. The derivation was performed using an iterative order-by-order perturbation method in a Frenet–Serret beamline coordinate system. We produced a C program edabrt for calculation of the first and second order aberrations using the formulas that we derived. The electrostatic deflector aberration formulas from Wollnik (1965) disagree with COSY INFINITY and the aberration formulas that we derived here, and they also deviate from the first and second order symplecticity conditions. The reason for this discrepancy is that the aberration formulas from Wollnik (1965) consider only motion in the main field of the deflector and do not properly account for fringe field effects, in particular the necessarily occurring change of kinetic energy due to the change in potential. The code GIOS uses electrostatic aberration formulas from Wollnik (1965) and exhibits the same discrepancy. We performed a comparison of first and second order electrostatic deflector aberrations for (1) the analytic formulas derived in this work, (2) differential-algebraic (DA) numerical integration of the equations of motion using the code COSY INFINITY, (3) the aberration formulas from Wollnik (1965), (4) aberrations computed using the code GIOS, and (5) the aberration formulas from Wollnik (1965) adjusted to account for the occurring change in potential. An electrostatic spherical deflector and an electrostatic cylindrical deflector were used as test cases for this comparison. There is excellent agreement between methods (1), (2), and (5), and these three methods satisfy the first and second order symplecticity conditions.

43 PARTICLE ACCELERATORS↗
DOE OSTI2016

Investigations on RELAP5-3D to RELAP5-3D coupling methodology by PVMEXEC

In the framework of a BEPU (Best Estimate Plus Uncertainty) approach within the licensing process of a nuclear power plant, the need to extend the resources of nuclear system thermal-hydraulics codes, such as RELAP5-3D, arises to allow more detailed simulations of the complex 3D reality of Nuclear Power Plants (NPPs), either under normal steady-state or during various accident scenarios. Currently, it is not possible to achieve the same degree of detail for a whole nuclear system when it is simulated with RELAP5-3D and this is due to the inherent limitations in the number of components and volumes to be used for the analysis. For this reason, it is of extreme interest the use of tools for codes coupling that enable the use of different codes for the simulation of different portions of a system in a unified analysis. In this paper the attention will be focused on the decomposition of the thermal-hydraulic domain of a system into subsystems to be simulated by different instances of the same code (e.g. RELAP5-3D) coupled together by means of PVMEXEC program and parallel virtual machine (PVM) technology. Explicit and semi-implicit solution algorithms were used for the analyses. Among the analyzed cases, the following will be discussed in detail with the aim to provide additional guidelines for the use of the PVMEXEC tool: (i) the Edward's pipe blowdown test, (ii) a simplified countercurrent heat exchanger, (iii) different hydraulics and heat structure coupling schemes for a shell-tube heat exchanger and (iv) a three-task coupled model of an adaptation of the Christensen subcooled boiling experiment. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2020

Novel Geometric Operations for Linear Programming

This report summarizes the work performed under the project "Linear Programming in Strongly Polynomial Time." Linear programming (LP) is a classic combinatorial optimization problem heavily used directly and as an enabling subroutine in integer programming (IP). Specifically IP is the same as LP except that some solution variables must take integer values (e.g. to represent yes/no decisions). Together LP and IP have many applications in resource allocation including general logistics, and infrastructure design and vulnerability analysis. The project was motivated by the PI's recent success developing methods to efficiently sample Voronoi vertices (essentially finding nearest neighbors in high-dimensional point sets) in arbitrary dimension. His method seems applicable to exploring the high-dimensional convex feasible space of an LP problem. Although the project did not provably find a strongly-polynomial algorithm, it explored multiple algorithm classes. The new medial simplex algorithms may still lead to solvers with improved provable complexity. We describe medial simplex algorithms and some relevant structural/complexity results. We also designed a novel parallel LP algorithm based on our geometric insights and implemented it in the Spoke-LP code. A major part of the computational step is many independent vector dot products. Our parallel algorithm distributes the problem constraints across processors. Current commercial and high-quality free LP solvers require all problem details to fit onto a single processor or multicore. Our new algorithm might enable the solution of problems too large for any current LP solvers. We describe our new algorithm, give preliminary proof-of-concept experiments, and describe a new generator for arbitrarily large LP instances.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Development and experimental qualification of a calculation scheme for the evaluation of gamma heating in experimental reactors. Application to MARIA and Jules Horowitz (JHR) MTR Reactors

