Search NASA⌕ Search

SEARCH · Search NASA

Results for “COMPUTER ARCHITECTURE”

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 289 records · Page 16
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

PaRSEC: Scalability, flexibility, and hybrid architecture support for task-based applications in ECP

This paper highlights the most significant enhancements made to PaRSEC, a scalable task-based runtime system designed for hybrid machines, during the Exascale Computing Project (ECP). The enhancements focus on expanding the capabilities of PaRSEC to address the evolving landscape of parallel computing. Notable achievements include the integration of support for three major types of accelerators (NVIDIA, AMD, and Intel GPUs), the refinement and increased flexibility of the communication subsystem, and the introduction of new programming interfaces tailored for irregular applications. Additionally, the project resulted in the development of powerful debugging and performance analysis tools aimed at assisting users in understanding and optimizing their applications. We present a comprehensive demonstration of these advancements through a series of benchmarks and applications within ECP and beyond, thereby showcasing the enhanced capabilities of PaRSEC across the diverse architectures within the ECP, providing valuable insights into the runtime system’s adaptability and performance across varied computing environments.

Bouteiller, Aurelien↗
DOE OSTI2026

Pre-training Vision Models for the Classification of Alerts from Wide-field Time-domain Surveys

Modern wide-field time-domain surveys facilitate the study of transient, variable and moving phenomena by conducting image differencing and relaying alerts to their communities. Machine learning tools have been used on data from these surveys and their precursors for more than a decade, and convolutional neural networks (CNNs), which make predictions directly from input images, saw particularly broad adoption through the 2010s. Since then, continually rapid advances in computer vision have transformed the standard practices around using such models. It is now commonplace to use standardized architectures pre-trained on large corpora of everyday images (e.g., ImageNet). In contrast, time-domain astronomy studies still typically design custom CNN architectures and train them from scratch. Here, we explore the effects of adopting various pre-training regimens and standardized model architectures on the performance of alert classification. We find that the resulting models match or outperform a custom, specialized CNN like what is typically used for filtering alerts. Moreover, our results show that pre-training on galaxy images from Galaxy Zoo tends to yield better performance than pre-training on ImageNet or training from scratch. We observe that the design of standardized architectures are much better optimized than the custom CNN baseline, requiring significantly less time and memory for inference despite having more trainable parameters. On the eve of the Legacy Survey of Space and Time and other image-differencing surveys, these findings advocate for a paradigm shift in the creation of vision models for alerts, demonstrating that greater performance and efficiency, in time and in data, can be achieved by adopting the latest practices from the computer vision field.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Optoelectronic polymer memristors with dynamic control for power-efficient in-sensor edge computing

Abstract As the demand for edge platforms in artificial intelligence increases, including mobile devices and security applications, the surge in data influx into edge devices often triggers interference and suboptimal decision-making. There is a pressing need for solutions emphasizing low power consumption and cost-effectiveness. In-sensor computing systems employing memristors face challenges in optimizing energy efficiency and streamlining manufacturing due to the necessity for multiple physical processing components. Here, we introduce low-power organic optoelectronic memristors with synergistic optical and mV-level electrical tunable operation for a dynamic “control-on-demand” architecture. Integrating signal sensing, featuring, and processing within the same memristors enables the realization of each in-sensor analogue reservoir computing module, and minimizes circuit integration complexity. The system achieves 97.15% fingerprint recognition accuracy while maintaining a minimal reservoir size and ultra-low energy consumption. Furthermore, we leverage wafer-scale solution techniques and flexible substrates for optimal memristor fabrication. By centralizing core functionalities on the same in-sensor platform, we propose a resilient and adaptable framework for energy-efficient and economical edge computing.

