Search NASA⌕ Search

SEARCH · Search NASA

Results for “Memory Optimization”

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 919 records · Page 51
All sourcesNASA NTRSDOE OSTI
DOE OSTI2015

Geant4 Computing Performance Benchmarking and Monitoring

Performance evaluation and analysis of large scale computing applications is essential for optimal use of resources. As detector simulation is one of the most compute intensive tasks and Geant4 is the simulation toolkit most widely used in contemporary high energy physics (HEP) experiments, it is important to monitor Geant4 through its development cycle for changes in computing performance and to identify problems and opportunities for code improvements. All Geant4 development and public releases are being profiled with a set of applications that utilize different input event samples, physics parameters, and detector configurations. Results from multiple benchmarking runs are compared to previous public and development reference releases to monitor CPU and memory usage. Observed changes are evaluated and correlated with code modifications. Besides the full summary of call stack and memory footprint, a detailed call graph analysis is available to Geant4 developers for further analysis. The set of software tools used in the performance evaluation procedure, both in sequential and multi-threaded modes, include FAST, IgProf and Open|Speedshop. In conclusion, the scalability of the CPU time and memory performance in multi-threaded application is evaluated by measuring event throughput and memory gain as a function of the number of threads for selected event samples.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Deterministic Current–Induced Perpendicular Switching in Epitaxial Co/Pt Layers without an External Field

Current–induced spin–orbit torques (SOTs) have emerged as a powerful tool to control magnetic elements and non–uniform magnetic textures such as domain walls and skyrmions. SOT–induced switching of perpendicular magnetization generally requires an external field to break the rotational symmetry of the spin–orbit effective fields responsible for the deterministic reversal. The proposed mechanisms to eliminate this requirement often rely on complex multilayer structures that necessitate laborious optimization in the material and spin transport properties, making them less attractive for applications. Herein, current–induced, external field–free switching of an epitaxial MgO/Pt/Co trilayer with an extremely large perpendicular anisotropy in excess of 3 Tesla is reported. It is found that switching occurs due to the interplay of strong SOTs, local anisotropy fluctuations, and the Dzyaloshinkii–Moriya interaction inherent to this epitaxial system. Finally, given that these layers constitute the base stack of a magnetic tunnel junction, this switching mechanism offers the most technologically viable path toward devices such as field–free SOT–based magnetic random–access memories.

36 MATERIALS SCIENCE↗
DOE OSTI2016

PARVMEC: An Efficient, Scalable Implementation of the Variational Moments Equilibrium Code

The ability to sustain magnetically confined plasma in a state of stable equilibrium is crucial for optimal and cost-effective operations of fusion devices like tokamaks and stellarators. The Variational Moments Equilibrium Code (VMEC) is the de-facto serial application used by fusion scientists to compute magnetohydrodynamics (MHD) equilibria and study the physics of three dimensional plasmas in confined configurations. Modern fusion energy experiments have larger system scales with more interactive experimental workflows, both demanding faster analysis turnaround times on computational workloads that are stressing the capabilities of sequential VMEC. In this paper, we present PARVMEC, an efficient, parallel version of its sequential counterpart, capable of scaling to thousands of processors on distributed memory machines. PARVMEC is a non-linear code, with multiple numerical physics modules, each with its own computational complexity. A detailed speedup analysis supported by scaling results on 1,024 cores of a Cray XC30 supercomputer is presented. Depending on the mode of PARVMEC execution, speedup improvements of one to two orders of magnitude are reported. PARVMEC equips fusion scientists for the first time with a state-of-theart capability for rapid, high fidelity analyses of magnetically confined plasmas at unprecedented scales.