Precise computational determination of nuclear heating which consists predominantly of gamma heating (more than 80 %) is one of the challenges in material testing reactor exploitation. Due to sophisticated construction and conditions of experimental programs planned in JHR it became essential to use most accurate and precise gamma heating model. Before the JHR starts to operate, gamma heating evaluation methods need to be developed and qualified in other experimental reactor facilities. This is done inter alia using OSIRIS, MINERVE or EOLE research reactors in France. Furthermore, MARIA - Polish material testing reactor - has been chosen to contribute to the qualification of gamma heating calculation schemes/tools. This reactor has some characteristics close to those of JHR (beryllium usage, fuel element geometry). To evaluate gamma heating in JHR and MARIA reactors, both simulation tools and experimental program have been developed and performed. For gamma heating simulation, new calculation scheme and gamma heating model of MARIA have been carried out using TRIPOLI4 and APOLLO2 codes. Calculation outcome has been verified by comparison to experimental measurements in MARIA reactor. To have more precise calculation results, model of MARIA in TRIPOLI4 has been made using the whole geometry of the core. This has been done for the first time in the history of MARIA reactor and was complex due to cut cone shape of all its elements. Material composition of burnt fuel elements has been implemented from APOLLO2 calculations. An experiment for nuclear heating measurements and calculation verification has been done in September 2014. This involved neutron, photon and nuclear heating measurements at selected locations in MARIA reactor using in particular Rh SPND, Ag SPND, Ionization Chamber (all three from CEA), KAROLINA calorimeter (NCBJ) and Gamma Thermometer (CEA/SCK CEN). Measurements were done in forty points using four channels. Maximal nuclear heating evaluated from measurements is of the order of 2.5 W/g at half of the possible MARIA power - 15 MW. The approach and the detailed program for experimental verification of calculations will be presented. The following points will be discussed: - Development of a gamma heating model of MARIA reactor with TRIPOLI 4 (coupled neutron-photon mode) and APOLLO2 model taking into account the key parameters like: configuration of the core, experimental loading, control rod location, reactor power, fuel depletion); - Design of specific measurement tools for MARIA experiments including for instance a new single-cell calorimeter called KAROLINA calorimeter; - MARIA experimental program description and a preliminary analysis of results; - Comparison of calculations for JHR and MARIA cores with experimental verification analysis, calculation behavior and n-γ 'environments'. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2024

Sum-of-Fractions Methodology for Actinides in Water- and Polyethylene-Moderated and -Reflected Systems

Sum-of-fractions is a method intended to make sure a subcritical margin for aqueous solutions and slurries of fissionable isotopes exists. The method indicates that a system is subcritical if the sum of the ratios of the mass of each isotope (in a mixture) to its individual minimum subcritical mass limit is less than or equal to one. Historically, the basis of the sum-of-fractions has been derived from allowances given in the American National Standards Institute (ANSI)/ American Nuclear Society (ANS)-8.15-1981. However, the allowance was removed in ANSI/ANS-8.15-2014 due to a lack of technical basis. A methodology was developed to assess the validity of using the sum-of-fractions for water- or polyethylene-moderated systems for the following nuclides: 232 U, 233 U, 234 U, 235 U, 237 Np, 236 Pu, 238 Pu, 239 Pu, 240 Pu, 241 Pu, 242 Pu, 241 Am, 242 m Am, 243 Am, 242 Cm, 243 Cm, 244 Cm, 245 Cm, 246 Cm, 247 Cm, 249 Cf, and 251 Cf. The methodology uses available benchmark data for mixtures of 233 U, 235 U, and 239 Pu to establish the calculational margin, and a mass limit reduction to establish the margin of subcriticality. Water- or polyethylene-moderated and -reflected mixtures containing the nuclides are evaluated with the code system, SCALE 6.2.4. Including the calculational margin, subcritical mass limits for each nuclide were computed for optimally water- or polyethylene-moderated and fully reflected systems. These masses were used to create nuclide mixtures in which the sum of the mass to subcritical mass limit ratios is one. The various nuclide mixtures were modeled over a range of moderation and demonstrate the keff does not exceed the calculational margin. For additional assurance of subcriticality, a significant mass reduction is applied to each computed minimum critical mass of the nuclides without adequate benchmark data consistent with the method in ANSI/ANS-8.15-2014.