Optics↗
DOE OSTI2025

DeepHyper: A Python Package for Massively Parallel Hyperparameter Optimization in Machine Learning

Machine learning models are increasingly applied across scientific disciplines, yet their effectiveness often hinges on heuristic decisions—such as data transformations, training strategies, and model architectures—that are not learned by the models themselves. Automating the selection of these heuristics and analyzing their sensitivity is crucial for building robust and efficient learning workflows. DeepHyper addresses this challenge by democratizing hyperparameter optimization, providing accessible tools to streamline and enhance machine learning workflows from a laptop to the largest supercomputer in the world. Building on top of hyperparameter optimization, it unlocks new capabilities around ensembles of models for improved accuracy and uncertainty quantification. All of these organized around efficient parallel computing.

ensemble↗
DOE OSTI2025

Micropolar Elastoplasticity Using a Fast Fourier Transform‐Based Solver

ABSTRACT This work presents a micromechanical spectral formulation for obtaining the full‐field and homogenized response of elastoplastic micropolar composites. A closed‐form radial‐return mapping is derived from thermodynamics‐based micropolar elastoplastic constitutive equations to determine the increment of plastic strain necessary to return the generalized stress state to the yield surface, and the algorithm implementation is verified using the method of numerically manufactured solutions. Then, size‐dependent material response and micro‐plasticity are shown as features that may be efficiently simulated in this micropolar elastoplastic framework. The computational efficiency of the formulation enables the generation of large datasets in reasonable computing times.

42 ENGINEERING↗
DOE OSTI2025

SchrödingerNet: A Universal Neural Network Solver for the Schrödinger Equation

Recent advances in machine learning have facilitated numerically accurate solution of the electronic Schrödinger equation (SE) by integrating various neural network (NN)-based wave function ansatzes with variational Monte Carlo methods. Nevertheless, such NN-based methods are all based on the Born–Oppenheimer approximation (BOA) and require computationally expensive training for each nuclear configuration. In this work, we propose a novel NN architecture, SchrödingerNet, to solve the full electronic-nuclear SE by defining a loss function designed to equalize local energies across the system. This approach is based on a translationally, rotationally and permutationally symmetry-adapted total wave function ansatz that includes both nuclear and electronic coordinates. Furthermore, this strategy not only allows for an efficient and accurate generation of a continuous potential energy surface at any geometry within the well-sampled nuclear configuration space, but also incorporates non-BOA corrections, through a single training process. Comparison with benchmarks of atomic and small molecular systems demonstrates its accuracy and efficiency.

Chemical calculations↗
DOE OSTI2026

Magnon confinement in epitaxial antiferromagnetic oxide heterostructures.

Magnons, the quanta of spin waves, have been extensively studied in a range of materials for spintronics, particularly for non-volatile logic-in-memory devices. Controlling magnons in conventional antiferromagnets and harnessing them in practical applications, however, remains a challenge. Here, we demonstrate highly efficient magnon transport in a LaFeO3/ BiFeO3/ LaFeO3 all-antiferromagnetic system, which can be controlled electrically, making it highly desirable for energy-efficient computation. Leveraging spin-orbit-driven spin-charge transduction, we demonstrate that this material architecture permits magnon confinement in ultrathin antiferromagnets, enhancing the output voltage generated by magnon transport by several orders of magnitude, which provides a pathway to enable magnetoelectric memory and logic functionalities. Additionally, the non-volatility of output voltage enables ultralowpower logic-in-memory processing, where magnonic devices can be efficiently reconfigured via electrically controlled magnon spin currents within magnetoelectric channels.

Husain, Sajid↗
DOE OSTI2025

Empirically-calibrated H100 node power models for accurate AI training energy estimation

Accurately quantifying the energy use of artificial intelligence (AI) training is critical for infrastructure planning, carbon accounting, and sustainable data center operation, but few studies have directly measured the power consumption of production workloads on contemporary hardware. By combining empirical measurements from Brookhaven National Laboratory during AI training on 8-graphics-processing-unit H100 systems with open-source benchmarking data, we develop statistical models relating computational intensity to node-level power consumption. We measure the gap between manufacturer-rated thermal design power (TDP) and actual power demand during AI training. Our analysis reveals that even computationally intensive workloads operate at only 76% of the 10.2 kW TDP rating. Our architecture-specific model, calibrated to floating-point operations, predicts energy consumption with 11.4% mean absolute percentage error, significantly outperforming TDP-based approaches (27%–37% error). We identified distinct power signatures between transformer and convolutional neural network architectures, with transformers showing characteristic fluctuations that may impact grid stability. These results provide a measurement-grounded basis for improving AI training energy estimates, enabling more reliable infrastructure sizing, cost projections, and environmental impact assessments.