Seal, Sudip K↗
DOE OSTI2024

Design and fabrication of robust hybrid photonic crystal cavities

Abstract Heterogeneously integrated hybrid photonic crystal cavities enable strong light–matter interactions with solid state, optically addressable quantum memories. A key challenge to realizing high quality factor ( Q ) hybrid photonic crystals is the reduced index contrast on the substrate compared to suspended devices in air. This challenge is particularly acute for color centers in diamond because of diamond’s high refractive index, which leads to increased scattering loss into the substrate. Here, we develop a design methodology for hybrid photonic crystals utilizing a detailed understanding of substrate-mediated loss, which incorporates sensitivity to fabrication errors as a critical parameter. Using this methodology, we design robust, high-Q, GaAs-on-diamond photonic crystal cavities, and by optimizing our fabrication procedure, we experimentally realize cavities with Q approaching 30,000 at a resonance wavelength of 955 nm.

Abulnaga, Alex↗
DOE OSTI2022

Feasibility Study of Millimeter Wave Radars For Safeguards Applications

Containment and surveillance are fundamental measures in nuclear safeguards. Techniques such as video surveillance and laser curtain for containment provide effective monitoring in areas where maintaining continuity of knowledge is required. These systems, however, can be susceptible to loss of monitoring capabilities under certain environmental conditions such as poor visibility (i.e. low light conditions, smoke, fog, etc.) or extended power loss past the duration that the backup power system is designed for. Brookhaven National Laboratory has been investigating the feasibility of millimeter waves (mmWave) as a new perimeter seal in which radio frequency waves in the range of 60-64 GHz are used to detect and monitor objects of interest. Signals in this frequency range are not susceptible to environmental conditions. For proof-of-concept tests, mmWave sensors from Texas Instruments (TI), specificallyIWR6843, are used in a test bed at BNL's Waste Management facility to simulate the operations at nuclear facilities. The unique design of TI mmWave sensors requires less memory and power consumption compared to counterpart systems. These devices are capable of exporting 3D point-cloud data, which is visualized graphically and compared to videos recorded at the same time to validate the performance of the mmWave sensor. A set of experiments were planned to test the feasibility of the mmWave in this application, including monitoring static containers in a storage area and detecting intrusions at the boundaries of the area. In addition, the experiments also identify potential blind spots relative to sensor position and utilize multiple operating sensors simultaneously to reduce or eliminate such blind spots. The optimal positioning of multiple sensors was determined for the experimental room configuration. In this paper, we will discuss the details of this novel perimeter sealing concept and present the test results.

98 NUCLEAR DISARMAMENT, SAFEGUARDS, AND PHYSICAL P↗
DOE OSTI2024

Optimizing Distributed Training on Frontier for Large Language Models

Large language models (LLMs) have demonstrated remarkable success as foundational models, benefiting various downstream applications through fine-tuning. Loss scaling studies have demonstrated the superior performance of larger LLMs compared to their smaller counterparts. Nevertheless, training LLMs with billions of parameters poses significant challenges and requires considerable computational resources. For example, training a one trillion parameter GPT-style model on 20 trillion tokens requires a staggering 120 million exaflops. This research explores efficient distributed training strategies to extract this computation from Frontier, the world's first exascale supercomputer. We enable and investigate various model and data parallel training techniques, such as tensor parallelism, pipeline parallelism, and sharded data parallelism, to facilitate training a trillion-parameter model on Frontier. We empirically assess these techniques and their associated parameters to determine their impact on memory footprint, communication latency, and GPU's computational efficiency. We analyze the complex interplay among these techniques and find a strategy to combine them to achieve high throughput through hyperparameter tuning. We have identified efficient strategies for training large LLMs of varying sizes through empirical analysis and hyperparameter tuning. For 22 Billion, 175 Billion, and 1 Trillion parameters, we achieved GPU throughputs of 38.38%, 36.14%, and 31.96%, respectively. For the training of the 175 Billion parameter model and the 1 Trillion parameter model, we achieved 100% weak scaling efficiency on 1024 and 3072 Mi250X GPUs, respectively. We also achieved strong scaling efficiencies of 89% and 87% for these two models. We trained these models only tens of iterations instead of training till completion.

