Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “compute in memory”

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 541 records · Page 30
All sourcesNASA NTRSDOE OSTI
NASA NTRS2003

Iterative pass optimization of sequence data

The problem of determining the minimum-cost hypothetical ancestral sequences for a given cladogram is known to be NP-complete. This "tree alignment" problem has motivated the considerable effort placed in multiple sequence alignment procedures. Wheeler in 1996 proposed a heuristic method, direct optimization, to calculate cladogram costs without the intervention of multiple sequence alignment. This method, though more efficient in time and more effective in cladogram length than many alignment-based procedures, greedily optimizes nodes based on descendent information only. In their proposal of an exact multiple alignment solution, Sankoff et al. in 1976 described a heuristic procedure--the iterative improvement method--to create alignments at internal nodes by solving a series of median problems. The combination of a three-sequence direct optimization with iterative improvement and a branch-length-based cladogram cost procedure, provides an algorithm that frequently results in superior (i.e., lower) cladogram costs. This iterative pass optimization is both computation and memory intensive, but economies can be made to reduce this burden. An example in arthropod systematics is discussed. c2003 The Willi Hennig Society. Published by Elsevier Science (USA). All rights reserved.

NASA Discipline Evolutionary Biology↗
NASA NTRS2004

Using an Extended Kalman Filter Learning Algorithm for Feed-Forward Neural Networks to Describe Tracer Correlations

In this study a new extended Kalman filter (EKF) learning algorithm for feed-forward neural networks (FFN) is used. With the EKF approach, the training of the FFN can be seen as state estimation for a non-linear stationary process. The EKF method gives excellent convergence performances provided that there is enough computer core memory and that the machine precision is high. Neural networks are ideally suited to describe the spatial and temporal dependence of tracer-tracer correlations. The neural network performs well even in regions where the correlations are less compact and normally a family of correlation curves would be required. For example, the CH4-N2O correlation can be well described using a neural network trained with the latitude, pressure, time of year, and CH4 volume mixing ratio (v.m.r.). The neural network was able to reproduce the CH4-N2O correlation with a correlation coefficient between simulated and training values of 0.9997. The neural network Fortran code used is available for download.

Lary, David J.↗
NASA NTRS1994

Computer driven optical keratometer and method of evaluating the shape of the cornea

An apparatus and method for measuring the shape of the cornea utilize only one reticle to generate a pattern of rings projected onto the surface of a subject's eye. The reflected pattern is focused onto an imaging device such as a video camera and a computer compares the reflected pattern with a reference pattern stored in the computer's memory. The differences between the reflected and stored patterns are used to calculate the deformation of the cornea which may be useful for pre-and post-operative evaluation of the eye by surgeons.

Baroth, Edmund C.↗
NASA NTRS2005

Highly Reliable, High-Speed, Unidirectional Serial Data Links

Highly reliable, high-speed, unidirectional serial data-communication subsystems have been proposed to be installed in an upgrade of the computing systems aboard the space shuttles. The basic design concept of these serial data links is also adaptable to terrestrial use in applications in which there are requirements for highly reliable serial data communications. The hardware and software aspects of the architecture of the data links are dictated largely by a requirement, in the original space-shuttle application, for one computer to monitor the memory transactions and memory contents of other computers in real time with high reliability and without reliance on requests for retransmission. To minimize weight while affording a capability to transfer data at a required rate of 2.56 x 10(exp 8) bits per second, it was decided that the links would be serial ones of the fiber-channel type. [Fiber channel denotes a type of serial computer bus that is used to connect a computer (usually a supercomputer) with a high-speed data storage device. Depending on the specific application, the physical connection between the transmitter and receiver could be made via an optical fiber or a twisted pair of wires.] Heretofore, fiber-channel links have ordinarily been bidirectional and have operated under protocols that provide for receiving stations to detect errors and request retransmission when necessary. In the present case, the time taken by processing to request retransmission would conflict with the requirement for real-time transfer of data. To ensure reliability without retransmission, a link according to the proposal would utilize a modified version of the normal fiberchannel character set in conjunction with forward error correction by means of a Reed-Solomon code (see figure). The Reed-Solomon encoding and decoding and the translations between the normal and modified character sets would be effected by logic circuitry external to the fiber-channel transmitter and receiver, which would be commercial off-the-shelf units. The receiving end of the link could detect and correct errors at a rate as high as 4 million times per second, if necessary. The receiver detects uncorrectable double-byte errors. It has been estimated that uncorrectable-error rate would amount to one failure in about 10(exp 19) characters.

