Search NASA⌕ Search

SEARCH · Search NASA

Results for “Low Precision”

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 55 records · Page 3
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Shifting Between Compute and Memory Bounds: A Compression-Enabled Roofline Model

In the evolving landscape of high-performance computing, especially to fight the end of Moore’s Law and Dennard’s Scaling, the ability to shift between compute-bound and memory-bound states is critical for enhancing adaptability and flexibility to diverse system and domain-specific architectures. Such capability is vital for optimizing performance across distinguished hardware configurations, such as accelerators, memory hierarchies, and cache systems. Despite that ad hoc optimization techniques, such as compressed/approximate computation, have been enabled for compute-/data-intensive computing for improved performance in distinct hardware settings, there lacks an understanding of 1) the rational behind performance improvement; 2) capability of different optimizations; 3) what optimization to respond to specific computational and memory demands. This work proposes a compression-enabled roofline model to facilitate this adaptability with data compression techniques to balance and transform between computational and memory demands. This model enables applications to adjust in response to the specific strengths and limitations of the underlying hardware and system to optimize resource utilization. The effectiveness of this approach is demonstrated with matrix multiplication kernels on different input sizes, with turning on/off various compression techniques, including 1) low-precision floating point; 2) sparse matrix formulation; and 3) compressed arrays with ZFP. By reducing memory transfer volumes and cache misses and increasing data locality and computational intensity through compression, the specific roofline model can transform between compute and memory bounds to align more efficiently with system capabilities. This advancement not only improves overall performance but also maximizes adaptability in diverse computing environments.

Naraparaju, Ramasoumya [University of Washington]↗
DOE OSTI2020

PANTHER: A Programmable Architecture for Neural Network Training Harnessing Energy-Efficient ReRAM

The wide adoption of deep neural networks has been accompanied by ever-increasing energy and performance demands due to the expensive nature of training them. Additionally, numerous special-purpose architectures have been proposed to accelerate training: both digital and hybrid digital-analog using resistive RAM (ReRAM) crossbars. ReRAM-based accelerators have demonstrated the effectiveness of ReRAM crossbars at performing matrix-vector multiplication operations that are prevalent in training. However, they still suffer from inefficiency due to the use of serial reads and writes for performing the weight gradient and update step. A few works have demonstrated the possibility of performing outer products in crossbars, which can be used to realize the weight gradient and update step without the use of serial reads and writes. However, these works have been limited to low precision operations which are not sufficient for typical training workloads. Moreover, they have been confined to a limited set of training algorithms for fully-connected layers only. To address these limitations, we propose a bit-slicing technique for enhancing the precision of ReRAM-based outer products, which is substantially different from bit-slicing for matrix-vector multiplication only. We incorporate this technique into a crossbar architecture with three variants catered to different training algorithms. To evaluate our design on different types of layers in neural networks (fully-connected, convolutional, etc.) and training algorithms, we develop PANTHER, an ISA-programmable training accelerator with compiler support. Our design can also be integrated into other accelerators in the literature to enhance their efficiency. Our evaluation shows that PANTHER achieves up to 8.02×, 54.21×, and 103× energy reductions as well as 7.16×, 4.02×, and 16× execution time reductions compared to digital accelerators, ReRAM-based accelerators, and GPUs, respectively.

42 ENGINEERING↗
DOE OSTI2022

Dissecting Tensor Cores via Microbenchmarks: Latency, Throughput and Numeric Behaviors

Tensor Cores have been an important unit to accelerate Fused Matrix Multiplication Accumulation (MMA) in all NVIDIA GPUs since Volta Architecture. To program Tensor Cores, users have to use either legacy wmma APIs or current mma APIs. Legacy wmma APIs are more easy-to-use but can only exploit limited features and power of Tensor Cores. Specifically, wmma APIs support fewer operand shapes and can not leverage the new sparse matrix multiplication feature of the newest Ampere Tensor Cores. However, the performance of current programming interface has not been well explored. Furthermore, the computation numeric behaviors of low-precision floating points (TF32, BF16, and FP16) supported by the newest Ampere Tensor Cores are also mysterious. In this paper, we explore the throughput and latency of current programming APIs. Further, we intuitively study the numeric behaviors of Tensor Cores MMA and profile the intermediate operations including multiplication, addition of inner product, and accumulation. All codes used in this work can be found in https://github.com/sunlex0717/DissectingTensorCores.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Randomized Algorithms for Scientific Computing (RASC)

