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 1,261 records · Page 70
All sourcesNASA NTRSDOE OSTI
NASA NTRS1998

Molecular Nanotechnology and Space Settlement

Atomically precise manipulation of matter is becoming increasingly common in laboratories around the world. As this control moves into aerospace systems, huge improvements in computers, high-strength materials, and other systems are expected. For example, studies suggest that it may be possible to build: 10(exp 18) MIPS computers, 10(exp 15) bytes/sq cm write once memory, $153-412/kg-of-cargo single- stage-to-orbit launch vehicles and active materials which sense their environment and react intelligently. All of NASA's enterprises should benefit significantly from molecular nanotechnology. Although the time may be measured in decades and the precise path to molecular nanotechnology is unclear, all paths (diamondoid, fullerene, self-assembly, biomolecular, etc.) will require very substantial computation. This talk will discuss fullerene nanotechnology and early work on hypothetical active materials consisting of large numbers of identical machines. The speaker will also discuss aerospace applications, particularly missions leading to widespread space settlement (e.g., small near-Earth - object retrieval). It is interesting to note that control of the tiny - individual atoms and molecules - may lead to colonization of the huge -first the solar system, then the galaxy.

Globus, Al↗
NASA NTRS2000

Message Passing vs. Shared Address Space on a Cluster of SMPs

The convergence of scalable computer architectures using clusters of PCs (or PC-SMPs) with commodity networking has become an attractive platform for high end scientific computing. Currently, message-passing and shared address space (SAS) are the two leading programming paradigms for these systems. Message-passing has been standardized with MPI, and is the most common and mature programming approach. However message-passing code development can be extremely difficult, especially for irregular structured computations. SAS offers substantial ease of programming, but may suffer from performance limitations due to poor spatial locality, and high protocol overhead. In this paper, we compare the performance of and programming effort, required for six applications under both programming models on a 32 CPU PC-SMP cluster. Our application suite consists of codes that typically do not exhibit high efficiency under shared memory programming. due to their high communication to computation ratios and complex communication patterns. Results indicate that SAS can achieve about half the parallel efficiency of MPI for most of our applications: however, on certain classes of problems SAS performance is competitive with MPI. We also present new algorithms for improving the PC cluster performance of MPI collective operations.

Shan, Hongzhang↗
DOE OSTI2022

A second-order distributed memory parallel fast sweeping method for the Eikonal equation

The Eikonal equation is used to calculate wave propagation and distance fields, and due to its complexity requires numerical treatment for its solution. In this work, we present a second-order distributed memory parallel fast sweeping method. The second-order solution switches on a two-point stencil when two upwind points are available, and reverts to first-order otherwise. In all examples, the second-order method improves the solution over the first-order, allowing for significant savings in memory while achieving the same accuracy. Parallelization over distributed memory saw good weak scaling with optimal convergence. The computational time for second-order was approximately 2.5 times slower than first-order, where the largest amount of mesh points ran on 144 cores (512 GB) was ≈20 billion. The savings in memory from the second-order method combined with the distributed memory algorithm result in the ability to solve problems much larger than are possible with the serial first-order method.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
NASA NTRS2017

Computational and Experimental Fluid-Structure Interaction Analysis of a High-Lift Wing with a Slat-Cove Filler for Noise Reduction

Airframe noise is a significant component of overall noise produced by transport aircraft during landing and approach (low speed maneuvers). A significant source for this noise is the cove of the leading-edge slat. The slat-cove filler (SCF) has been shown to be effective at mitigating slat noise. The objective of this work is to understand the fluid-structure interaction (FSI) behavior of a superelastic shape memory alloy (SMA) SCF in flow using both computational and physical models of a high-lift wing. Initial understanding of flow around the SCF and wing is obtained using computational fluid dynamics (CFD) analysis at various angles of attack. A framework compatible with an SMA constitutive model (implemented as a user material subroutine) is used to perform FSI analysis for multiple flow and configuration cases. A scaled physical model of the high-lift wing is constructed and tested in the Texas A&M 3 ft-by-4-foot wind tunnel. Initial validation of both CFD and FSI analysis is conducted by comparing lift, drag and pressure distributions with experimental results.

