Search NASA⌕ Search

SEARCH · Search NASA

Results for “near memory computing”

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

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

At least 19 records

A survey on processing-in-memory techniques: Advances and challenges

Processing-in-memory (PIM) techniques have gained much attention from computer architecture researchers, and significant research effort has been invested in exploring and developing such techniques. Increasing the research activity dedicated to improving PIM techniques will hopefully help deliver PIM’s promise to solve or significantly reduce memory access bottleneck problems for memory-intensive applications. We also believe it is imperative to track the advances made in PIM research to identify open challenges and enable the research community to make informed decisions and adjust future research directions. In this survey, we analyze recent studies that explored PIM techniques, summarize the advances made, compare recent PIM architectures, and identify target application domains and suitable memory technologies. We also discuss proposals that address unresolved issues of PIM designs (e.g., address translation/mapping of operands, workload analysis to identify application segments that can be accelerated with PIM, OS/runtime support, and coherency issues that must be resolved to incorporate PIM). We believe this work can serve as a useful reference for researchers exploring PIM techniques.

97 MATHEMATICS AND COMPUTING↗

A Survey on the Expanding Scope and Interdisciplinary Opportunities for Processing-in-Memory Techniques

Processing-in-Memory (PIM) is emerging as a practical path to overcome the limitations of traditional von Neumann architectures. At its core, PIM systems implement computing primitives such as logic operations and multiply-accumulate acceleration through compute-in-memory, near-memory processing, or hybrid designs. The role of memory cells varies widely across technologies, acting as inputs, outputs, or analog accumulators through bit-lines and sense amplifiers. This diversity creates trade-offs in precision, bandwidth, latency, and programmability, making it difficult to build a unified understanding on the progress of the field. In this survey, we organize recent advances of PIM into three areas. First, we discuss the progress on the architectural optimizations of PIM and its integration with both DRAM and emerging non-volatile memories. Second, we examine how PIM is being used to accelerate key computing domains, including generative AI workloads and high-performance kernels, along with new approaches. Third, we highlight the growing adoption of PIM in computational sciences, where it is being applied to solve interdisciplinary problems such as genome analysis, mRNA quantification, mass spectrometry, quantum circuit simulation, wave modeling, and secure computation. Finally, we synthesize the major challenges that continue to slow PIM adoption, including manufacturing constraints, power delivery, thermal reliability, data consistency, runtime and memory-management coordination, and the difficulty of building portable software abstractions without sacrificing commercial viability. This work provides an updated, structured perspective on PIM’s potential across computing and computational sciences and the barriers that must be solved for it to reach its full impact.

Asifuzzaman, Kazi [Oak Ridge National Laboratory (↗

Nitrogen Vacancies Induce Fatigue in Ferroelectric Al 0.93 B 0.07 N

Wurtzite ferroelectrics (e.g., Al 0.93 B 0.07 N) are being explored for high-temperature and emerging near-, or in-compute, memory architectures due to the material advantages offered by their large remanent polarization and robust chemical stability. Despite these advantages, current Al0.93B0.07N devices do not have sufficient endurance lifetime to meet roadmap targets. To identify the defects responsible for this limited endurance, a combination of electronic measurements and optical spectroscopies characterized the evolution of defect states within Al 0.93 B 0.07 N with cycling. Ultrathin (∼10 nm) metal contacts were used to optically probe regions subject to ferroelectric switching; photoluminescence spectroscopy identified the emergence of a transition near 2.1 eV whose intensity scaled with the non-switching polarization quantified via positive-up negative-down (PUND) measurements. Accompanying thermally stimulated depolarization current (TSDC) and modulus spectroscopy measurements also observed the strengthening of a state near 2.1 eV. The origin of this feature is ascribed to transitions between a nitrogen vacancy and another defect deeper in the bandgap. Recognizing that the impurity concentration is largely fixed, strengthening of this transition indicates an increase in the number of nitrogen vacancies. Switching, therefore, creates vacancies in Al 0.93 B 0.07 N likely due to hot-atom damage induced by the aggressive fields necessary to switch wurtzite materials that ultimately limits endurance.