Randomized algorithms have propelled advances in artificial intelligence (AI) and represent a foundational research area in advancing AI for Science. Future advancements in DOE Office of Science priority areas such as climate science, astrophysics, fusion, advanced materials, combustion, and quantum computing all require randomized algorithms for surmounting challenges of complexity, robustness, and scalability. Advances in data collection and numerical simulation have changed the dynamics of scientific research and motivate the need for randomized algorithms. For instance, advances in imaging technologies such as X-ray ptychography, electron microscopy, electron energy loss spectroscopy, or adaptive optics lattice light-sheet microscopy collect hyperspectral imaging and scattering data in terabytes, at breakneck speed enabled by state-of-the-art detectors. The data collection is exceptionally fast compared with its analysis. Likewise, advances in high-performance architectures have made exascale computing a reality and changed the economies of scientific computing in the process. Floating-point operations that create data are essentially free in comparison with data movement. Thus far, most approaches have focused on creating faster hardware. Ironically, this faster hardware has exacerbated the problem by making data still easier to create. Under such an onslaught, scientists often resort to heuristic deterministic sampling schemes (e.g., low-precision arithmetic, sampling every nth element) and sacrifice potentially valuable accuracy. Dramatically better results can be achieved via randomized algorithms, reducing the data size as much as or more than naive deterministic subsampling can achieve, while retaining the high accuracy of computing on the full data set. By randomized algorithms we mean those algorithms that employ some form of randomness in internal algorithmic decisions to accelerate time to solution, increase scalability, or improve reliability. Examples include matrix sketching for solving large-scale least-squares problems (see Figure 1) and stochastic gradient descent for training machine learning models. We are not recommending heuristic methods but rather randomized algorithms that have certificates of correctness and probabilistic guarantees of optimality and near-optimality. Such approaches can be useful beyond acceleration, for example, in understanding how to avoid measure zero worst-case scenarios that plague methods such as QR matrix factorization.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

ICP-MS For Analysis of Lithium Isotopic Ratios in Materials Highly Enriched in 7Li - CRADA 626 (Abstract)

The purpose of this project is to develop analytical methods that can be deployed for analysis of lithium isotope enrichment products. One method involves rapid, high-throughput, relatively low precision analysis for direct monitoring of enrichment processes. A second method involves slower, but more precise analysis of final products of the enrichment process. We will use commercially available analytical equipment (quadrupole base inductively coupled plasma mass spectrometry and multi-collector magnetic sector inductively coupled plasma mass spectrometry) so that the analytical methods developed can be applied broadly.

07 ISOTOPE AND RADIATION SOURCES↗
DOE OSTI2023

pnnl/lpNNPS4SPH

Low precision-based nearest neighboring particle searching algorithm.This work introduces a computationally efficient Nearest Neighbor Particles Searching (NNPS) algorithm tailored for the meshfree Smoothed Particle Hydrodynamics (SPH) method in large-deformation problems simulations. The innovation lies in the strategic use of low-precision float-point 16 (FP16) in NNPS for efficiency purpose.

Mao, Zirui↗
DOE OSTI2025

Precision Mass Measurements Reveal Low Neutron Pairing in Tin beyond 𝑁=82 and Its Impact on Stellar Nucleosynthesis

We present a study on neutron-rich tin (𝑍 =50) isotopes beyond the doubly closed shell of 𝑁 = 82 through high-precision mass measurements, including the first-ever measurements of the masses of 136 Sn, 137 Sn, and 138 Sn isotopes. These measurements enhance our understanding of the nuclear structure and astrophysical nucleosynthesis in this previously unexplored region. The new mass data are used for evaluation of the final abundances of mass numbers 𝐴 =135 and 137 in 𝑟-process network calculations. Our findings reveal a notable change in the empirical pairing gap for tin isotopes beyond the 𝑁 = 82 closed shell and a shift in the two-neutron-separation energy slope compared to heavier elements above the shell closure. A new set of ab initio calculations effectively describes these observed trends.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2020

Improving radiograph analysis throughput through transfer learning and object detection

