Search NASA⌕ Search

SEARCH · Search NASA

Results for “computer system benchmarking”

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 487 records · Page 27

Quantifying and reducing model-form uncertainties in Reynolds-averaged Navier–Stokes simulations: A data-driven, physics-informed Bayesian approach

Despite their well-known limitations, Reynolds-Averaged Navier–Stokes (RANS) models are still the workhorse tools for turbulent flow simulations in today's engineering analysis, design and optimization. While the predictive capability of RANS models depends on many factors, for many practical flows the turbulence models are by far the largest source of uncertainty. As RANS models are used in the design and safety evaluation of many mission-critical systems such as airplanes and nuclear power plants, quantifying their model-form uncertainties has significant implications in enabling risk-informed decision-making. In this work we develop a data-driven, physics-informed Bayesian framework for quantifying model-form uncertainties in RANS simulations. Uncertainties are introduced directly to the Reynolds stresses and are represented with compact parameterization accounting for empirical prior knowledge and physical constraints (e.g., realizability, smoothness, and symmetry). An iterative ensemble Kalman method is used to assimilate the prior knowledge and observation data in a Bayesian framework, and to propagate them to posterior distributions of velocities and other Quantities of Interest (QoIs). We use two representative cases, the flow over periodic hills and the flow in a square duct, to evaluate the performance of the proposed framework. Both cases are challenging for standard RANS turbulence models. Simulation results suggest that, even with very sparse observations, the obtained posterior mean velocities and other QoIs have significantly better agreement with the benchmark data compared to the baseline results. At most locations the posterior distribution adequately captures the true model error within the developed model form uncertainty bounds. The framework is a major improvement over existing black-box, physics-neutral methods for model-form uncertainty quantification, where prior knowledge and details of the models are not exploited. This approach has potential implications in many fields in which the governing equations are well understood but the model uncertainty comes from unresolved physical processes. - Highlights: • Proposed a physics–informed framework to quantify uncertainty in RANS simulations. • Framework incorporates physical prior knowledge and observation data. • Based on a rigorous Bayesian framework yet fully utilizes physical model. • Applicable for many complex physical systems beyond turbulent flows.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Energy dataset of Frontier supercomputer for waste heat recovery

The Hewlett Packard Enterprise–Cray EX Frontier is the world’s first and fastest exascale supercomputer, hosted at the Oak Ridge Leadership Computing Facility in Tennessee, United States. Frontier is a significant electricity consumer, drawing 8–30 MW; this massive energy demand produces significant waste heat, requiring extensive cooling measures. Although harnessing this waste heat for campus heating is a sustainability goal at Oak Ridge National Laboratory (ORNL), the 30 °C–38 °C waste heat temperature poses compatibility issues with standard HVAC systems. Heat pump systems, prevalent in residential settings and some industries, can efficiently upgrade low-quality heat to usable energy for buildings. Thus, heat pump technology powered by renewable electricity offers an efficient, cost-effective solution for substantial waste heat recovery. However, a major challenge is the absence of benchmark data on high-performance computing (HPC) heat generation and waste heat profiles. This paper reports power demand and waste heat measurements from an ORNL HPC data centre, aiming to guide future research on optimizing waste heat recovery in large-scale data centres, especially those of HPC calibre.

97 MATHEMATICS AND COMPUTING↗

Using scalable computer vision to automate high-throughput semiconductor characterization

Abstract High-throughput materials synthesis methods, crucial for discovering novel functional materials, face a bottleneck in property characterization. These high-throughput synthesis tools produce 10 4 samples per hour using ink-based deposition while most characterization methods are either slow (conventional rates of 10 1 samples per hour) or rigid (e.g., designed for standard thin films), resulting in a bottleneck. To address this, we propose automated characterization (autocharacterization) tools that leverage adaptive computer vision for an 85x faster throughput compared to non-automated workflows. Our tools include a generalizable composition mapping tool and two scalable autocharacterization algorithms that: (1) autonomously compute the band gaps of 200 compositions in 6 minutes, and (2) autonomously compute the environmental stability of 200 compositions in 20 minutes, achieving 98.5% and 96.9% accuracy, respectively, when benchmarked against domain expert manual evaluation. These tools, demonstrated on the formamidinium (FA) and methylammonium (MA) mixed-cation perovskite system FA 1−x MA x PbI 3 , 0 ≤ x ≤ 1, significantly accelerate the characterization process, synchronizing it closer to the rate of high-throughput synthesis.