Scholten, William D.↗
DOE OSTI2020

Efficient discontinuous Galerkin implementations and preconditioners for implicit unsteady compressible flow simulations

This work presents and compares efficient implementations of high-order discontinuous Galerkin methods: a modal matrix-free discontinuous Galerkin (DG) method, a hybridizable discontinuous Galerkin (HDG) method, and a primal formulation of HDG, applied to the implicit solution of unsteady compressible flows. The matrix-free implementation allows for a reduction of the memory footprint of the solver when dealing with implicit time-accurate discretizations. HDG reduces the number of globally-coupled degrees of freedom relative to DG, at high order, by statically condensing element-interior degrees of freedom from the system in favor of face unknowns. The primal formulation further reduces the element-interior degrees of freedom by eliminating the gradient as a separate unknown. This paper introduces a p-multigrid preconditioner implementation for these discretizations and presents results for various flow problems. Benefits of the p-multigrid strategy relative to simpler, less expensive, preconditioners are observed for stiff systems, such as those arising from low-Mach number flows at high-order approximation. The p-multigrid preconditioner also shows excellent scalability for parallel computations. Finally, additional savings in both speed and memory occur with a matrix-free/reduced version of the preconditioner.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2005

Automated Synthesis of Long Communication Delays for Testing

Planetary-Ohio Network Emulator (p- ONE) is a computer program for local laboratory testing of high bandwidth data-communication systems subject to long delays in propagation over interplanetary distances. p-ONE is installed on a personal computer connected to two bidirectional Ethernet interfaces, denoted A and B, that represent local-area networks at opposite ends of a long propagation path. Traffic that is to be passed between A and B is encapsulated in IP (Internet Protocol) packets (e.g., User Data Protocol, UDP). Intercepting this traffic between A and B in both directions, p-ONE time-tags each packet and stores it in memory or on the hard disk of the computer for a user-specified interval that equals the propagation delay to be synthesized. At the expiration of its storage time, each such packet is sent to its destination (that is, if it was received from A, it is sent to B, or vice versa). The accuracy of the p-ONE software is very high, with zero packet loss through the system and negligible latency. Optionally, p-ONE can be configured to delay all network traffic to and from all network addresses on each Ethernet interface or to selectively delay traffic between specific addresses or traffic of specific types. p-ONE works well with Linux and is also designed to be compatible with other operating systems.

Seibert, Marc↗
NASA NTRS1974

Performance of convolutional codes on fading channels typical of planetary entry missions

The performance of convolutional codes in fading channels typical of the planetary entry channel is examined in detail. The signal fading is due primarily to turbulent atmospheric scattering of the RF signal transmitted from an entry probe through a planetary atmosphere. Short constraint length convolutional codes are considered in conjunction with binary phase-shift keyed modulation and Viterbi maximum likelihood decoding, and for longer constraint length codes sequential decoding utilizing both the Fano and Zigangirov-Jelinek (ZJ) algorithms are considered. Careful consideration is given to the modeling of the channel in terms of a few meaningful parameters which can be correlated closely with theoretical propagation studies. For short constraint length codes the bit error probability performance was investigated as a function of E sub b/N sub o parameterized by the fading channel parameters. For longer constraint length codes the effect was examined of the fading channel parameters on the computational requirements of both the Fano and ZJ algorithms. The effects of simple block interleaving in combatting the memory of the channel is explored, using the analytic approach or digital computer simulation.

Modestino, J. W.↗
DOE OSTI2022

The Exploitation of Data Reduction for Visualization

The disparity between the computational speed and storage bandwidth, as demonstrated in Figure 1, is a well known problem that grows with each successive generation. The visualization community is principally responding to this issue by using in situ to reduce which data must be written to storage. However, other communities are taking different, possibly complementary approaches. In particular, data compression is a common general approach to reduce storage demands. Data compression technologies are typically not designed with post processing in mind. The principal metrics measured are compression ratio, the improved bandwidth to storage, and the error introduced. It is assumed that data is inflated to its full size before any post processing can happen. Although when talking about bandwidth disparities, HPC’s dirty little secret is that no part of the memory nor interconnect hardware is increasing at the rate of computation. For example, the Summit supercomputer has a peak computation rate almost 10 times its predecessor, Titan, but only about 4 times the memory, less than twice the aggregate memory bandwidth, and almost no improvement in the interconnect bisection bandwidth. Naively inflating data for post processing does not help with limitations in the memory and interconnect systems.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2007