SIGN Fracture Care International partners with surgeons in low-resource hospitals worldwide to provide access to effective orthopedic care by donating educational materials and innovatively designed surgical implants. Over two decades, SIGN’s Online Surgical Database (SOSD) has grown to contain over 500,000 medical images, with radiographs holding the majority share. One challenge in working with hospitals worldwide is that both the radiographs uploaded to the SOSD and the data entry accompanying the uploads vary in quality. To improve the accuracy of data in the SOSD, we trained a model to detect surgical implants in radiographs. We first developed a tool to automatically detect radiographs, then trained an object detection model to determine the number and placement of surgical implants visible in the radiograph. Active learning was used to generate a training set containing 2,510 radiographs with screws, nails, and plates labeled by bounding boxes. Training a model to simultaneously recognize all three classes of implants gave a low average precision (AP) for the plate class, likely due to the low number of plate instances in our training set and the large variety of surgical plates used by SIGN-partnered surgeons. Applying standard image augmentation techniques to increase the plate count in our training set did not appreciably increase the AP of plate detection. To improve plate detection, we redrew the bounding boxes to account for correlations between the screw and plate classes. Training one model to detect nails and screws and a separate model to detect plates increased the AP of plate detection by 78.8 percentage points. The AP of each class was 80.7% for screws, 93.6% for nails, and 92.6% for plates; meanwhile, the sensitivity was 92% for screws, 86% for nails, and 81% for plates. We show that object detection methods can be used to detect surgical implants in radiographs of varying quality; however, the detection ability is dependent on the type of implant, and some implants, in our case plates, must be treated differently than others. Such tools can improve the throughput of radiograph analysis, assisting physicians and surgeons with the treatment of bone fractures.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2024

Low Energy Neutron-induced Charged-particle (Z) (LENZ) instrument development with a focus on pulse shape discrimination for low-energy charged particles

To study neutron-induced charged-particle reactions with high precision, the Low Energy Neutron-induced Charged-particle (Z) instrument (LENZ) was developed at the Los Alamos Neutron Science Center. For the interest of measuring (n,p) and (n, α) reactions simultaneously, Pulse Shape Discrimination methods were investigated to identify different charged particles in the energy range of 3 - 20 MeV and improve signal-to-background ratios using Double-sided Silicon Strip Detectors and waveform digitizers. The risetime and pulse shape properties of detected charged particles were characterized for various silicon detector’s thickness with different orientations. During the post-processing of waveforms, we implemented different digital filters for effective particle identifications and improved energy- and timing- resolutions. We validated the optimized digital filters and pulse shape analyses, via measurements with 228 Th and 229 Th calibration sources, proton-induced reactions on a 7 LiF target, and neutron-induced reactions on CH 2 , Ta 2 O 5 , 58 Ni, and 6 LiF/ 59 Ni targets at the time-of-flight facility, LANSCE. In conclusion, the summary of effective thresholds and Figure of Merits on separating different charged particles is reported.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2022

Cryogenic probe for low-noise, high-frequency electronic measurements

The design and performance of a low-noise, modular cryogenic probe, which is applicable to a wide range of measurements over a broad range of working frequencies, temperatures, and magnetic fields, is presented. The design of the probe facilitates the exchange of sample holders and sample-stage amplifiers, which, combined with its characteristic low transmission and reflection loss, make this design suitable for high precision or low sensitivity measurements. The specific example of measuring the shot noise of magnetic tunnel junctions is discussed. We highlight various design characteristics chosen specifically to expand the applicability of the probe to measurement techniques such as nuclear magnetic resonance.

