Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “computer hardware”

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 1,009 records · Page 56

3D-ReG: A 3D ReRAM-based Heterogeneous Architecture for Training Deep Neural Networks

Deep neural network (DNN) models are being expanded to a broader range of applications. The computational capability of traditional hardware platforms cannot accommodate the growth of model complexity. Among recent technologies to accelerate DNN, resistive memory (ReRAM)-based processing-in-memory (PIM) emerged as a promising solution for DNN inference due to its high efficiency for matrix-based computation. We face two major technical challenges in extending the use of ReRAM-based accelerators for training: (1) full-precision data is essential in back-propagation; (2) the need to support both feed-forward and back-propagation aggravates the data-movement burden. We propose a heterogeneous architecture named as 3D-ReG, which leverages full-precision GPU to ensure training accuracy and low-overhead 3D integration to provide low-cost data movements. Moreover, we introduce conservative and aggressive task-mapping schemes, which partition the computation phases in different ways to balance execution efficiency and training accuracy. We evaluate 3D-ReG implemented with two 3D integration technologies, through-silicon vias (TSVs) and monolithic inter-tier vias (MIVs), and compare them with GPU-only and PIM-only counterparts. Various GPU-only platforms using two main-memory technologies (DRAM, ReRAM) and three interconnect technologies (2D, TSV, MIV) are evaluated as well. Experimental results show that 3D-ReG can achieve on average 5.64× training speedup and 3.56× higher energy efficiency compared with the GPU with DRAM as main memory, at the cost of 0.05%–3.39% accuracy drop. We define a new metric, gain-loss ratio (GLR), which quantitatively evaluates the capability of a DNN training hardware in terms of the model accuracy and hardware efficiency. The results of our comparison show that the aggressive task-mapping scheme on MIV-based 3D-ReG outperforms the other methods.

Computer Science↗

NeuroCoreX: Brain-Inspired Computing from Code to Circuit

NeuroCoreX is an open-source codebase that enables the implementation of brain-inspired, energy-efficient neuromorphic computing models on FPGA hardware. Designed to support real-time learning, all-to-all neural connectivity, and flexible network architectures, NeuroCoreX offers a hands-on, accessible platform for exploring biologically inspired models of neural computation. It empowers researchers, students, and developers to implement and experiment with adaptive systems—bringing the power of neuromorphic computing to a broader community through a low-cost, scalable, and reconfigurable framework.

Gautam, Ashish [Oak Ridge National Laboratory (ORN↗

Design and implementation of a medium speed communications interface and protocol for a low cost, refreshed display computer

The design and implementation of hardware and software systems involved in using a 40,000 bit/second communication line as the connecting link between an IMLAC PDS 1-D display computer and a Univac 1108 computer system were described. The IMLAC consists of two independent processors sharing a common memory. The display processor generates the deflection and beam control currents as it interprets a program contained in the memory; the minicomputer has a general instruction set and is responsible for starting and stopping the display processor and for communicating with the outside world through the keyboard, teletype, light pen, and communication line. The processing time associated with each data byte was minimized by designing the input and output processes as finite state machines which automatically sequence from each state to the next. Several tests of the communication link and the IMLAC software were made using a special low capacity computer grade cable between the IMLAC and the Univac.

Phyne, J. R.↗

A Linux Workstation for High Performance Graphics

The primary goal of this effort was to provide a low-cost method of obtaining high-performance 3-D graphics using an industry standard library (OpenGL) on PC class computers. Previously, users interested in doing substantial visualization or graphical manipulation were constrained to using specialized, custom hardware most often found in computers from Silicon Graphics (SGI). We provided an alternative to expensive SGI hardware by taking advantage of third-party, 3-D graphics accelerators that have now become available at very affordable prices. To make use of this hardware our goal was to provide a free, redistributable, and fully-compatible OpenGL work-alike library so that existing bodies of code could simply be recompiled. for PC class machines running a free version of Unix. This should allow substantial cost savings while greatly expanding the population of people with access to a serious graphics development and viewing environment. This should offer a means for NASA to provide a spectrum of graphics performance to its scientists, supplying high-end specialized SGI hardware for high-performance visualization while fulfilling the requirements of medium and lower performance applications with generic, off-the-shelf components and still maintaining compatibility between the two.