36 MATERIALS SCIENCE↗

Filament-Free Bulk Resistive Memory Enables Deterministic Analogue Switching

Digital computing is nearing its physical limits as computing needs and energy consumption rapidly increase. Analogue-memory-based neuromorphic computing can be orders of magnitude more energy efficient at data-intensive tasks like deep neural networks, but has been limited by the inaccurate and unpredictable switching of analogue resistive memory. Filamentary resistive random access memory (RRAM) suffers from stochastic switching due to the random kinetic motion of discrete defects in the nanometer-sized filament. Here, this stochasticity is overcome by incorporating a solid electrolyte interlayer, in this case, yttria-stabilized zirconia (YSZ), toward eliminating filaments. Filament-free, bulk-RRAM cells instead store analogue states using the bulk point defect concentration, yielding predictable switching because the statistical ensemble behavior of oxygen vacancy defects is deterministic even when individual defects are stochastic. Both experiments and modeling show bulk-RRAM devices using TiO2-X switching layers and YSZ electrolytes yield deterministic and linear analogue switching for efficient inference and training. Bulk-RRAM solves many outstanding issues with memristor unpredictability that have inhibited commercialization, and can, therefore, enable unprecedented new applications for energy-efficient neuromorphic computing. Beyond RRAM, this work shows how harnessing bulk point defects in ionic materials can be used to engineer deterministic nanoelectronic materials and devices.

36 MATERIALS SCIENCE↗

A Survey of Architectural Techniques for Near-Threshold Computing

Energy efficiency has now become the primary obstacle in scaling the performance of all classes of computing systems. In low-voltage computing and specifically, near-threshold voltage computing (NTC), which involves operating the transistor very close to and yet above its threshold voltage, holds the promise of providing many-fold improvement in energy efficiency. However, use of NTC also presents several challenges such as increased parametric variation, failure rate and performance loss etc. Our paper surveys several re- cent techniques which aim to offset these challenges for fully leveraging the potential of NTC. By classifying these techniques along several dimensions, we also highlight their similarities and differences. Ultimately, we hope that this paper will provide insights into state-of-art NTC techniques to researchers and system-designers and inspire further research in this field.

97 MATHEMATICS AND COMPUTING↗

Physics-informed latent neural operator for real-time predictions of time-dependent parametric PDEs

Deep operator network (DeepONet) has shown significant promise as surrogate models for systems governed by partial differential equations (PDEs), enabling accurate mappings between infinite-dimensional function spaces. However, when applied to systems with high-dimensional input-output mappings arising from large numbers of spatial and temporal collocation points, these models often require heavily overparameterized networks, leading to long training times. Latent DeepONet addresses some of these challenges by introducing a two-step approach: first learning a reduced latent space using a separate model, followed by operator learning within this latent space. While efficient, this method is inherently data-driven and lacks mechanisms for incorporating physical laws, limiting its robustness and generalizability in data-scarce settings. Here, in this work, we propose PI-Latent-NO, a physics-informed latent neural operator framework that integrates governing physics directly into the learning process. Our architecture features two coupled DeepONets trained end-to-end: a Latent-DeepONet that learns a low-dimensional representation of the solution, and a Reconstruction-DeepONet that maps this latent representation back to the physical space. By embedding PDE constraints into the training via automatic differentiation, our method eliminates the need for labeled training data and ensures physics-consistent predictions. The proposed framework is both memory and compute-efficient, exhibiting near-constant scaling with problem size and demonstrating significant speedups over traditional physics-informed operator models. We validate our approach on a range of parametric PDEs, showcasing its accuracy, scalability, and suitability for real-time prediction in complex physical systems.

Latent representations↗

Distributed seams for gigapixel panoramas

Gigapixel panoramas are an increasingly popular digital image application. They are often created as a mosaic of many smaller images. The mosaic acquisition can take many hours causing the individual images to differ in exposure and lighting conditions. A blending operation is often necessary to give the appearance of a seamless image. The blending quality depends on the magnitude of discontinuity along the image boundaries. Often, new boundaries, or seams, are first computed that minimize this transition. Current techniques based on multi-labeling Graph Cuts are too slow and memory intensive for gigapixel sized panoramas. In this paper, we present a parallel, out-of-core seam computing technique that is fast, has small memory footprint, and is capable of running efficiently on different types of parallel systems. Its maximum memory usage is configurable, in the form of a cache, which can improve performance by reducing redundant disk I/O and computations. It shows near-perfect scaling on symmetric multiprocessing systems and good scaling on clusters and distributed shared memory systems. Lastly, our technique improves the time required to compute seams for gigapixel imagery from many hours (or even days) to just a few minutes, while still producing boundaries with energy that is on-par with Graph Cuts.