Two-Way Communication Using RFID Equipment and Techniques

Equipment and techniques used in radio-frequency identification (RFID) would be extended, according to a proposal, to enable short-range, two-way communication between electronic products and host computers. In one example of a typical contemplated application, the purpose of the short-range radio communication would be to transfer image data from a user s digital still or video camera to the user s computer for recording and/or processing. The concept is also applicable to consumer electronic products other than digital cameras (for example, cellular telephones, portable computers, or motion sensors in alarm systems), and to a variety of industrial and scientific sensors and other devices that generate data. Until now, RFID has been used to exchange small amounts of mostly static information for identifying and tracking assets. Information pertaining to an asset (typically, an object in inventory to be tracked) is contained in miniature electronic circuitry in an RFID tag attached to the object. Conventional RFID equipment and techniques enable a host computer to read data from and, in some cases, to write data to, RFID tags, but they do not enable such additional functions as sending commands to, or retrieving possibly large quantities of dynamic data from, RFID-tagged devices. The proposal would enable such additional functions. The figure schematically depicts an implementation of the proposal for a sensory device (e.g., a digital camera) that includes circuitry that converts sensory information to digital data. In addition to the basic sensory device, there would be a controller and a memory that would store the sensor data and/or data from the controller. The device would also be equipped with a conventional RFID chipset and antenna, which would communicate with a host computer via an RFID reader. The controller would function partly as a communication interface, implementing two-way communication protocols at all levels (including RFID if needed) between the sensory device and the memory and between the host computer and the memory. The controller would perform power V

Jedry, Thomas↗
NASA NTRS1972

LSI computer design - SUMC/DV.

Advanced Technology Laboratories has designed, built, and tested the first CMOS general-purpose computer. The computer executes 32 instructions that are operation-code compatible with the IBM 360 and RCA Spectra 70 series of computers. The computer is supported by 2k words of main-memory LSI storage and a TTY peripheral for input and output operations. Additionally, a simulator, an assembler, and a loader have been constructed to provide software support to the hardware. This paper provides an overall picture of this computer project.

Clapp, W. A.↗
DOE OSTI2023

Software-Hardware Co-design of Heterogeneous SmartNIC System for Recommendation Models Inference and Training

Deep Learning Recommendation Models (DLRMs) are critical applications in various domains and have evolved as one of the single largest machine learning applications. Trillions of DLRM parameters exceed the on-chip memory capacity of GPUs. Large-scale multi-node systems are required for distributed DLRM inference and training, which suffer from the all-to-all communication bottleneck, mainly limiting the scalability of ever-growing DLRMs. In recent years, SmartNICs have evolved with coupled computation and communication capabilities providing opportunities for a powerful heterogeneous device in the system. However, there isn't such a distributed system that fully leverages the abundant smartNIC resources that resolve the scalability issue of DLRMs. In this work, we proposed a software-hardware co-design of a heterogeneous smartNIC system that resolves the communication bottleneck of distributed DLRMs, mitigates the memory bandwidth pressure, and improves computation efficiency. We provide a set of smartNIC designs of cache systems (including local cache and remote cache) and smartNIC computation kernels which reduce data movement, relieve memory lookup intensity, and improve the GPU's computation efficiency. In addition, we propose a graph algorithm that improves the data locality of queries within batches which optimizes the overall system performance with higher data reuse. Our evaluation shows that our system achieves 2.1x latency speedup for inference and 1.6x throughput speedup for training.

Guo, Anqi↗
NASA NTRS1993

Parallel spatial direct numerical simulations on the Intel iPSC/860 hypercube

The implementation and performance of a parallel spatial direct numerical simulation (PSDNS) approach on the Intel iPSC/860 hypercube is documented. The direct numerical simulation approach is used to compute spatially evolving disturbances associated with the laminar-to-turbulent transition in boundary-layer flows. The feasibility of using the PSDNS on the hypercube to perform transition studies is examined. The results indicate that the direct numerical simulation approach can effectively be parallelized on a distributed-memory parallel machine. By increasing the number of processors nearly ideal linear speedups are achieved with nonoptimized routines; slower than linear speedups are achieved with optimized (machine dependent library) routines. This slower than linear speedup results because the Fast Fourier Transform (FFT) routine dominates the computational cost and because the routine indicates less than ideal speedups. However with the machine-dependent routines the total computational cost decreases by a factor of 4 to 5 compared with standard FORTRAN routines. The computational cost increases linearly with spanwise wall-normal and streamwise grid refinements. The hypercube with 32 processors was estimated to require approximately twice the amount of Cray supercomputer single processor time to complete a comparable simulation; however it is estimated that a subgrid-scale model which reduces the required number of grid points and becomes a large-eddy simulation (PSLES) would reduce the computational cost and memory requirements by a factor of 10 over the PSDNS. This PSLES implementation would enable transition simulations on the hypercube at a reasonable computational cost.