Geist, Robert↗

The CMS Event-Builder System for LHC Run 3 (2021-23)

The data acquisition system (DAQ) of the CMS experiment at the CERN Large Hadron Collider (LHC) assembles events of 2MB at a rate of 100 kHz. The event builder collects event fragments from about 750 sources and assembles them into complete events which are then handed to the highlevel trigger (HLT) processes running on O(1000) computers. The aging eventbuilding hardware will be replaced during the long shutdown 2 of the LHC taking place in 2019/20. The future data networks will be based on 100 Gb/s interconnects using Ethernet and Infiniband technologies. More powerful computers may allow to combine the currently separate functionality of the readout and builder units into a single I/O processor handling simultaneously 100 Gb/s of input and output trac. It might be beneficial to preprocess data originating from specific detector parts or regions before handling it to generic HLT processors. Therefore, we will investigate how specialized coprocessors, e.g. GPUs, could be integrated into the event builder. We will present the envisioned changes to the event-builder compared to today’s system. Initial measurements of the performance of the data networks under the event-building trac pattern will be shown. Implications of a folded network architecture for the event building and corresponding changes to the software implementation will be discussed. Corresponding author: remigius.mommsen@cern.c

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Proceedings, Conference on the Computing Environment for Mathematical Software

Recent advances in software and hardware technology which make it economical to create computing environments appropriate for specialized applications are addressed. Topics included software tools, FORTRAN standards activity, and features of languages, operating systems, and hardware that are important for the development, testing, and maintenance of mathematical software.

Source record↗

Orthopedic metallic hardware in routine abdomino-pelvic CT scans: occurrence and clinical significance

Purpose: To study the occurrence of orthopedic metallic hardware in routine abdomen/pelvic computed tomography (CT) scans and their impact on image quality (IQ) and diagnostic evaluation. Material and methods: In this retrospective single institution study, we analyzed 3500 consecutive abdomen/pelvis CT scans for occurrence of orthopedic metallic hardware. In the cohort of patients with metallic hardware detected on CT scans, subjective and objective IQ analysis was performed to estimate diagnostic acceptability (DA, 4-point scale), subjective noise (SN, 3-point scale), presence of artifacts (PA, 4-point scale) and objective noise. The clinical significance of metallic hardware was determined by evaluating the impact of artifacts on radiological diagnosis according to the clinical indication and disease type. Results: Orthopedic metallic hardware was encountered in 4.97% of abdomino-pelvic CT scans (n = 174/3500), and artifacts related to the hardware in the region of clinical interest were identified in 82% (n = 144/174) of scans. The overall mean DA was 2.66 (n = 174), and it was severely limited (score < 2) in 32% of cases particularly affecting patients with bilateral hip implants (92.6%, n = 25/27). The artifacts due to hardware significantly limited diagnostic evaluation in 58.6% of cases (PA score ≥ 3), and the image noise was unacceptable in 71% of cases (SN score > 2) in the region of clinical interest. Conclusion: Orthopedic metallic hardware is encountered in nearly 5% of abdomino-pelvic CT scans and causes significant image degradation limiting diagnostic evaluation in the region of clinical interest.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

EHW Approach to Temperature Compensation of Electronics

Efforts are under way to apply the concept of evolvable hardware (EHW) to compensate for variations, with temperature, in the operational characteristics of electronic circuits. To maintain the required functionality of a given circuit at a temperature above or below the nominal operating temperature for which the circuit was originally designed, a new circuit would be evolved; moreover, to obtain the required functionality over a very wide temperature range, there would be evolved a number of circuits, each of which would satisfy the performance requirements over a small part of the total temperature range. The basic concepts and some specific implementations of EHW were described in a number of previous NASA Tech Briefs articles, namely, "Reconfigurable Arrays of Transistors for Evolvable Hardware" (NPO-20078), Vol. 25, No. 2 (February 2001), page 36; Evolutionary Automated Synthesis of Electronic Circuits (NPO- 20535), Vol. 26, No. 7 (July 2002), page 37; "Designing Reconfigurable Antennas Through Hardware Evolution" (NPO-20666), Vol. 26, No. 7 (July 2002), page 38; "Morphing in Evolutionary Synthesis of Electronic Circuits" (NPO-20837), Vol. 26, No. 8 (August 2002), page 31; "Mixtrinsic Evolutionary Synthesis of Electronic Circuits" (NPO-20773) Vol. 26, No. 8 (August 2002), page 32; and "Synthesis of Fuzzy-Logic Circuits in Evolvable Hardware" (NPO-21095) Vol. 26, No. 11 (November 2002), page 38. To recapitulate from the cited prior articles: EHW is characterized as evolutionary in a quasi-genetic sense. The essence of EHW is to construct and test a sequence of populations of circuits that function as incrementally better solutions of a given design problem through the selective, repetitive connection and/or disconnection of capacitors, transistors, amplifiers, inverters, and/or other circuit building blocks. The connection and disconnection can be effected by use of field-programmable transistor arrays (FPTAs). The evolution is guided by a search-andoptimization algorithm (in particular, a genetic algorithm) that operates in the space of possible circuits to find a circuit that exhibits an acceptably close approximation of the desired functionality. The evolved circuits can be tested by mathematical modeling (that is, computational simulation) only, tested in real hardware, or tested in combinations of computational simulation and real hardware.