97 MATHEMATICS AND COMPUTING↗

Adaptive LES Methodology for Turbulent Flow Simulations

Although turbulent flows are common in the world around us, a solution to the fundamental equations that govern turbulence still eludes the scientific community. Turbulence has often been called one of the last unsolved problem in classical physics, yet it is clear that the need to accurately predict the effect of turbulent flows impacts virtually every field of science and engineering. As an example, a critical step in making modern computational tools useful in designing aircraft is to be able to accurately predict the lift, drag, and other aerodynamic characteristics in numerical simulations in a reasonable amount of time. Simulations that take months to years to complete are much less useful to the design cycle. Much work has been done toward this goal (Lee-Rausch et al. 2003, Jameson 2003) and as cost effective accurate tools for simulating turbulent flows evolve, we will all benefit from new scientific and engineering breakthroughs. The problem of simulating high Reynolds number (Re) turbulent flows of engineering and scientific interest would have been solved with the advent of Direct Numerical Simulation (DNS) techniques if unlimited computing power, memory, and time could be applied to each particular problem. Yet, given the current and near future computational resources that exist and a reasonable limit on the amount of time an engineer or scientist can wait for a result, the DNS technique will not be useful for more than 'unit' problems for the foreseeable future (Moin & Kim 1997, Jimenez & Moin 1991). The high computational cost for the DNS of three dimensional turbulent flows results from the fact that they have eddies of significant energy in a range of scales from the characteristic length scale of the flow all the way down to the Kolmogorov length scale. The actual cost of doing a three dimensional DNS scales as Re{sup 9/4} due to the large disparity in scales that need to be fully resolved. State-of-the-art DNS calculations of isotropic turbulence have recently been completed at the Japanese Earth Simulator (Yokokawa et al. 2002, Kaneda et al. 2003) using a resolution of 40963 (approximately 10{sup 11}) grid points with a Taylor-scale Reynolds number of 1217 (Re {approx} 10{sup 6}). Impressive as these calculations are, performed on one of the world's fastest super computers, more brute computational power would be needed to simulate the flow over the fuselage of a commercial aircraft at cruising speed. Such a calculation would require on the order of 10{sup 16} grid points and would have a Reynolds number in the range of 108. Such a calculation would take several thousand years to simulate one minute of flight time on today's fastest super computers (Moin & Kim 1997). Even using state-of-the-art zonal approaches, which allow DNS calculations that resolve the necessary range of scales within predefined 'zones' in the flow domain, this calculation would take far too long for the result to be of engineering interest when it is finally obtained. Since computing power, memory, and time are all scarce resources, the problem of simulating turbulent flows has become one of how to abstract or simplify the complexity of the physics represented in the full Navier-Stokes (NS) equations in such a way that the 'important' physics of the problem is captured at a lower cost. To do this, a portion of the modes of the turbulent flow field needs to be approximated by a low order model that is cheaper than the full NS calculation. This model can then be used along with a numerical simulation of the 'important' modes of the problem that cannot be well represented by the model. The decision of what part of the physics to model and what kind of model to use has to be based on what physical properties are considered 'important' for the problem. It should be noted that 'nothing is free', so any use of a low order model will by definition lose some information about the original flow.

42 ENGINEERING↗

PDG: A Composite Method Based on the Resolution of the Identity

The Gaussian-3 (G3) composite approach for thermochemical properties is revisited in light of the enhanced computational efficiency and reduced memory costs by applying the resolution-of-the-identity (RI) approximation for two-electron repulsion integrals (ERIs) to the computationally demanding component methods in the G3 model: the energy and gradient computations via the second-order Møller–Plesset perturbation theory (MP2) and the energy computations using the coupled-cluster singles–doubles method augmented with noniterative triples corrections [CCSD(T)]. Efficient implementation of the RI-based methods is achieved by employing a hybrid distributed/shared memory model based on MPI and OpenMP. The new variant of the G3 composite approach based on the RI approximation is termed the RI-G3 scheme, or alternatively the PDG method. The accuracy of the new RI-G3/PDG scheme is compared to the “standard” G3 composite approach that employs the memory-expensive four-center ERIs in the MP2 and CCSD(T) calculations. Taking the computation of the heats of formation of the closed-shell molecules in the G3/99 test set as a test case, it is demonstrated that the RI approximation introduces negligible changes to the mean absolute errors relative to the standard G3 model (less than 0.1 kcal/mol), while the standard deviations remain unaltered. Furthermore, the efficiency and memory requirements for the RI-MP2 and RI-CCSD(T) methods are compared to the standard MP2 and CCSD(T) approaches, respectively. The hybrid MPI/OpenMP-based RI-MP2 energy plus gradient computation is found to attain a 7.5× speedup over the standard MP2 calculations. For the most demanding CCSD(T) calculations, the application of the RI approximation is found to nearly halve the memory demand, confer about a 4–5× speedup for the CCSD iterations, and reduce the computational time for the compute-intensive triples correction step by several hours.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Lightweight storage and overlay networks for fault tolerance.