Science & Technology - Other Topics↗

Cholesky-decomposed density MP2 with density fitting: Accurate MP2 and double-hybrid DFT energies for large systems

Our recently developed QQR-type integral screening is introduced in our Cholesky-decomposed pseudo-densities Møller-Plesset perturbation theory of second order (CDD-MP2) method. We use the resolution-of-the-identity (RI) approximation in combination with efficient integral transformations employing sparse matrix multiplications. The RI-CDD-MP2 method shows an asymptotic cubic scaling behavior with system size and a small prefactor that results in an early crossover to conventional methods for both small and large basis sets. We also explore the use of local fitting approximations which allow to further reduce the scaling behavior for very large systems. The reliability of our method is demonstrated on test sets for interaction and reaction energies of medium sized systems and on a diverse selection from our own benchmark set for total energies of larger systems. Timings on DNA systems show that fast calculations for systems with more than 500 atoms are feasible using a single processor core. Parallelization extends the range of accessible system sizes on one computing node with multiple cores to more than 1000 atoms in a double-zeta basis and more than 500 atoms in a triple-zeta basis.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Exploring Hilbert space on a budget: Novel benchmark set and performance metric for testing electronic structure methods in the regime of strong correlation

This work explores the ability of classical electronic structure methods to efficiently represent (compress) the information content of full configuration interaction (FCI) wave functions. We introduce a benchmark set of four hydrogen model systems of different dimensionalities and distinctive electronic structures: a 1D chain, a 1D ring, a 2D triangular lattice, and a 3D close-packed pyramid. To assess the ability of a computational method to produce accurate and compact wave functions, we introduce the accuracy volume, a metric that measures the number of variational parameters necessary to achieve a target energy error. Using this metric and the hydrogen models, we examine the performance of three classical deterministic methods: (i) selected configuration interaction (sCI) realized both via an a posteriori (ap-sCI) and variational selection of the most important determinants, (ii) an a posteriori singular value decomposition (SVD) of the FCI tensor (SVD-FCI), and (iii) the matrix product state representation obtained via the density matrix renormalization group (DMRG). We find that the DMRG generally gives the most efficient wave function representation for all systems, particularly in the 1D chain with a localized basis. For the 2D and 3D systems, all methods (except DMRG) perform best with a delocalized basis, and the efficiency of sCI and SVD-FCI is closer to that of DMRG. For larger analogs of the models, the DMRG consistently requires the fewest parameters but still scales exponentially in 2D and 3D systems, and the performance of SVD-FCI is essentially equivalent to that of ap-sCI.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Deep potential generation scheme and simulation protocol for the Li 10 GeP 2 S 12 -type superionic conductors

We report solid-state electrolyte materials with superior lithium ionic conductivities are vital to the next-generation Li-ion batteries. Molecular dynamics could provide atomic scale information to understand the diffusion process of Li-ion in these superionic conductor materials. Here, we implement the deep potential generator to set up an efficient protocol to automatically generate interatomic potentials for Li 10 GeP 2 S 12 -type solid-state electrolyte materials (Li 10 GeP 2 S 12 , Li 10 SiP 2 S 12 , and Li 10 SnP 2 S 12 ). The reliability and accuracy of the fast interatomic potentials are validated. With the potentials, we extend the simulation of the diffusion process to a wide temperature range (300 K–1000 K) and systems with large size (~1000 atoms). Important technical aspects such as the statistical error and size effect are carefully investigated, and benchmark tests including the effect of density functional, thermal expansion, and configurational disorder are performed. The computed data that consider these factors agree well with the experimental results, and we find that the three structures show different behaviors with respect to configurational disorder. Our work paves the way for further research on computation screening of solid-state electrolyte materials.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Accelerated kinetic model for global macro stability studies of high-beta fusion reactors