Newkirk, Alex C↗
DOE OSTI2024

Machine Learning-Enabled Image Classification for Automated Electron Microscopy

Abstract Traditionally, materials discovery has been driven more by evidence and intuition than by systematic design. However, the advent of “big data” and an exponential increase in computational power have reshaped the landscape. Today, we use simulations, artificial intelligence (AI), and machine learning (ML) to predict materials characteristics, which dramatically accelerates the discovery of novel materials. For instance, combinatorial megalibraries, where millions of distinct nanoparticles are created on a single chip, have spurred the need for automated characterization tools. This paper presents an ML model specifically developed to perform real-time binary classification of grayscale high-angle annular dark-field images of nanoparticles sourced from these megalibraries. Given the high costs associated with downstream processing errors, a primary requirement for our model was to minimize false positives while maintaining efficacy on unseen images. We elaborate on the computational challenges and our solutions, including managing memory constraints, optimizing training time, and utilizing Neural Architecture Search tools. The final model outperformed our expectations, achieving over 95% precision and a weighted F-score of more than 90% on our test data set. This paper discusses the development, challenges, and successful outcomes of this significant advancement in the application of AI and ML to materials discovery.

Materials Science↗
DOE OSTI2024

CommBench: Micro-Benchmarking Hierarchical Networks with Multi-GPU, Multi-NIC Nodes

Modern high-performance computing systems have multiple GPUs and network interface cards (NICs) per node. The resulting network architectures have multilevel hierarchies of subnetworks with different interconnect and software technologies. These systems offer multiple vendor-provided communication capabilities and library implementations (IPC, MPI, NCCL, RCCL, OneCCL) with APIs providing varying levels of performance across the different levels. Understanding this performance is currently difficult because of the wide range of architectures and programming models (CUDA, HIP, OneAPI). We present CommBench, a library with cross-system portability and a high-level API that enables developers to easily build microbenchmarks relevant to their use cases and gain insight into the performance (bandwidth & latency) of multiple implementation libraries on different networks. We demonstrate CommBench with three sets of microbenchmarks that profile the performance of six systems. Our experimental results reveal the effect of multiple NICs on optimizing the bandwidth across nodes and also present the performance characteristics of four available communication libraries within and across nodes of NVIDIA, AMD, and Intel GPU networks.

Hidayetoglu, Mert↗
DOE OSTI2026

The landscape of regulatory element evolution in a C4 perennial grass

Gene regulatory evolution is a well-known source of phenotypic diversity and adaptive evolution. Although cis-regulatory elements (CREs) play a vital role in gene expression evolution, the molecular evolution of CREs remains mostly unknown due to the difficulty in identifying and characterizing these functional elements. Comparative genomic analyses of noncoding DNA can be leveraged to identify conserved noncoding sequences (CNS), many of which may harbor functional CREs conserved by purifying selection. However, purely computational inference of CREs from putative CNS can be erroneous due to the complex genomic architecture in plants. One promising experimental approach to identify CREs is by profiling accessible chromatin regions (ACRs) that are often associated with the location of CREs. In this study, we use comparative genomics along with the profiling of ACRs to study the molecular evolution of putative functional noncoding regulatory regions in Panicoid grasses. We identified sets of CNS that varied in relationship to the degree of evolutionary divergence among the studied taxa, including identifying core-Panicoid-CNS. We augmented this analysis by profiling ACRs in Panicum hallii ecotypes using ATAC-seq. ACRs had low SNP density at the summit, harbored a high frequency of core-Panicoid-CNS, and were enriched with expression QTL. These data help to annotate the P. hallii genome for putative functional elements and suggest that a large proportion of these ACRs are evolving under purifying selection. Turnover in CNS and ACR between ecotypes of P. hallii identifies a small set of putatively divergent CREs that may underlie differences in gene regulation between genotypes from inland and coastal habitats. In summary, we profiled ACRs in Panicoid grasses and integrated this data with our putative CNS prediction framework, which provides unique insight into patterns of polymorphism and divergence in CREs in C4 perennial grasses.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Thermo-Fluid Modeling Framework for Supercomputing Digital Twins: Part 2, Automated Cooling Models