Yin, Junqi↗
DOE OSTI2018

Basic Research Needs for Microelectronics: Report of the Office of Science Workshop on Basic Research Needs for Microelectronics, October 23 – 25, 2018

Moore’s Law — realized via a combination of device physics advances, technology investments, and economic returns— allowed the number of transistors on a chip to double roughly every two years for over five decades. During those five decades, the cost of a unit of computing dropped by eight orders of magnitude. Those declines and the associated computing advances have dramatically affected every aspect of society, from science and technology through business and health to national security. Today, further computing advances are limited by the economics of chip fabrication and the physical limits on transistors, electrical interconnects, and memory elements at the nanoscale. If we are to evolve new generations of computing systems over the next decades that are both faster and more energy efficient, a complete reorganization of the science and technology underlying computing is needed. Within the Department of Energy’s (DOE’s) mission, which includes a pivotal and historical role in the evolution of computing, the needs are critical as well. Advanced computing and simulation underpin all aspects of DOE missions in energy, the environment, and national security, requiring energy-efficient computing beyond the exascale range. Edge computing, low-power computing technologies, and computers optimized for artificial intelligence are key to next-generation scientific facilities for high-energy physics research, as well as neutron and x-ray facilities. Finally, microelectronics will also play a major role in reshaping the U.S. electricity grid: from its current state to one that is cleaner, more efficient, cyber-secure, and resilient to widespread events, both natural and manmade. Realizing this vision will require advancements in both system architecture and power electronics. To enable continued advances in computing technologies, a fundamental rethinking is needed of the science behind the materials, synthesis and placement technologies, architectures, and algorithms. This cannot be modular and linear, as it has been in the past. Rather, these advances must be developed collectively, in a spirit of co-design, where each scientific discipline informs and engages the other to achieve orders of magnitude improvements in system-level performance. To explore these challenges, the Office of Science convened a Basic Research Needs Workshop for Microelectronics in October 2018 and charged workshop participants to conduct a thorough assessment of the scientific issues associated with advanced microelectronics technologies for applications relevant to the DOE mission. The workshop examined research relevant to the extension of complementary metal oxide semiconductor (CMOS) and beyond CMOS technologies; however, topics of direct relevance to quantum information science and quantum computing were outside the scope of this workshop.

42 ENGINEERING↗
DOE OSTI2016

Computing Maximum Cardinality Matchings in Parallel on Bipartite Graphs via Tree-Grafting

It is difficult to obtain high performance when computing matchings on parallel processors because matching algorithms explicitly or implicitly search for paths in the graph, and when these paths become long, there is little concurrency. In spite of this limitation, we present a new algorithm and its shared-memory parallelization that achieves good performance and scalability in computing maximum cardinality matchings in bipartite graphs. This algorithm searches for augmenting paths via specialized breadth-first searches (BFS) from multiple source vertices, hence creating more parallelism than single source algorithms. Algorithms that employ multiple-source searches cannot discard a search tree once no augmenting path is discovered from the tree, unlike algorithms that rely on single-source searches. We describe a novel tree-grafting method that eliminates most of the redundant edge traversals resulting from this property of multiple-source searches. We also employ the recent direction-optimizing BFS algorithm as a subroutine to discover augmenting paths faster. Our algorithm compares favorably with the current best algorithms in terms of the number of edges traversed, the average augmenting path length, and the number of iterations. Here, we provide a proof of correctness for our algorithm. Our NUMA-aware implementation is scalable to 80 threads of an Intel multiprocessor and to 240 threads on an Intel Knights Corner coprocessor. On average, our parallel algorithm runs an order of magnitude faster than the fastest algorithms available. The performance improvement is more significant on graphs with small matching number.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

XploreNAS : Explore Adversarially Robust and Hardware-efficient Neural Architectures for Non-ideal Xbars