The field reversed configuration (FRC), such as studied in the C-2W experiment at TAE Technologies, is an attractive candidate for realizing a nuclear fusion reactor. In an FRC, kinetic ion effects play the majority role in macroscopic stability, which allows global stability studies to make use of fluid-kinetic hybrid (also referred to as Ohm's law) models wherein ions are treated kinetically while electrons are treated as a fluid. The development and validation of such a hybrid particle-in-cell algorithm in the Exascale Computing Project code WarpX are reported here. Implementation of this model in the WarpX framework benefits from the numerical efficiency of WarpX as well as its scalability on large HPC systems and portability to different architectures. Performance benchmarks of the new algorithm for large, 3-dimensional, full device simulations from the Perlmutter supercomputer are presented. Results of a series of FRC simulations are discussed in which the impact of two-fluid effects on the tilt-mode growth rate was studied. It was observed that, in agreement with previous Hall-MHD studies, two-fluid effects have a stabilizing impact on the tilt mode.

Physics↗

Real-Time Machine Learning Inferencing with Edge Computing Devices from Google and Intel

Machine Learning (ML) has been demonstrated to accelerate discovery and improve the accuracy of many measurements in astrophysics and high energy physics research. As an increasingly commonplace tool, we have reached a new set of challenges in the deployment of ML, at the forefront of which is their computational complexity. The goal of this project is to explore the viability of dedicated machine learning hardware, such as Google’s Coral/Edge TPU and Intel’s Movidius Myriad devices, in astrophysics and high energy physics applications. ML algorithms, while faster than traditional methods, can become the bottleneck in data processing pipelines --- which is a critical issue in systems that require low latency. Within the context of high energy physics and astrophysics, we test this hardware in an attempt to find if it is able to reduce the computational cost of machine learning. We have benchmarked some of these devices, specifically the Google Coral Dev Board and th e Google AIY Vision Kit (Containing Intel’s Movidius Myriad 2).This was done via a strong gravitational lens finding application and a known data set of lenses, simulating a realistic astrophysics application of machine learning, and compare their results to that of traditional computer hardware.

Hawks, Benjamin↗

Implementation and evaluation of permeability-porosity and tortuosity-porosity relationships linked to mineral dissolution-precipitation

Changes of porosity, permeability, and tortuosity due to physical and geochemical processes are of vital importance for a variety of hydrogeological systems, including passive treatment facilities for contaminated groundwater, engineered barrier systems (EBS), and host rocks for high-level nuclear waste (HLW) repositories. Due to the nonlinear nature and chemical complexity of the problem, in most cases, it is impossible to verify reactive transport codes analytically, and code intercomparisons are the most suitable method to assess code capabilities and model performance. This paper summarizes model intercomparisons for six hypothetical scenarios with generally increasing geochemical or physical complexity using the reactive transport codes CrunchFlow, HP1, MIN 3 P, PFlotran, and TOUGHREACT. Benchmark problems include the enhancement of porosity and permeability through mineral dissolution, as well as near complete clogging due to localized mineral precipitation, leading to reduction of permeability and tortuosity. Processes considered in the benchmark simulations are advective-dispersive transport in saturated media, kinetically controlled mineral dissolution-precipitation, and aqueous complexation. Porosity changes are induced by mineral dissolution-precipitation reactions, and the Carman-Kozeny relationship is used to describe changes in permeability as a function of porosity. Archie’s law is used to update the tortuosity and the pore diffusion coefficient as a function of porosity. Results demonstrate that, generally, good agreement is reached amongst the computer models despite significant differences in model formulations. As a result, some differences are observed, in particular for the more complex scenarios involving clogging; however, these differences do not affect the interpretation of system behavior and evolution.