Joslin, Ronald D.↗
DOE OSTI2025

Emulation of Synaptic Plasticity in WO 3 ‐Based Ion‐Gated Transistors

Neuromorphic systems, inspired by the human brain, promise significant advancements in computational efficiency and power consumption by integrating processing and memory functions, thereby addressing the von Neumann bottleneck. This paper explores the synaptic plasticity of a WO3-based ion-gated transistor (IGT) in [EMIM][TFSI] and a 0.1 mol L −1 LiTFSI in [EMIM][TFSI] for neuromorphic computing applications. Cyclic voltammetry (CV), transistor characteristics, and atomic force microscopy (AFM) force–distance (FD) profiling analyses reveal that Li + brings about ion intercalation, together with higher mobility and conductance, and slower response time (τ). WO 3 IGTs exhibit spike amplitude-dependent plasticity (SADP), spike number-dependent plasticity (SNDP), spike duration-dependent plasticity (SDDP), frequency-dependent plasticity (FDP), and paired-pulse facilitation (PPF), which are all crucial for mimicking biological synaptic functions and understanding how to achieve different types of plasticity in the same IGT. The findings underscore the importance of selecting the appropriate ionic medium to optimize the performance of synaptic transistors, enabling the development of neuromorphic systems capable of adaptive learning and real-time processing, which are essential for applications in artificial intelligence (AI).

36 MATERIALS SCIENCE↗
DOE OSTI2022

Scalable access to shared files in a distributed system

Logic may pre-populate cache of compute nodes allocated to a user's session. Logic may access an image file of a global filesystem, the image file to comprise a first set of one or more files associated with a user session and a second set of one or more files upon which execution of the first set will depend. Logic may store the image file in local data storage of each compute node of a spanning tree of compute nodes allocated for the user session. Logic may unpack the image file based on the data. Logic may configure the local environment of each compute node in the spanning tree to identify a location in local memory of each of the files. And logic may initiate execution of a local filesystem service in each of the compute nodes of the spanning tree.

Oganezov, Alexander↗
DOE OSTI2025

Self‐Strain Suppression of the Metal‐to‐Insulator Transition in Phase‐Change Oxide Devices

Strongly correlated materials exhibiting phase transitions which can be controlled through external stimuli, such as electric fields, are promising for future computing technologies beyond conventional semiconductor transistors. Devices that take advantage of structural phase transitions have inherent built‐in memory, reminiscent of synapses and neurons, and are thus natural candidates for neuromorphic computing. Of particular interest are phase‐change oxides, which allow for control over the metal‐to‐insulator transition. Here, X‐ray nano‐diffraction structural imaging of micro‐devices fabricated with the archetypal phase‐change material vanadium sesquioxide (V 2 O 3 ) is reported. The devices contain a Ga ion‐irradiated region where the metal‐to‐insulator transition critical temperature is lowered, a useful feature for controlling neuron‐like spiking behavior. Results show that strain, induced by crystal lattice mismatch between the pristine and irradiated material, leads to a suppression of the metal‐to‐insulator‐transition. Suppression occurs within the irradiated region or along its edges, depending on the defect‐distribution and the size of the region. The observed self‐straining effect can extend to other phase‐change oxides and dominate as device dimensions are reduced and become too small to dissipate strain within the irradiated region. The findings are important for phase engineering in phase‐change devices and highlight the necessity to study phase transitions at the nanoscale.

77 NANOSCIENCE AND NANOTECHNOLOGY↗
NASA NTRS1985

Complementary-Logic Fault Detector

Circuit for checking two-line complementary-logic bits for single faults used as building block for self-checking memory interface for Hammingcoded data. Intended for such applications as fault-tolerant computing, data handling, and data transmission. Circuit performs exclusive-OR function. Many such circuits combined produce complete memory interface with both detection and correction abilities.

Wawrzynek, J. C.↗