Stoica, Adrian↗

The Other Artificial Intelligence Hardware Problem

The consolidation of leading-edge semiconductor manufacturing in one company in one country poses two problems: a short-term strategic risk and a long-term innovation risk. Furthermore, both problems remain underexplored because advances in artificial intelligence (AI) algorithms have compensated for a relative lack of advances in AI hardware.

97 MATHEMATICS AND COMPUTING↗

HPC - Platforms Penta Chart

Strategy, Planning, Acquiring- very large scale computing platforms come and go and planning for immensely scalable machines often precedes actual procurement by 3 years. Procurement can be another year or more. Integration- After Acquisition, machines must be integrated into the computing environments at LANL. Connection to scalable storage via large scale storage networking, assuring correct and secure operations. Management and Utilization – Ongoing operations, maintenance, and trouble shooting of the hardware and systems software at massive scale is required.

Computer Hardware↗

A NASA Perspective on Quantum Computing: Algorithmic Opportunities and Challenges

In the last couple of decades, the world has seen several stunning instances of quantum algorithms that provably outperform the best classical algorithms. For most problems, however, it is currently unknown whether quantum algorithms can provide an advantage, and if so how to design quantum algorithms that realize such advantages. Today, classical heuristics are used to solve many of the most challenging computational problems arising in the practical world, algorithms that have been shown to be effective empirically but have not been mathematically proven to outperform other approaches. With the advent of quantum advantage, the ability of current quantum hardware to do certain computations beyond the ability of even that largest supercomputers, we have an unprecedented opportunity to explore heuristic quantum algorithms. The next few years will be exciting as empirical testing of quantum heuristic algorithms becomes more and more feasible. The talk will begin overview of the NASA QuAIL team’s ongoing quantum computing investigations, and then focus on both near-term and longer term algorithms for optimization, including distributed algorithms.

quantum computing↗

Design considerations for a 10-kW integrated hydrogen-oxygen regenerative fuel cell system

Integration of an alkaline fuel cell subsystem with an alkaline electrolysis subsystem to form a regenerative fuel cell (RFC) system for low earth orbit (LEO) applications characterized by relatively high overall round trip electrical efficiency, long life, and high reliability is possible with present state of the art technology. A hypothetical 10 kW system computer modeled and studied based on data from ongoing contractual efforts in both the alkaline fuel cell and alkaline water electrolysis areas. The alkaline fuel cell technology is under development utilizing advanced cell components and standard Shuttle Orbiter system hardware. The alkaline electrolysis technology uses a static water vapor feed technique and scaled up cell hardware is developed. The computer aided study of the performance, operating, and design parameters of the hypothetical system is addressed.

Hoberecht, M. A.↗

Design considerations for a 10-kW integrated hydrogen-oxygen regenerative fuel cell system

Integration of an alkaline fuel cell subsystem with an alkaline electrolysis subsystem to form a regenerative fuel cell (RFC) system for low earth orbit (LEO) applications characterized by relatively high overall round trip electrical efficiency, long life, and high reliability is possible with present state of the art technology. A hypothetical 10 kW system computer modeled and studied based on data from ongoing contractual efforts in both the alkaline fuel cell and alkaline water electrolysis areas. The alkaline fuel cell technology is under development utilizing advanced cell components and standard Shuttle Orbiter system hardware. The alkaline electrolysis technology uses a static water vapor feed technique and scaled up cell hardware is developed. The computer aided study of the performance, operating, and design parameters of the hypothetical system is addressed.

Hoberecht, M. A.↗

Implementation of Ada protocols on Mil-STD-1553 B data bus

Standardization activity of data communication in avionic systems started in 1968 for the purpose of total system integration and the elimination of heavy wire bundles carrying signals between various subassemblies. The growing complexity of avionic systems is straining the capabilities of MIL-STD-1553 B (first issued in 1973), but a much greater challenge to it is posed by Ada, the standard language adopted for real-time, computer embedded-systems. Hardware implementation of Ada communication protocols in a contention/token bus or token ring network is proposed. However, during the transition period when the current command/response multiplex data bus is still flourishing and the development environment for distributed multi-computer Ada systems is as yet lacking, a temporary accomodation of the standard language with the standard bus could be very useful and even highly desirable. By concentrating all status informtion and decisions at the bus controller, it was found to be possible to construct an elegant and efficient harware impelementation of the Ada protocols at the bus interface. This solution is discussed.

Ruhman, Smil↗