47 OTHER INSTRUMENTATION↗
DOE OSTI2020

Functional Photoresists for Energy Applications. Final Report

Monolithic ultralow-density porous bulk materials have recently attracted much interest due to many emerging applications in the areas of catalysis, energy storage and conversion, and thermal insulation. They are also important components of high energy density (HED) and inertial confinement fusion (ICF) targets. However, despite tremendous progress that has been made in the synthesis of porous materials, deterministic and independent control over microscopic architecture, density and composition remain key issues, and their integration in high precision devices requires cost and time-intensive mechanical machining that not only reduces reproducibility by generating debris but also limits the complexity of the 3D shapes that can be realized. In this project, we overcame these limitations by developing a universal templating capability that provides deterministic and independent control over density, composition, architecture, and macroscopic sample shape. This was achieved by developing the technology to 1) 3D print ultrahigh resolution, ultra-high precision polymeric micro-lattice templates, 2) coat these templates with the desired materials, and 3) removing the template (Fig. 1a). Atomic layer deposition (ALD) provides the atomic scale coating thickness accuracy required for precisely controlling density. While this templating approach had been demonstrated in prior work, limitations in suitable photoresists, 3D print technologies, print design, and template removal techniques did not allow the fabrication of millimeter-sized high-precision parts with sub-micron resolution. To enable this technology, we developed 1) two-photon polymerization (TPP) print designs that enable the fabrication of millimeter-sized, mechanically robust polymeric templates with sub-micron resolution and 2) a continuum level TPP printing simulation capability for additional print design guidance; 3) atomistic models to study photoresist polymerization kinetics and network topography, 4) refractive index matched polymeric and preceramic TPP photoresists, and 5) functional TPP photoresists including porous voxel structures and self-immolative polymer photoresist chemistries; and 6) damage free template removal techniques that enable the fabrication of defect-free high-precision low-density foam components. We also developed a templating approach for pure carbon microlattices with a unique tube-in-tube ligament morphology. As a test platform, we pursued the fabrication of foam liners that promise to further increase the neutron yield in indirect drive ICF experiments by improving implosion symmetry control and coupling between the laser and the deuterium-tritium fuel. This application requires fabrication and integration of a ultra-high precision, millimeter-sized, thin-walled (200-400 micrometer thick), low-density (10-30 mg/cc), high atomic number (high Z) cylindrical foam tube into the gold hohlraum of an indirect drive ICF target (Fig. 1b). While our hohlraum liner test case will mainly find application in HED and ICF experiments, the underlying science will also directly apply to previously developed nanoparticle and additive manufacturing technologies and will advance those techniques as well.

36 MATERIALS SCIENCE↗
DOE OSTI2023

A hybrid 3D/2D field response calculation for liquid argon detectors with PCB based anode plane

Liquid Argon Time Projection Chamber (LArTPC) technology is commonly utilized in neutrino detector designs. It enables detailed reconstruction of neutrino events with high spatial precision and low energy threshold. Its field response (FR) model describes the time-dependent electric currents induced in the anode-plane electrodes when ionization electrons drift nearby. An accurate and precise FR is a crucial input to LArTPC detector simulations and charge reconstruction. Established LArTPC designs have been based on parallel wire planes. It allows accurate and computationally economic two-dimensional (2D) FR models utilizing the translational symmetry along the direction of the wires. Recently, novel LArTPC designs utilize electrodes formed on printed circuit board (PCB) in the shape of strips with through holes. The translational symmetry is no longer a good approximation near the electrodes and a new FR calculation that employs regions with three dimensions (3D) has been developed. Extending the 2D models to 3D would be computationally expensive. Fortuitously, the nature of strips with through holes allows for a computationally economic approach based on the finite-difference method (FDM). In this paper, we present a new software package pochoir that calculates LArTPC field response for these new strip-based anode designs. This package combines 3D calculations in the volume near the electrodes with 2D far-field solutions to achieve fast and precise field response computation. We apply the resulting FR to simulate and reconstruct samples of cosmic-ray muons and 39 Ar decays from a Vertical Drift (VD) detector prototype operated at CERN. We find the difference between real and simulated data within 5%. Current state-of-the-art LArTPC software requires a 2D FR which we provide by averaging over one dimension and estimate that variations lost in this average are smaller than 7%.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Energy dependence of prompt fissions neutron multiplicity in the 239 Pu(n, ƒ) reaction

Accurate multiplicities of prompt fission neutrons emitted in neutron-induced fission on a large energy range are essential for fundamental and applied nuclear physics. Measuring them to high precision for radioactive fissioning nuclides remains, however, an experimental challenge. In this work, the average prompt-neutron multiplicity emitted in the 239 Pu(n, ƒ) reaction was extracted as a function of the incident-neutron energy, over the range 1-700 MeV, with a novel technique, which allowed to minimize and correct for the main sources of bias and thus achieve unprecedented precision. At low energies, our data validate for the first time the ENDF/B-VIII.0 nuclear data evaluation with an independent measurement and reduce the evaluated uncertainty by up to 60%. This work opens up the possibility of precisely measuring prompt fission neutron multiplicities on highly radioactive nuclei relevant for an essential component of energy production world-wide.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2020

Benefits of MeV-scale reconstruction capabilities in large liquid argon time projection chambers