Compute In-Memory platforms such as memristive crossbars are gaining focus as they facilitate acceleration of Deep Neural Networks (DNNs) with high area and compute efficiencies. However, the intrinsic non-idealities associated with the analog nature of computing in crossbars limits the performance of the deployed DNNs. Furthermore, DNNs are shown to be vulnerable to adversarial attacks leading to severe security threats in their large-scale deployment. Thus, finding adversarially robust DNN architectures for non-ideal crossbars is critical to the safe and secure deployment of DNNs on the edge. This work proposes a two-phase algorithm-hardware co-optimization approach called XploreNAS that searches for hardware efficient and adversarially robust neural architectures for non-ideal crossbar platforms. We use the one-shot Neural Architecture Search approach to train a large Supernet with crossbar-awareness and sample adversarially robust Subnets therefrom, maintaining competitive hardware efficiency. Our experiments on crossbars with benchmark datasets (SVHN, CIFAR10, CIFAR100) show up to ~8–16% improvement in the adversarial robustness of the searched Subnets against a baseline ResNet-18 model subjected to crossbar-aware adversarial training. We benchmark our robust Subnets for Energy-Delay-Area-Products (EDAPs) using the Neurosim tool and find that with additional hardware efficiency–driven optimizations, the Subnets attain ~1.5–1.6× lower EDAPs than ResNet-18 baseline.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2014

Considerations Related To Human Intrusion In The Context Of Disposal Of Radioactive Waste-The IAEA HIDRA Project

The principal approaches for management of radioactive waste are commonly termed ‘delay and decay’, ‘concentrate and contain’ and ‘dilute and disperse’. Containing the waste and isolating it from the human environment, by burying it, is considered to increase safety and is generally accepted as the preferred approach for managing radioactive waste. However, this approach results in concentrated sources of radioactive waste contained in one location, which can pose hazards should the facility be disrupted by human action in the future. The International Commission on Radiological Protection (ICRP), International Atomic Energy Agency (IAEA), and Organization for Economic Cooperation and Development/Nuclear Energy Agency (OECD/NEA) agree that some form of inadvertent human intrusion (HI) needs to be considered to address the potential consequences in the case of loss of institutional control and loss of memory of the disposal facility. Requirements are reflected in national regulations governing radioactive waste disposal. However, in practice, these requirements are often different from country to country, which is then reflected in the actual implementation of HI as part of a safety case. The IAEA project on HI in the context of Disposal of RadioActive waste (HIDRA) has been started to identify potential areas for improved consistency in consideration of HI. The expected outcome is to provide recommendations on how to address human actions in the safety case in the future, and how the safety case may be used to demonstrate robustness and optimize siting, design and waste acceptance criteria within the context of a safety case.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2025

Reducing Communication Overhead in Federated Learning for Network Anomaly Detection with Adaptive Client Selection

Communication overhead in federated learning (FL) poses a significant challenge for network anomaly detection systems, where the myriad of client configurations and network conditions can severely impact system efficiency and detection accuracy. While existing approaches attempt to address this through individual optimization techniques, they often fail to maintain the delicate balance between reduced overhead and detection performance. This paper presents an adaptive FL framework that dynamically combines batch size optimization, client selection, and asynchronous updates to achieve efficient anomaly detection. Through extensive profiling and experimental analysis on two distinct datasets-UNSW-NBIS for general network traffic and ROAD for automotive networks-our framework reduces communication overhead by 97.6%; (from 700.0s to 16.8s) compared to synchronous baseline approaches while maintaining comparable detection accuracy (95.10%; vs. 95.12%;). Statistical validation using Mann-Whitney U test confirms significant improvements (p < 0.05) over existing FL approaches across both datasets, demonstrating the framework's adaptability to different network security contexts. Detailed profiling analysis reveals the efficiency gains through dramatic reductions in GPU operations and memory transfers while maintaining robust detection performance under varying client conditions.