The development of digital twins for the purpose of improving the energy efficiency of supercomputing facilities is a non-trivial endeavor that is complicated by the difficulty of creating physics-based thermo-fluid cooling system models (CSMs). Within ExaDigit---an open-source framework for liquid-cooled supercomputing digital twins---a thermo-fluid modeling framework is being developed. This effort has been segmented into two with two companion papers describing each portion of the overall effort. Part 1 focuses on the development of a cooling system library in Dymola for the Frontier supercomputer at Oak Ridge National Laboratory {\cite{Kumar2024}. Part 2, this paper, describes an effort to create a template-based auto-generation methodology for CSMs, called \textit{AutoCSM}. In this paper, an overview of the initial AutoCSM architecture and workflow is provided, along with a practical example using the Oak Ridge Leadership Computing Facility's (OLCF) Frontier supercomputer CSM. AutoCSM will (1) improve ExaDigiT's user accessibility by providing a flexible workflow for modularizing the creation of the CSM system and control logic, (2) decrease the development time of CSMs, and (3) standardize the method for incorporating CSMs into the ExaDigiT framework.

Greenwood, Scott↗
DOE OSTI2026

There and Back Again: Reimagining Cryogenic Cooling for Scalable Arrays of Dilution Refrigerators for future Quantum Datacenters

While pulse tube cryocoolers enabled the rapid expansion of dilution refrigerator technology over the past two decades, the transition to large-scale quantum systems is now driving a reassessment of the DR’s higher-temperature-stage cooling strategies and how these systems can be effectively scaled in a modular way. Quasi-wet architectures based on centralized cryoplants and forced-flow helium distribution offer compelling advantages in energy efficiency, operational cost, and scalability. With appropriate redundancy, standardized interfaces, and optimized distribution system designs, these architectures will provide a practical and robust path forward for the next generation of quantum computing infrastructure.

Hansen, B. [Fermilab]↗
DOE OSTI2025

BULKI-Store v0.3.2

BULKI-Store is a distributed object storage system optimized for high-performance computing environments. Built with a Rust core and Python bindings, it efficiently manages scientific and machine learning datasets across HPC clusters. The system employs a client-server architecture with MPI integration, enabling seamless scaling on supercomputers like Perlmutter. BULKI-Store's object-oriented approach provides intuitive data organization with rich metadata support, contrasting with traditional file-based solutions. Key optimizations include selective checkpoint loading, unified checkpoint files, and object chunking for large data transfers. For machine learning workloads, BULKI-Store offers advantages through fine-grained access patterns, dynamic data sharing between training instances, and reduced memory pressure. Memory management features include strategic Python GC calls, minimized data copies, and batch processing capabilities. The system leverages Rayon's thread pool for asynchronous data prefetching and supports multiple CPU architectures (ARM64, x86, AMD, RISC-V). By combining performance optimizations with developer-friendly APIs, BULKI-Store addresses the complex data management challenges of modern HPC applications while maintaining compatibility across heterogeneous computing environments.