Cole, Robert M.↗
NASA NTRS2013

ULTRA: Underwater Localization for Transit and Reconnaissance Autonomy

This software addresses the issue of underwater localization of unmanned vehicles and the inherent drift in their onboard sensors. The software gives a 2 to 3 factor of improvement over the state-of-the-art underwater localization algorithms. The software determines the localization (position, heading) of an AUV (autonomous underwater vehicle) in environments where there is no GPS signal. It accomplishes this using only the commanded position, onboard gyros/accelerometers, and the bathymetry of the bottom provided by an onboard sonar system. The software does not rely on an onboard bathymetry dataset, but instead incrementally determines the position of the AUV while mapping the bottom. In order to enable long-distance underwater navigation by AUVs, a localization method called ULTRA uses registration of the bathymetry data products produced by the onboard forward-looking sonar system for hazard avoidance during a transit to derive the motion and pose of the AUV in order to correct the DR (dead reckoning) estimates. The registration algorithm uses iterative point matching (IPM) combined with surface interpolation of the Iterative Closest Point (ICP) algorithm. This method was used previously at JPL for onboard unmanned ground vehicle localization, and has been optimized for efficient computational and memory use.

Huntsberger, Terrance L.↗
NASA NTRS2013

PhoneSat - The Smartphone Nanosatellite

PhoneSat 2.4, carried into space on November 19, 2013 aboard a Minotaur I rocket from the Mid-Atlantic Regional Spaceport at NASAs Wallops Flight Facility in Virginia, is the first of the PhoneSat family to use a two-way S-band radio to allow engineers to command the satellite from Earth. This mission also serves as a technology demonstration for a novel attitude determination and control system (ADCS) that establishes and stabilizes the satellites attitude relative to Earth. Unlike the earlier PhoneSats that used a Nexus One, PhoneSat 2.4 uses the Nexus S smartphone, which runs Googles Android operating system, and is made by Samsung Electronics Co., Suwon, So. Korea. The smartphone provides many of the functions needed by the satellite such as a central computer, data memory, ready-made interfaces for communications, navigation and power all pre-assembled in a rugged electronics package.

PhoneSat↗
NASA NTRS2016

Parameter Uncertainty for Aircraft Aerodynamic Modeling using Recursive Least Squares

A real-time method was demonstrated for determining accurate uncertainty levels of stability and control derivatives estimated using recursive least squares and time-domain data. The method uses a recursive formulation of the residual autocorrelation to account for colored residuals, which are routinely encountered in aircraft parameter estimation and change the predicted uncertainties. Simulation data and flight test data for a subscale jet transport aircraft were used to demonstrate the approach. Results showed that the corrected uncertainties matched the observed scatter in the parameter estimates, and did so more accurately than conventional uncertainty estimates that assume white residuals. Only small differences were observed between batch estimates and recursive estimates at the end of the maneuver. It was also demonstrated that the autocorrelation could be reduced to a small number of lags to minimize computation and memory storage requirements without significantly degrading the accuracy of predicted uncertainty levels.

Grauer, Jared A.↗
NASA NTRS2000

Quantitative analysis of charge injection and discharging of Si nanocrystals and arrays by electrostatic force microscopy

NASA requirements for computing and memory for microspacecraft emphasizes high density, low power, small size, and radiation hardness. The distributed nature of a storage elements in nanocrystal floating-gate memories leads to instrinsic fault tolerance and radiation-hardness. Nanocrystal-based memories also offer the possibility of a faster, lower power operation.

atomic↗
NASA NTRS2017

Gaussian Mixture Models for Temporal Depth Fusion

Sensing the 3D environment of a moving robot is essential for collision avoidance. Most 3D sensors produce dense depth maps, which are subject to imperfections due to various environmental factors. Temporal fusion of depth maps is crucial to overcome those. Temporal fusion is traditionally done in 3D space with voxel data structures, but it can be approached by temporal fusion in image space, with potential benefits in reduced memory and computational cost for applications like reactive collision avoidance for micro air vehicles. In this paper, we present an efficient Gaussian Mixture Models based depth map fusion approach, introducing an online update scheme for dense representations. The environment is modeled from an ego-centric point of view, where each pixel is represented by a mixture of Gaussian inverse-depth models. Consecutive frames are related to each other by transformations obtained from visual odometry. This approach achieves better accuracy than alternative image space depth map fusion techniques at lower computational cost.

Matthies, Larry↗
DOE OSTI2023

Multichannel angular spectrometer

A multichannel angular spectrometer includes an array of fiber pickups having an arcuate arrangement and focused about a sample volume. A broadband light source is configured to illuminate a sample within the sample volume. At least one dispersion element is in optical communication with the array of fiber pickups. An imaging sensor is in optical communication with the array of fiber pickups. The imaging sensor is configured to image the broadband light received by the array of fiber pickups and dispersed by the at least one dispersion element. A processor is in electrical communication with the imaging sensor. The processor has a power supply and computer-readable memory.