Marfo, William [University of Texas at El Paso]↗
DOE OSTI2017

Experiments and Analyses of Data Transfers Over Wide-Area Dedicated Connections

Dedicated wide-area network connections are increasingly employed in high-performance computing and big data scenarios. One might expect the performance and dynamics of data transfers over such connections to be easy to analyze due to the lack of competing traffic. However, non-linear transport dynamics and end-system complexities (e.g., multi-core hosts and distributed filesystems) can in fact make analysis surprisingly challenging. We present extensive measurements of memory-to-memory and disk-to-disk file transfers over 10 Gbps physical and emulated connections with 0–366 ms round trip times (RTTs). For memory-to-memory transfers, profiles of both TCP and UDT throughput as a function of RTT show concave and convex regions; large buffer sizes and more parallel flows lead to wider concave regions, which are highly desirable. TCP and UDT both also display complex throughput dynamics, as indicated by their Poincare maps and Lyapunov exponents. For disk-to-disk transfers, we determine that high throughput can be achieved via a combination of parallel I/O threads, parallel network threads, and direct I/O mode. Our measurements also show that Lustre filesystems can be mounted over long-haul connections using LNet routers, although challenges remain in jointly optimizing file I/O and transport method parameters to achieve peak throughput.

Rao, Nageswara S.↗
DOE OSTI2016

Enabling NVM for Data-Intensive Scientific Services

Specialized, transient data services are playing an increasingly prominent role in data-intensive scientific computing. These services offer flexible, on-demand pairing of applications with storage hardware using semantics that are optimized for the problem domain. Concurrent with this trend, upcoming scientific computing and big data systems will be deployed with emerging NVM technology to achieve the highest possible price/productivity ratio. Clearly, therefore, we must develop techniques to facilitate the confluence of specialized data services and NVM technology. In this work we explore how to enable the composition of NVM resources within transient distributed services while still retaining their essential performance characteristics. Our approach involves eschewing the conventional distributed file system model and instead projecting NVM devices as remote microservices that leverage user-level threads, RPC services, RMA-enabled network transports, and persistent memory libraries in order to maximize performance. We describe a prototype system that incorporates these concepts, evaluate its performance for key workloads on an exemplar system, and discuss how the system can be leveraged as a component of future data-intensive architectures.

Carns, Philip↗
DOE OSTI2025

A kinetic-based regularization method for data science applications

We propose a physics-based regularization technique for function learning, inspired by statistical mechanics. By drawing an analogy between optimizing the parameters of an interpolator and minimizing the energy of a system, we introduce corrections that impose constraints on the lower-order moments of the data distribution. This minimizes the discrepancy between the discrete and continuum representations of the data, in turn allowing to access more favorable energy landscapes, thus improving the accuracy of the interpolator. Our approach improves performance in both interpolation and regression tasks, even in high-dimensional spaces. Unlike traditional methods, it does not require empirical parameter tuning, making it particularly effective for handling noisy data. We also show that thanks to its local nature, the method offers computational and memory efficiency advantages over Radial Basis Function interpolators, especially for large datasets.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2019

Reverse-Mode Algorithmic Differentiation of an OpenMP-Parallel Compressible Flow Solver

y Reverse-mode algorithmic differentiation (AD) is an established method for obtaining adjoint derivatives of computer simulation applications. In computational fluid dynamics (CFD), adjoint derivatives of a cost function output such as drag or lift with respect to design parameters such as surface coordinates or geometry control points are a key ingredient for shape optimization, uncertainty quantification and flow control. The computational cost of CFD applications and their derivatives makes it essential to use high-performance computing hardware efficiently, including multi- and many-core architectures. Nevertheless, OpenMP is not supported in most AD tools, and previously shown methods achieve poor scalability of the derivative code. We present the AD of an OpenMP-parallelized finite volume compressible flow solver for unstructured meshes. Our approach enables us to reuse the parallelization of the original code in the computation of adjoint derivatives. The method works by identifying code segments that can be differentiated in reverse-mode without changing their memory access pattern. The OpenMP parallelization is integrated into the derivative code during the build process in a way that is robust to modifications of the original code and independent of the OpenMP support of the differentiation tool. We show the scalability of our adjoint CFD solver on test cases ranging from thousands to millions of finite volume mesh cells on CPUs with up to 16 threads as well as on an Intel XeonPhi card with 236 threads. We demonstrate that our approach is more practical to implement for production-sized CFD codes and produces more efficient adjoint derivative code than previously shown AD methods.