58 GEOSCIENCES↗

Unprecedented cloud resolution in a GPU-enabled full-physics atmospheric climate simulation on OLCF’s summit supercomputer

Clouds represent a key uncertainty in future climate projection. While explicit cloud resolution remains beyond our computational grasp for global climate, we can incorporate important cloud effects through a computational middle ground called the Multi-scale Modeling Framework (MMF), also known as Super Parameterization. This algorithmic approach embeds high-resolution Cloud Resolving Models (CRMs) to represent moist convective processes within each grid column in a Global Climate Model (GCM). The MMF code requires no parallel data transfers and provides a self-contained target for acceleration. This study investigates the performance of the Energy Exascale Earth System Model-MMF (E3SM-MMF) code on the OLCF Summit supercomputer at an unprecedented scale of simulation. Hundreds of kernels in the roughly 10K lines of code in the E3SM-MMF CRM were ported to GPUs with OpenACC directives. A high-resolution benchmark using 4600 nodes on Summit demonstrates the computational capability of the GPU-enabled E3SM-MMF code in a full physics climate simulation.

58 GEOSCIENCES↗

DeepHyper: Asynchronous Hyperparameter Search for Deep Neural Networks

Hyperparameters employed by deep learning (DL) methods play a substantial role in the performance and reliability of these methods in practice. Unfortunately, finding performance-optimizing hyperparameter settings is a notoriously difficult task. Hyperparameter search methods typically have limited production-strength implementations or do not target scalability within a highly parallel machine, portability across different machines, experimental comparison between different methods, and tighter integration with workflow systems. In this paper, we present DeepHyper, a Python package that provides a common interface for the implementation and study of scalable hyperparameter search methods. It adopts the Balsam workflow system to hide the complexities of running large numbers of hyperparameter configurations in parallel on high-performance computing (HPC) systems. We implement and study asynchronous model-based search methods that consist of sampling a small number of input hyperparameter configurations and progressively fitting surrogate models over the input-output space until exhausting a user-defined budget of evaluations. We evaluate the efficacy of these methods relative to approaches such as random search, genetic algorithms, Bayesian optimization, and hyperband on DL benchmarks on CPU- and GPU-based HPC systems.

Balaprakash, Prasanna↗

Enabling High Performance Application I/O

This work provides software that enables scientific applications to more efficiently access available storage resources at different levels of interfaces. We developed scalable techniques and optimizations for PVFS parallel file systems, MPI I/O, and parallel netCDF I/O library. These implementations were evaluated using production application I/O kernels as well as popular I/O benchmarks and demonstrated promising results. The software developed under this work has been made available to the public via MCS, ANL web sites.

97 MATHEMATICS AND COMPUTING↗

Quantifying uncertainties due to irreducible three-body forces in deuteron-nucleus reactions

Deuteron-induced nuclear reactions are an essential tool for probing the structure of nuclei as well as astrophysical information such as (n, γ) cross sections. The deuteron-nucleus system is typically described within a Faddeev three-body model consisting of a neutron (n), a proton (p), and the target nucleus (A) interacting through pairwise phenomenological potentials. While Faddeev techniques enable the exact description of the three-body dynamics, their predictive power is limited in part by the omission of irreducible neutron-proton-nucleus three-body force (n–p–A 3BF). Here, our goal is to quantify systematic uncertainties stemming from the reduction of deuteron-nucleus (d + A) dynamics to a picture of three pointlike nuclear clusters interacting via pairwise nucleon-nucleus forces, using as testing grounds d + α scattering and the 6 Li ground state. We particularly focus on quantifying uncertainties arising from the full antisymmetrization of the (A + 2)-body system with the target nucleus fixed in its ground state. We adopt the ab initio no-core shell model coupled with the resonating group method (NCSM/RGM) to compute microscopic n–α and p–α interactions, and use them in a three-body description of the d + α system by means of momentum-space Faddeev-type equations. Simultaneously, we also carry out ab initio calculations of d + α scattering and 6 Li ground state by means of six-body NCSM/RGM calculations to serve as a benchmark for the three-body model predictions given by the Faddeev calculations. By comparing the Faddeev and NCSM/RGM results, we show that the irreducible n–p–α 3BF has a non-negligible effect on bound state and scattering observables alike. Specifically, the Faddeev approach yields a 6 Li ground state that is approximately 600 keV shallower than the one obtained with the NCSM/RGM. Additionally, the Faddeev calculations for d + α scattering yield a 3 + resonance that is located approximately 400 keV higher in energy compared to the NCSM/RGM result. The shape of the d + α angular distributions computed using the two approaches also differ, owing to the discrepancy in the predictions of the 3 + resonance energy. The Faddeev three-body model predictions for d + α scattering and 6 Li using microscopic n–α and p–α potentials differ from those computed microscopically with the NCSM/RGM. These discrepancies are due to the n–p–α 3BF, which arises from two-nucleon exchange terms in the microscopic d–α interaction and are not accounted for in the three-body model Faddeev calculations. This study lays the foundation for future parametrizations of the 3BF due to Pauli exclusion principle effects in improved three-body calculations of deuteron-induced reactions.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Active space selection with self-healing diffusion Monte Carlo algorithms for periodic solids