Zhang, Wei [Lawrence Berkeley National Laboratory ↗
DOE OSTI2024

Homologous recombination shapes the architecture and evolution of bacterial genomes

Homologous recombination is a key evolutionary force that varies considerably across bacterial species. However, how the landscape of homologous recombination varies across genes and within individual genomes has only been studied in a few species. Here, we used Approximate Bayesian Computation to estimate the recombination rate along the genomes of 145 bacterial species. Our results show that homologous recombination varies greatly along bacterial genomes and shapes many aspects of genome architecture and evolution. The genomic landscape of recombination presents several key signatures: rates are highest near the origin of replication in most species, patterns of recombination generally appear symmetrical in both replichores (i.e. replicational halves of circular chromosomes) and most species have genomic hotspots of recombination. Furthermore, many closely related species share conserved landscapes of recombination across orthologs indicating that recombination landscapes are conserved over significant evolutionary distances. We show evidence that recombination drives the evolution of GC-content through increasing the effectiveness of selection and not through biased gene conversion, thereby contributing to an ongoing debate. Finally, we demonstrate that the rate of recombination varies across gene function and that many hotspots of recombination are associated with adaptive and mobile regions often encoding genes involved in pathogenicity.

Torrance, Ellis L [University of North Carolina, G↗
DOE OSTI2025

A Scalable Multi-Modal Framework for High-Fidelity Distributed Human Mobility Simulations

The development of data-driven models for human mobility in urban settings requires access to substantial and diverse real-world data. However, existing historical data often presents challenges such as limited volume, variety, and veracity, as well as missing data and privacy preservation concerns. Also, urban mobility modeling is inherently time-variant, complex, and multi-modal, encompassing everything from individual walking and running to private road travel and large-scale public transportation. These challenges call for innovative solutions to overcome data limitations and compute needs to model mobility behaviors accurately. To address these challenges, we propose a distributed, co-simulation-based architecture DURMOSim that integrates real-world data with scalable, high-fidelity simulations, demonstrating distributed co-simulation feasibility with existing mobility models. DURMOSim underpins a modular integration that would enable using any available mobility simulators for greater extensibility and scalability in performing various urban scenarios. In this paper, we present the design, implementation, and performance evaluation of DURMOSim, highlighting its capability to model population-scale mobility patterns. Our initial results show its ability to dynamically synchronize multiple simulation models at runtime with negligible computational overhead. We believe DURMOSim could be a robust tool for advancing urban mobility research and intelligent transportation systems.

Yoginath, Srikanth [ORNL] (ORCID:0000000184236050)↗
DOE OSTI2024

Domain-decomposition nonlinear manifold reduced order model

This software combines nonlinear-manifold reduced order models (NM-ROMs) with domain decomposition (DD) techniques. NM-ROMs, which utilize a shallow, sparse autoencoder trained with full order model (FOM) snapshot data, approximate the FOM state on a nonlinear manifold. These models offer advantages over linear-subspace ROMs (LS-ROMs) particularly in scenarios with slowly decaying Kolmogorov n-width. However, the training of NM-ROMs involves a number of parameters that scale with the size of the FOM, and storing high-dimensional FOM snapshots can significantly increase the cost of ROM training for extreme-scale problems. To mitigate these costs, the software employs DD to partition the FOM into smaller subdomains, computes NM-ROMs for each, and then integrates these to form a global NM-ROM. This strategy offers multiple benefits: it enables parallel training of subdomain NM-ROMs, reduces the number of parameters needed, decreases the dimensional requirements of subdomain FOM training data, and allows for customization to the unique characteristics of each FOM subdomain. The use of a shallow, sparse autoencoder architecture in each subdomain NM-ROM facilitates the application of hyper-reduction (HR), simplifying the nonlinear complexities and enhancing computational speed. This software marks the inaugural application of NM-ROM combined with HR to a DD problem. It features an algebraic DD reformulation of the FOM, training of NM-ROMs with HR for each subdomain, and employs a sequential quadratic programming (SQP) solver for the evaluation of the coupled global NMROM. The effectiveness of the DD NM-ROM with HR is numerically demonstrated on the 2D steady-state Burgers' equation, showing an order of magnitude improvement in accuracy over the DD LS-ROM with HR.

Diaz, AlejandroN↗