The next generation of capability-class, massively parallel processing (MPP) systems is expected to have hundreds of thousands to millions of processors, In such environments, it is critical to have fault-tolerance mechanisms, including checkpoint/restart, that scale with the size of applications and the percentage of the system on which the applications execute. For application-driven, periodic checkpoint operations, the state-of-the-art does not provide a scalable solution. For example, on today's massive-scale systems that execute applications which consume most of the memory of the employed compute nodes, checkpoint operations generate I/O that consumes nearly 80% of the total I/O usage. Motivated by this observation, this project aims to improve I/O performance for application-directed checkpoints through the use of lightweight storage architectures and overlay networks. Lightweight storage provide direct access to underlying storage devices. Overlay networks provide caching and processing capabilities in the compute-node fabric. The combination has potential to signifcantly reduce I/O overhead for large-scale applications. This report describes our combined efforts to model and understand overheads for application-directed checkpoints, as well as implementation and performance analysis of a checkpoint service that uses available compute nodes as a network cache for checkpoint operations.

99 GENERAL AND MISCELLANEOUS↗

SEISMIC SIMULATIONS USING PARALLEL COMPUTING AND THREE-DIMENSIONAL EARTH MODELS TO IMPROVE NUCLEAR EXPLOSION PHENOMENOLOGY AND MONITORING

The development of accurate numerical methods to simulate wave propagation in three-dimensional (3D) earth models and advances in computational power offer exciting possibilities for modeling the motions excited by underground nuclear explosions. This presentation will describe recent work to use new numerical techniques and parallel computing to model earthquakes and underground explosions to improve understanding of the wave excitation at the source and path-propagation effects. Firstly, we are using the spectral element method (SEM, SPECFEM3D code of Komatitsch and Tromp, 2002) to model earthquakes and explosions at regional distances using available 3D models. SPECFEM3D simulates anelastic wave propagation in fully 3D earth models in spherical geometry with the ability to account for free surface topography, anisotropy, ellipticity, rotation and gravity. Results show in many cases that 3D models are able to reproduce features of the observed seismograms that arise from path-propagation effects (e.g. enhanced surface wave dispersion, refraction, amplitude variations from focusing and defocusing, tangential component energy from isotropic sources). We are currently investigating the ability of different 3D models to predict path-specific seismograms as a function of frequency. A number of models developed using a variety of methodologies are available for testing. These include the WENA/Unified model of Eurasia (e.g. Pasyanos et al 2004), the global CUB 2.0 model (Shapiro and Ritzwoller, 2002), the partitioned waveform model for the Mediterranean (van der Lee et al., 2007) and stochastic models of the Yellow Sea Korean Peninsula region (Pasyanos et al., 2006). Secondly, we are extending our Cartesian anelastic finite difference code (WPP of Nilsson et al., 2007) to model the effects of free-surface topography. WPP models anelastic wave propagation in fully 3D earth models using mesh refinement to increase computational speed and improve memory efficiency. Thirdly, we are modeling non-linear near-source shock wave propagation with GEODYN, an Eulerian Godunov finite-difference code (Antoun et al., 2001). This code accounts for shock wave propagation and a variety of effects including cavity formation, rock fracture and plastic deformation. We are exploring the coupling of GEODYN to WPP to propagate motions from the near-source (non-linear) region to the (linear anelastic) region where seismic observations are made at local, regional and teleseismic distances. This effort has just begun and we show preliminary results in this paper (with more to follow in our poster). These simulation tools are supported by massively parallel computers operated by Livermore Computing.

58 GEOSCIENCES↗

FPGA-based HPC accelerators: An evaluation on performance and energy efficiency

Hardware specialization is a promising direction for the future of digital computing. Reconfigurable technologies enable hardware specialization with modest non-recurring engineering cost, but their performance and energy efficiency compared to state-of-the-art processor architectures remain an open question. In this article, we use FPGAs to evaluate the benefits of building specialized hardware for numerical kernels found in scientific applications. In order to properly evaluate performance, we not only compare Intel Arria 10 and Xilinx U280 performance against Intel Xeon, Intel Xeon Phi, and NVIDIA V100 GPUs, but we also extend the Empirical Roofline Toolkit (ERT) to FPGAs in order to assess our results in terms of the Roofline model. We show design optimization and tuning techniques for peak FPGA performance at reasonable hardware usage and power consumption. As FPGA peak performance is known to be far less than that of a GPU, we also benchmark the energy efficiency of each platform for the scientific kernels comparing against microbenchmark and technological limits. Results show that while FPGAs struggle to compete in absolute terms with GPUs on memory- and compute-intensive kernels, they require far less power and can deliver nearly the same energy efficiency.