07 ISOTOPE AND RADIATION SOURCES↗
DOE OSTI2014

Experimental and Computational Studies of High Energy Density Plasma Streams Ablated from Fine Wires

This report summarizes progress during the final year of this project to study the physics of high energy density (HED) plasma streams of 10^17-10^20/cm3 density and high velocity (~100-500 km/s). Such streams are produced from 5-250 micrometer diameter wires heated and ionized by a 1 MA, 250 ns current pulse on the COBRA pulsed power facility at Cornell University. Plasma is ablated from the wires and is driven away to high velocity by unbalanced JxB force. A wire, or an array of wires, can persist as an essentially stationary, continuous source of this streaming plasma for >200 ns, even with driving magnetic fields of many Tesla and peak current densities in the plasma of many MA/cm2. At the heart of the ablation stream generation is the continuous transport of mass from the relatively cold, near-solid-density wire "core" into current-carrying plasma within 1 mm of the wire, followed by the magnetic acceleration of that plasma and its trapped flux to form a directed stream. In the first two years of this program, an advancing understanding of ablation physics led to the discovery of several novel wire ablation experimental regimes. In the final year, one of these new HED plasma regimes has been studied in quantitative detail. This regime studies highly reproducible magnetic reconnection in strongly radiating plasma with supersonic and superalfvenic flow, and shock structures in the outflow. The key discovery is that very heavy wires, e.g. 250 micrometer diameter Al or 150 micrometer Cu, behave in a qualitatively different way than the lighter wires typically used in wire-array Z-pinches. Such wires can be configured to produce a static magnetic X-point null geometry that stores magnetic and thermal energy; reconnection and outflow are triggered when the current begins to decrease and the electric field reverses. The reconnecting flow is driven by both magnetic and thermal pressure forces, and it has been found to be possible to vary the configuration so that one or the other dominates. The magnetic null extends into a current sheet that is heated and radiates strongly, with supersonic outflows. This is the first study of reconnection in this HED plasma regime. This compressible, radiative regime, and the triggering mechanism, may be relevant to solar and astrophysical processes. The PERSEUS extended MHD code has been developed for simulation of these phenomena, and will continue to be used and further developed to help interpret and understand experimental results, as well as to guide experimental design. The code is well-suited to simulations of shocks, and includes Hall and electron inertia physics that appear to be of importance in a number of ablation flow regimes, and definitely in the reconnection regime when gradient scales are comparable to the ion inertial scale. During the final year, our graduate student supported by this grant completed a new version of PERSEUS with the finite volume computational scheme replaced by a discontinuous Galerkin method that gives much less diffusive behavior and allows faster run time and higher spatial resolution. The code is now being used to study shock structures produced in the outflow region of the reconnection regime.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

OpenACC unified programming environment for GPU and FPGA multi-hybrid acceleration

Attached accelerators have been frequently used in recent High Per- formance Computing (HPC) systems because of their high performance/power ratio. In particular, the Graphics Processing Unit (GPU) is the most popu- lar accelerator owing to its high peak FLOPS performance and high memory bandwidth supported by HBM2, etc. However, the performance of GPU depends highly on a large degree of SIMD parallelism and has difficulty sustaining a high performance on programs with frequent branch operations or a partially low degree of parallelism.By contrast, a Field Programmable Gate Array (FPGA) has received attention as a different type of accelerator than GPU as a fully reconfigurable processor fitting the target applications. The high performance of FPGA is mainly provided by a pipelined operation and optimized circuit suitable for any operation even with frequent conditional branches. We have been focusing on the flexibility of FPGA to compensate for the weakness of GPU. We believe that the coupling of GPU with FPGA can result in one of the most powerful accelerating platforms available.However, the program coding of GPU and FPGA coupling can be quite difficult for application users. Traditionally, CUDA by NVIDIA has been the most popular programming language with the largest share of GPUs used in HPC, whereas a hardware description language such as Verilog HDL has been used in FPGA programming. OpenCL coding has recently become available even on high-end FPGAs. Moreover, several recent studies have also enabled the OpenACC coding for use in FPGA. In this study, we provide a unified programming system based on OpenACC for a platform equipped with both GPU and FPGA aiming at the next-generation accelerated supercomputer framework. Our programming environment is called Multi-Hybrid OpenACC Translator (MHOAT), and in this paper, we describe the basic concept and prototype system of MHOAT based on an evaluation on the amount of coding required and the performance of a hybrid multi-device accelerated system.