Using truth-level Monte Carlo simulations of particle interactions in a large volume of liquid argon, we demonstrate physics capabilities enabled by reconstruction of topologically compact and isolated low-energy features, or “blips,” in large liquid argon time projection chamber (LArTPC) events. These features are mostly produced by electron products of photon interactions depositing ionization energy. The blip identification capability of the LArTPC is enabled by its unique combination of size, position resolution precision, and low energy thresholds. Here, we show that consideration of reconstructed blips in LArTPC physics analyses can result in substantial improvements in calorimetry for neutrino and new physics interactions and for final-state particles ranging in energy from the MeV to the GeV scale. Blip activity analysis is also shown to enable discrimination between interaction channels and final-state particle types. In addition to demonstrating these gains in calorimetry and discrimination, some limitations of blip reconstruction capabilities and physics outcomes are also discussed.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2022

Complementary signals of lepton flavor violation at a high-energy muon collider

A muon collider would be a powerful probe of flavor violation in new physics. There is a strong complementary case for collider measurements and precision low-energy probes of lepton flavor violation (as well as CP violation). We illustrate this by studying the collider reach in a supersymmetric scenario with flavor-violating slepton mixing. We find that the collider could discover sleptons and measure the slepton and neutralino masses with high precision, enabling event reconstruction that could cleanly separate flavor-violating new physics signals from Standard Model backgrounds. The discovery reach of a high-energy muon collider would cover a comparably large, and overlapping, range of parameter space to future μ → e conversion and electron EDM experiments, and unlike precision experiments could immediately shed light on the nature of new physics responsible for flavor violation. This complementarity strengthens the case that a muon collider could be an ideal energy-frontier laboratory in the search for physics beyond the Standard Model.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2026

Carbon-Based Quantum Information Science with Symmetry Protected Topological States (Final Report, DOE-BES award DE-SC0023105)

This research program established the scientific foundation for the rational, bottom-up design, synthesis, isolation, and investigation of symmetry-protected topological (SPT) electron spin qubits embedded in graphene nanoribbons (GNRs). The work focused on integrating atomically precise low-dimensional carbon nanostructures with emerging quantum logic architectures, providing a pathway toward scalable quantum materials for next-generation computing and sensing technologies. A central component of the program was the elucidation of fundamental relationships between real-space molecular architecture, local spin density distributions, electronic band dispersion, and energy level alignment in atomically precise GNR systems. These correlations define key operational parameters of SPT qubits and were systematically investigated to establish quantitative benchmarks against established molecular and solid-state spin qubit platforms. Attention was given to properties critical for quantum device performance, e.g. decoherence times, spectral sharpness of energy transitions, and tunable exchange interactions between spin states. The research demonstrated that these parameters can be engineered with atomic precision through scalable bottom-up synthetic strategies. Theory-guided design played a central role in identifying candidate structures hosting topologically protected spin states. Experimental validation was performed using both ensemble measurements and single-molecule characterization. In addition to advances in quantum materials synthesis, the program developed and applied spin-sensitive scanning probe microscopy techniques capable of directly probing quantum states and dynamic processes with atomic-scale spatial resolution. These capabilities enabled direct observation and characterization of quantum structures at the single-atom level. While the research activities were primarily hypothesis-driven fundamental investigations, the program adopted a comprehensive materials-by-design framework aimed at translating scientific discoveries into technological concepts compatible with scalable and intelligent manufacturing approaches.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Development of Stochastic Voronoi Lattice Structures via Two-Photon Polymerization

Low-density polymer foams of varying sizes, shapes, and densities are of specific interest to the inertial confinement fusion (ICF) program and related high-energy density plasma physics research. Historically, these foams are comprised of polystyrene or other low atomic number materials and have densities in the 30 to 300 mg/cm 3 range. However, at the lower end of this density range, these traditional polymer foams become fragile and difficult to cast and machine into the geometries needed. Recently, the need by experimentalists for materials with densities below 30 mg/cm 3 has increased. To address these needs, we are developing three-dimensional (3-D) printing techniques to create high-precision, low-density, and repeatable complex lattice structures. Using two-photon polymerization 3-D printing, we recently developed the first 5 mg/cm3 low-density lattice structure having an annular hemispherical shape. These microscale to mesoscale structures were modeled and designed using the nTopology software, specifically utilizing the "Voronoi volume lattice" and "random points in body" option blocks. All printing operations were performed using the Nanoscribe Photonic Professional GT instrument. Characterization of these 3-D structures was conducted using various microscopic and X-ray tomographic imaging techniques. Furthermore, overall printed part sizes ranged from 1 to 5 mm in diameter and were composed of lattice ligaments having thicknesses in the 3- to 5-µm range. These structures have been incorporated into ICF targets recently shot on both the University of Rochester’s Laboratory of Laser Energetics Omega laser and the National Ignition Facility.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