97 MATHEMATICS AND COMPUTING↗

Proton Tunable Analog Transistor for Low Power Computing

This project was broadly motivated by the need for new hardware that can process information such as images and sounds right at the point of where the information is sensed (e.g. edge computing). The project was further motivated by recent discoveries by group demonstrating that while certain organic polymer blends can be used to fabricate elements of such hardware, the need to mix ionic and electronic conducting phases imposed limits on performance, dimensional scalability and the degree of fundamental understanding of how such devices operated. As an alternative to blended polymers containing distinct ionic and electronic conducting phases, in this LDRD project we have discovered that a family of mixed valence coordination compounds called Prussian blue analogue (PBAs), with an open framework structure and ability to conduct both ionic and electronic charge, can be used for inkjet-printed flexible artificial synapses that reversibly switch conductance by more than four orders of magnitude based on electrochemically tunable oxidation state. Retention of programmed states is improved by nearly two orders of magnitude compared to the extensively studied organic polymers, thus enabling in-memory compute and avoiding energy costly off-chip access during training. We demonstrate dopamine detection using PBA synapses and biocompatibility with living neurons, evoking prospective application for brain - computer interfacing. By application of electron transfer theory to in-situ spectroscopic probing of intervalence charge transfer, we elucidate a switching mechanism whereby the degree of mixed valency between N-coordinated Ru sites controls the carrier concentration and mobility, as supported by density functional theory (DFT) .

97 MATHEMATICS AND COMPUTING↗

ALD-Derived WO 3– x Leads to Nearly Wake-Up-Free Ferroelectric Hf 0.5 Zr 0.5 O 2 at Elevated Temperatures

Breaking the memory wall in advanced computing architectures will require complex 3D integration of emerging memory materials such as ferroelectrics─either within the back-end-of-line (BEOL) of CMOS front-end processes or through advanced 3D packaging technologies. Achieving this integration demands that memory materials exhibit high thermal resilience, with the capability to operate reliably at elevated temperatures, such as 125°C, due to the substantial heat generated by front-end transistors. However, silicon-compatible HfO 2 -based ferroelectrics tend to exhibit antiferroelectric-like behavior in this temperature range, accompanied by a more pronounced wake-up effect, posing significant challenges to their thermal reliability. Here, we report that by introducing a thin tungsten oxide (WO 3–x ) layer─known as an oxygen reservoir─and carefully tuning its oxygen content, ultrathin Hf 0.5 Zr 0.5 O 2 (5 nm) films can be made robust against the ferroelectric-to-antiferroelectric transition at elevated temperatures. This approach not only minimizes polarization loss in the pristine state but also effectively suppresses the wake-up effect, reducing the required wake-up cycles from 10 5 to only 10 at 125°C, a qualifying temperature for back-end memory integrated with front-end logic, as defined by the JEDEC standard. First-principles density functional theory (DFT) calculations reveal that WO 3 enhances the stability of the ferroelectric orthorhombic phase (o-phase) at elevated temperatures by increasing the tetragonal-to-orthorhombic phase energy gap and promoting favorable phonon mode evolution, thereby supporting o-phase formation under both thermodynamic and kinetic constraints.

36 MATERIALS SCIENCE↗

Polaron-induced metal-to-insulator transition in vanadium oxides from density functional theory calculations

Vanadium oxides have been extensively studied as phase-change memory units in artificial synapses for neuromorphic computing due to their metal-insulator transitions (MIT) at or near room temperature. Recently, injection of charge carriers into vanadium oxides, e.g., via optically via a heterostructure, has been proposed as an alternative switching mechanism and also potentially as a means to tune the MIT temperature. In this study, we explore the formation of small polarons in the low temperature (LT) insulating phases for V 3 O 5 ,VO 2 , and V 2 O 3 , and the barriers to their migration using density functional theory calculations. We find that V 3 O 5 exhibits very low hole and electron polaron migration barriers (<100 meV) compared to V 2 O 3 and VO 2 , leading to much higher estimated polaronic conductivity. We also link the relative migration barriers to the amount of distortion that has to travel when the polaron migrate from one site to another. Polarons in V 3 O 5 also have smaller binding energies to vanadium and oxygen vacancy defects. Furthermore, these results suggest that the triggering of the MIT via injection of charge carriers are due to the formation of small polarons that can migrate rapidly through the crystal.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