Tsunashima, Ryuta↗
DOE OSTI2022

Development of a New Fixed-source Sensitivity Tally Capability in the MCNP® code [Abstract]

The development of a new fixed-source sensitivity tally capability is currently underway in the MCNP code. In recent research and development efforts that utilize machine learning to both seek problematic nuclear data as well as design experiments optimized to improve the nuclear data, the adjoint-weighted k-eigenvalue sensitivity tally capabilities have been heavily essential. In this paper, the motivation to expand the sensitivity tally capabilities beyond k-eigenvalues toward diverse fixed-source problems along with preliminary results and verification will be discussed.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2012

Presentation of FP7 matter project: general overview

The 2010-2012 implementation plan of the European Sustainable Nuclear Industrial Initiative (ESNII), prepared in the frame of the Sustainable Nuclear Energy Technology Platform (SNETP), establishes a very tight time schedule for the start of construction of the European Gen IV prototypes; namely the construction of the LFR ETPP (European Technology Pilot Plant) MYRRHA will start in 2014 and that of the SFR Prototype ASTRID will start in 2017. The GEN IV reactors pose new challenges to the designers and scientists in terms of higher operating temperature, higher fuel burn-up, and in some cases more corrosive environment with respect to the present technologies and which impacts the materials performance. In this frame, the MATTER (Materials Testing and Rules) Project starts well targeted R and D activities to perform careful materials studies in GEN IV operational conditions and to find out criteria for the correct use of these materials in relevant reactor applications. Aim of the MATTER Project (that involved 27 partners and will end in 2015) is to complement the materials researches, in the frame of the European Energy Research Alliance (EERA) guidelines, with the implementation of pre-normative rules. The MATTER Project is divided in 3 technical Domains (called DM): DM1 - Development of test and evaluation guidelines for structural materials: to develop/establish best practice guidelines for testing and evaluation procedures, which are aimed to screen and characterize nuclear materials for innovative nuclear systems. DM2 - Pre-normative R and D for Codes and Standards: Pre-normative activities are performed, comprehensive of experiments, to revise and update the design rules (with an EU level consensus) in order to answer to some short term needs of the two projects ASTRID and MYRRHA with respect to the design and the construction of structural components. DM3 - Joint Program Scheme, implementation and Priorities: to optimise the effectiveness and efficiency of the EERA Joint Program on nuclear materials for innovative reactors and to support specific research activities related to fundamental understanding of ODS steels fabrication. ODS steels are considered candidate materials, in the medium-long term, for high fuel burn-up cladding application. After a brief presentation of DM1 and DM3, this paper mainly focuses on description of Pre-normative R and D activities for Codes and Standards (DM2). (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2024

A new portable random number generator wrapper library

Random number generator is an important component of many scientific projects. Many projects are written using programming models (like OpenMP and SYCL) to target different architectures. However, some programming models do not provide a random number generator. In this work, we introduce our random number generator wrapper. It is a header-only library that supports three distributions of random numbers: uniform, normal, and poisson. On the GPU backend, it wraps the cuRAND and rocRAND library, and supports various random number engines. It also wraps random123, a counterbased random number generator, on both CPU and GPU. With this library, we can generate random numbers with a few lines of code and target both GPU and multi-thread CPU with the same code. We also investigate the performance and scalability of this wrapper on different architectures with different engines and the number of cores.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Method of Test for Evaluating Building Performance Simulation Software

ANSI/ASHRAE Standard 140 Method of Test for Evaluating Building Performance Simulation Software specifies the method to test the core competency of building performance simulation (BPS) software, a broad class of software which includes building energy modeling (BEM) software. Standard 140 has been widely used by BEM software vendors to help diagnose and compare the results from the program with other modeling programs. The Standard is also referenced by codes, standards, government programs, and other incentive programs as a part of minimum requirements for qualifying BEM software. The explanatory information, summary tables and figures in this 140 User’s Manual (referred to as “this Manual”) are provided to help users in implementing various suites of tests specified in Standard 140-2023 (referred to in this manual as “Standard 140” or “the Standard”).