Multideterminant Diffusion Monte Carlo (DMC) displays improved accuracy over single determinant DMC. Self-Healing Diffusion Monte Carlo (SHDMC) is a DMC based method that iteratively improves a multideterminant trial wavefunction. Although configuration interaction or complete active space (CAS) methods are very accurate and computationally feasible for many systems, they are not optimal for application to solids. SHDMC is accurate and designed for application to solids, so developing SHDMC based active space selection algorithms is a worthy endeavor. Here, we present and compare active space selection algorithms that are designed for use in conjunction with SHDMC, without relying on external approaches. For benchmarking, we calculated the ground state energy of a small unit cell of graphene and compared the results with a complete basis set extrapolated selected CI and a reference SHDMC trajectory. We found that systematically expanding the active space using an “auto-branching” algorithm optimally balances accuracy with computational practicality. To the best of our knowledge, this is the first work that demonstrates completely self-contained DMC-based active space selection algorithms that do not depend on external methods for determinant selection.

Spanedda, Nicole [ORNL]↗

Final Report: DOE/ID/14215

The proposed straw separation system developed in the research project harvests the large internode sections of the straw which has the greater potential as a feedstock for lignocellulosic ethanol production while leaving the chaff and nodes in the field. This strategy ensures sustainable agriculture by preventing the depletion of soil minerals, and it restores organic matter to the soil in amounts and particle sizes that accommodate farmers’ needs to keep tillage and fertilizer costs low. A ton of these nutrient-rich plant tissues contains as much as $10.55 worth of fertilizer (economic and energy benefits), in terms of nitrogen, phosphorus, potassium, and other nutrients provided to the soil when incorporated by tillage instead of being burned. Biomass conversion to fermentable sugars for the purpose of producing fuels, chemicals, and other industrial products is well understood. Most bioenergy strategies rely on low-cost fermentable sugars for sustainability and economic viability in the marketplace. Exploitation of the “whole crop”—specifically, wheat straw or other plant material currently regarded as residue or waste—is a practical approach for obtaining a reliable and low-cost source of sugars. However, industrial-scale production of sugars from wheat straw, while technically feasible, is plagued by obstacles related to capital costs, energy consumption, waste streams, production logistics, and the quality of the biomass feedstock. Currently available separation options with combine harvesters are not able to achieve sufficient separation of the straw/stover and chaff streams to realize the full potential of selective harvest. Since ethanol yield is a function of feedstock structural carbohydrate content, biomass anatomical fractions of higher product yield can have a significant beneficial impact on minimum ethanol selling price. To address this advanced biomass separation computation engineering models were developed to more effectively and efficiently engineer high-fidelity and high throughput separation systems for biomass components. INL and Iowa State University developed a computational modeling strategy for simulating multi-phase flow with an integrated solver using various computational fluid dynamics (CFD) codes. ISU set up a classic multi-phase test problem to be solved by the various CFD codes. The benchmark case was based on experimental data for bubble gas holdup and bed expansion for a gas/solid fluidized bed. Preliminary fluidization experiments identified some unexpected fluidization behavior, where rather than the bed uniformly fluidizing, a “blow out” would occur where a hole would open up in the bed through which the air would preferentially flow, resulting in erratic fluidization. To improve understanding of this phenomena and aid in building a design tool, improved computational tools were developed. The virtual engineering techniques developed were tested and utilized to design a separation baffle in a CNH combine. A computational engineering approach involving modeling, analysis, and simulation was used in the form of virtual engineering to design a baffle separator capable of accomplishing the high-fidelity residue separation established by the performance targets. Through the use of the virtual engineering model, baffle designs were simulated to (1) determine the effect of the baffle on the airflow of the combine cleaning system, and (2) predict the effectiveness of the baffle in separating the residue streams. A baffle design was selected based on the virtual engineering modeling, built into the INL selective harvest test combine. The result of the baffle changes improved the crop separation capability of the combine, enabling downstream improvement in composition and theoretical ethanol yield. In addition, the positive results from the application of the virtual engineering tools to the CNH combine design resulted in further application of these tools to other INL areas of research. INL and the University of Idaho identified, characterized, and modified a key plant biosynthetic lignin gene, cinnamoyl CoA reductase (CCR), to assess its influence on the efficiency with which the resulting biomaterials interact with and are processed by engineering systems. The characterization of CCR genes and resulting antisense gene constructs enabled the modification and assessment of the impact of cell wall biosynthetic gene in wheat. This allows the model systems needed to study the structural and functional post-harvest properties of straw to be created and assessed. Progress was made toward identifying and using key CCR lignin biosynthesis gene sequences for controlling lignin biosynthesis with the goal of altering lignin content and composition in wheat. This was important because of the potential impact that reduced lignin could have on feedstock harvesting, transportation, storage, pretreatment, and processing.