47 OTHER INSTRUMENTATION↗
DOE OSTI2024

Adaptive Patching for High-resolution Image Segmentation with Transformers

Attention-based models are proliferating in the space of image analytics, including segmentation. The standard method of feeding images to transformer encoders is to divide the images into patches and then feed the patches to the model as a linear sequence of tokens. For high-resolution images, e.g. microscopic pathology images, the quadratic compute and memory cost prohibits the use of an attention-based model, if we are to use smaller patch sizes that are favorable in segmentation. The solution is to either use custom complex multi-resolution models or approximate attention schemes. We take inspiration from Adapative Mesh Refinement (AMR) methods in HPC by adaptively patching the images, as a pre-processing step, based on the image details to reduce the number of patches being fed to the model, by orders of magnitude. This method has a negligible overhead, and works seamlessly with any attention-based model, i.e. it is a pre-processing step that can be adopted by any attention-based model without friction. We demonstrate superior segmentation quality over SoTA segmentation models for realworld pathology datasets while gaining a geomean speedup of 6.9× for resolutions up to 64K2, on up to 2, 048 GPUs.

Zhang, Enzhi↗
DOE OSTI2024

Geometry-aware training of factorized layers in tensor Tucker format

Reducing parameter redundancies in neural network architectures is crucial for achieving feasible computational and memory requirements during train and inference of large networks. Given its easy implementation and flexibility, one promising approach is layer factorization, which reshapes weight tensors into a matrix format and parameterizes it as the product of two rank-r matrices. However, this family of approaches often requires an initial full-model warm-up phase, prior knowledge of a feasible rank, and it is sensitive to parameter initialization.In this work, we introduce a novel approach to train the factors of a Tucker decomposition of the weight tensors. Our training proposal proves to be optimal in locally approximating the original unfactorized dynamics and stable for the initialization. Furthermore, the rank of each mode is dynamically updated during training.We provide a theoretical analysis of the algorithm, showing convergence, approximation and local descent guarantees. The method's performance is further illustrated through a variety of experiments, showing remarkable training compression rates and comparable or even better performance than the full baseline and alternative layer factorization strategies.

Zangrando, Emanuele [Gran Sasso Science Institute ↗
NASA NTRS1991

Collecting address traces from parallel computers

Trace driven simulation is a well-established method of performance analysis for single processor computer systems. However, efficient and accurate memory address tracing for parallel computer systems is not well understood. In this paper we present a critical survey of recently implemented approaches to address tracing and highlight the issues specific to collection of traces for both shared and distributed memory parallel computers. These issues include potential distortion of the relative ordering of events by the address tracing activity, realistic interleaving of addresses generated by multiple processors, and I/O and storage problems associated with collecting traces for large parallel systems. The strengths and weaknesses of the parallel tracing approaches are described.

Stunkel, Craig B.↗
DOE OSTI2025

Dual-ion ECRAM as a stable and accurate analog synapse

Electrochemical random-access memory (ECRAM) works by tuning the bulk electronic conductance of functional materials via reversible, electrochemical insertion of ions, resulting in stable analog resistive switching, attractive for analog in-memory and neuromorphic computing. However, achieving fast programming for training and long retention for inference has been elusive. Protonic ECRAM demonstrates fast programming but insufficient retention, while oxygen-based ECRAM with excellent retention requires elevated programming temperatures. Cu-based ECRAM offers a compromise, with an activation energy (E A ) of ≈0.76 eV between protons (E A ≈ 0.4 eV) and oxygen (E A > 1 eV), enabling extensive retention and room temperature programming. Combining Cu 2+ ions with protons to form a dual-ion ECRAM, we demonstrate two distinct switching behaviors: fast switching at ≤5 V, (E A ≈ 0.45 eV) via protons, and nonvolatile, room temperature switching at ≥8 V, with E A ≈ 0.76 eV via Cu 2+ ions. In conclusion, the Cu-based state exhibits a wide conductance range, with excellent retention, low noise, and linear current-voltage behavior, achieving digital-equivalent ImageNet inference accuracy.

analog in-memory computing↗
NASA NTRS1973

Digital hardware and computer system for a digital image recorder.

Description of the digital memory and related computer system used at the Palomar Mountain Observatory. The memory was designed to have a large number of pixels, a high capacity for each pixel, and fast access for use with a variety of image sensors. As an image integration and storage device, this memory is shown to possess ideal properties, namely, it is strictly linear, has a wide dynamic range, and the image does not deteriorate with time. The observer has a visual display which enables him to monitor the image as it is being integrated.-

Dennison, E. W.↗