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

OVERVIEW OF PARTICLE DEPOSITION MODELS FOR SPENT NUCLEAR FUEL STORAGE SYSTEMS

Deposition models were built to evaluate contaminant deposition on spent nuclear fuel (SNF) canisters. The primary contaminant of concern is chloride, which is dispersed in the atmosphere and then deposits onto the canisters. During dry storage, the primary degradation process is likely to be Chloride Induced Stress Corrosion Cracking (CISCC) at the heat-affected zones of the canister welds. It is known that stainless steel canisters are susceptible to CISCC; however, the rate of chloride deposition onto the canisters is poorly known, based on sparse field data from a small number of sites. The models presented in this study could be useful for determining the rate of deposition on the canisters and the likelihood of CISCC to help with SNF canister ageing management. The deposition models were developed with the commercial computational fluid dynamics (CFD) code STAR-CCM+. Various deposition mechanisms were considered and incorporated into the models, and a sensitivity study was conducted to determine the most important mechanisms for deposition within a SNF storage system. The models included both a vertical and horizontal configuration storage system: NAC International’s Modular, Advanced Generation, Nuclear All-purpose STORage System (MAGNASTOR®) and a NUHOMS® horizontal storage module respectively. The resulting canister deposition on the horizontal canister is visually compared with inspection data taken onsite at the Calvert Cliffs Nuclear Power Plant. These models are preliminary, and development of the models will continue. Future validation exercises are currently being planned, including the Canister Deposition Field Demonstration (CDFD) effort funded by U.S. Department of Energy office of Nuclear Energy office of Spent Fuel Waste Science and Technology. The goal of the modeling presented is to demonstrate a potential modeling technique that could be used to plan and inform SNF canister ageing management programs with predictive models for the timing and occurrence of canister CISCC.

Suffield, Sarah R.↗
DOE OSTI2016

On-demand data analytics in HPC environments at leadership computing facilities: Challenges and experiences

Scaling-up scientific data analysis and machine learning algorithms for data-driven discovery is a grand challenge that we face today. Despite the growing need for analysis from science domains that are generating `Big Data' from instruments and simulations, building high-performance analytical workflows of data-intensive algorithms have been daunting because: (i) the `Big Data' hardware and software architecture landscape is constantly evolving, (ii) newer architectures impose new programming models, and (iii) data-parallel kernels of analysis algorithms and their performance facets on different architectures are poorly understood. To address these problems, we have: (i) identified scalable data-parallel kernels of popular data analysis algorithms, (ii) implemented `Mini-Apps' of those kernels using different programming models (e.g. Map Reduce, MPI, etc.), (iii) benchmarked and validated the performance of the kernels in diverse architectures. In this paper, we discuss two of those Mini-Apps and show the execution of principal component analysis built as a workflow of the Mini-Apps. We show that Mini-Apps enable scientists to (i) write domain-specific data analysis code that scales on most HPC hardware and (ii) and offers the ability (most times with over a 10x speed-up) to analyze data sizes 100 times the size of what off-the-shelf desktop/workstations of today can handle.

Harney, John↗
DOE OSTI2017

Benchmarking and Evaluating Unified Memory for OpenMP GPU Offloading

Here, the latest OpenMP standard offers automatic device offloading capabilities which facilitate GPU programming. Despite this, there remain many challenges. One of these is the unified memory feature introduced in recent GPUs. GPUs in current and future HPC systems have enhanced support for unified memory space. In such systems, CPU and GPU can access each other's memory transparently, that is, the data movement is managed automatically by the underlying system software and hardware. Memory over subscription is also possible in these systems. However, there is a significant lack of knowledge about how this mechanism will perform, and how programmers should use it. We have modified several benchmarks codes, in the Rodinia benchmark suite, to study the behavior of OpenMP accelerator extensions and have used them to explore the impact of unified memory in an OpenMP context. We moreover modified the open source LLVM compiler to allow OpenMP programs to exploit unified memory. The results of our evaluation reveal that, while the performance of unified memory is comparable with that of normal GPU offloading for benchmarks with little data reuse, it suffers from significant overhead when GPU memory is over subcribed for benchmarks with large amount of data reuse. Based on these results, we provide several guidelines for programmers to achieve better performance with unified memory.

97 MATHEMATICS AND COMPUTING↗