09 BIOMASS FUELS↗

Summer Internship Report: ARA2 Benchmarking

Over the past decade, the RISC-V Instruction Set Architecture (ISA) has emerged as a significant player in both academic and industrial processor design due to its open-source nature, modular extension system, and versatility across domains ranging from microcontrollers to high-performance computing (HPC). One of its most important recent advancements is the RISC-V Vector Extension (RVV), which enables explicit data-level parallelism through vector registers and vectorized instructions. Unlike traditional SIMD (Single Instruction, Multiple Data) architectures that fix vector lengths at design time, RVV uses the concept of VLEN (vector register length) as a hardware-independent parameter and allows software to adapt dynamically to the available vector width. This flexible approach ensures portability across implementations while enabling scalable performance. The ARA2 core is a parameterizable RISC-V vector processor developed at the Integrated Systems Lab at ETH Zürich and the University of Bologna. Designed as a tightly-coupled accelerator to a scalar RISC-V core, ARA2 implements the RVV 1.0 specification and offers tunable architectural parameters such as the number of vector lanes, VLEN, and cache sizes.

97 MATHEMATICS AND COMPUTING↗

Tools for Simulation and Benchmark Generation at Exascale

The path to exascale high-performance computing (HPC) poses several challenges related to power, performance, resilience, productivity, programmability, data movement, and data management. Investigating the performance of parallel applications at scale on future architectures and the performance impact of different architecture choices is an important component of HPC hardware/software co-design. Simulations using models of future HPC systems and communication traces from applications running on existing HPC systems can offer an insight into the performance of future architectures. This work targets technology developed for scalable application tracing of communication events and memory profiles, but can be extended to other areas, such as I/O, control flow, and data flow. It further focuses on extreme-scale simulation of millions of Message Passing Interface (MPI) ranks using a lightweight parallel discrete event simulation (PDES) toolkit for performance evaluation. Instead of simply replaying a trace within a simulation, the approach is to generate a benchmark from it and to run this benchmark within a simulation using models to reflect the performance characteristics of future-generation HPC systems. This provides a number of benefits, such as eliminating the data intensive trace replay and enabling simulations at different scales. The presented work utilizes the ScalaTrace tool to generate scalable trace files, the ScalaBenchGen tool to generate the benchmark, and the xSim tool to run the benchmark within a simulation.

Lagadapati, Mahesh↗