Huckelheim, Jan↗
DOE OSTI2016

iTOUGH2 v7.1

iTOUGH2 (inverse TOUGH2) provides inverse modeling capabilities for TOUGH2, a simulator for multi-dimensional, multi-phase, multi-component, non-isothermal flow and transport in fractured porous media. iTOUGH2 performs sensitivity analyses, data-worth analyses, parameter estimation, and uncertainty propagation analyses in geosciences and reservoir engineering and other application areas. iTOUGH2 supports a number of different combinations of fluids and components (equation-of-state (EOS) modules). In addition, the optimization routines implemented in iTOUGH2 can also be used for sensitivity analysis, automatic model calibration, and uncertainty quantification of any external code that uses text-based input and output files using the PEST protocol. iTOUGH2 solves the inverse problem by minimizing a non-linear objective function of the weighted differences between model output and the corresponding observations. Multiple minimization algorithms (derivative-free, gradient-based, and second-order; local and global) are available. iTOUGH2 also performs Latin Hypercube Monte Carlo simulations for uncertainty propagation analyses. A detailed residual and error analysis is provided. This upgrade includes (a) global sensitivity analysis methods, (b) dynamic memory allocation (c) additional input features and output analyses, (d) increased forward simulation capabilities, (e) parallel execution on multicore PCs and Linux clusters, and (f) bug fixes.

Finsterle, Stefan↗
DOE OSTI2020

End-to-end online performance data capture and analysis for scientific workflows

With the increased prevalence of employing workflows for scientific computing and a push towards exascale computing, it has become paramount that we are able to analyze characteristics of scientific applications to better understand their impact on the underlying infrastructure and vice-versa. Such analysis can help drive the design, development, and optimization of these next generation systems and solutions. Here, we present the architecture, integrated with existing well-established and newly developed tools, to collect online performance statistics of workflow executions from various, heterogeneous sources and publish them in a distributed database (Elasticsearch). Using this architecture, we are able to correlate online workflow performance data, with data from the underlying infrastructure, and present them in a useful and intuitive way via an online dashboard. We have validated our approach by executing two classes of real-world workflows, both under normal and anomalous conditions. The first is an I/O-intensive genome analysis workflow; the second, a CPU- and memory-intensive material science workflow. Based on the data collected in Elasticsearch, we are able to demonstrate that we can correctly identify anomalies that we injected. The resulting end-to-end data collection of workflow performance data is an important resource of training data for automated machine learning analysis.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2015

Proposal for a graphene-based all-spin logic gate

In this work, we present a graphene-based all-spin logic gate (G-ASLG) that integrates the functionalities of perpendicular anisotropy magnetic tunnel junctions (p-MTJs) with spin transport in graphene-channel. It provides an ideal integration of logic and memory. The input and output states are defined as the relative magnetization between free layer and fixed layer of p-MTJs. They can be probed by the tunnel magnetoresistance and controlled by spin transfer torque effect. Using lateral non-local spin valve, the spin information is transmitted by the spin-current interaction through graphene channels. By using a physics-based spin current compact model, the operation of G-ASLG is demonstrated and its performance is analyzed. It allows us to evaluate the influence of parameters, such as spin injection efficiency, spin diffusion length, contact area, the device length, and their interdependence, and to optimize the energy and dynamic performance. Compared to other beyond-CMOS solutions, longer spin information transport length (∼μm), higher data throughput, faster computing speed (∼ns), and lower power consumption (∼μA) can be expected from the G-ASLG.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