Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “ARRAY PROCESSORS”

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 307 records · Page 17

Unstructured Adaptive Grid Computations on an Array of SMPs

Dynamic load balancing is necessary for parallel adaptive methods to solve unsteady CFD problems on unstructured grids. We have presented such a dynamic load balancing framework called JOVE, in this paper. Results on a four-POWERnode POWER CHALLENGEarray demonstrated that load balancing gives significant performance improvements over no load balancing for such adaptive computations. The parallel speedup of JOVE, implemented using MPI on the POWER CHALLENCEarray, was significant, being as high as 31 for 32 processors. An implementation of JOVE that exploits 'an array of SMPS' architecture was also studied; this hybrid JOVE outperformed flat JOVE by up to 28% on the meshes and adaption models tested. With large, realistic meshes and actual flow-solver and adaption phases incorporated into JOVE, hybrid JOVE can be expected to yield significant advantage over flat JOVE, especially as the number of processors is increased, thus demonstrating the scalability of an array of SMPs architecture.

Biswas, Rupak↗

A cost-effective methodology for the design of massively-parallel VLSI functional units

In this paper we propose a generalized methodology for the design of cost-effective massively-parallel VLSI Functional Units. This methodology is based on a technique of generating and reducing a massive bit-array on the mask-programmable PAcube VLSI array. This methodology unifies (maintains identical data flow and control) the execution of complex arithmetic functions on PAcube arrays. It is highly regular, expandable and uniform with respect to problem-size and wordlength, thereby reducing the communication complexity. The memory-functional unit interface is regular and expandable. Using this technique functional units of dedicated processors can be mask-programmed on the naked PAcube arrays, reducing the turn-around time. The production cost of such dedicated processors can be drastically reduced since the naked PAcube arrays can be mass-produced. Analysis of the the performance of functional units designed by our method yields promising results.

Venkateswaran, N.↗

Evaluation of the OpenCL AES Kernel using the Intel FPGA SDK for OpenCL

The OpenCL standard is an open programming model for accelerating algorithms on heterogeneous computing system. OpenCL extends the C-based programming language for developing portable codes on different platforms such as CPU, Graphics processing units (GPUs), Digital Signal Processors (DSPs) and Field Programmable Gate Arrays (FPGAs). The Intel FPGA SDK for OpenCL is a suite of tools that allows developers to abstract away the complex FPGA-based development flow for a high-level software development flow. Users can focus on the design of hardware-accelerated kernel functions in OpenCL and then direct the tools to generate the low-level FPGA implementations. The approach makes the FPGA-based development more accessible to software users as the needs for hybrid computing using CPUs and FPGAs are increasing. It can also significantly reduce the hardware development time as users can evaluate different ideas with high-level language without deep FPGA domain knowledge. In this report, we evaluate the performance of the kernel using the Intel FPGA SDK for OpenCL and Nallatech 385A FPGA board. Compared to the M506 module, the board provides more hardware resources for a larger design exploration space. The kernel performance is measured with the compute kernel throughput, an upper bound to the FPGA throughput. The report presents the experimental results in details. The Appendix lists the kernel source code.

97 MATHEMATICS AND COMPUTING↗

Evaluation of CHO Benchmarks on the Arria 10 FPGA using Intel FPGA SDK for OpenCL

The OpenCL standard is an open programming model for accelerating algorithms on heterogeneous computing system. OpenCL extends the C-based programming language for developing portable codes on different platforms such as CPU, Graphics processing units (GPUs), Digital Signal Processors (DSPs) and Field Programmable Gate Arrays (FPGAs). The Intel FPGA SDK for OpenCL is a suite of tools that allows developers to abstract away the complex FPGA-based development flow for a high-level software development flow. Users can focus on the design of hardware-accelerated kernel functions in OpenCL and then direct the tools to generate the low-level FPGA implementations. The approach makes the FPGA-based development more accessible to software users as the needs for hybrid computing using CPUs and FPGAs are increasing. It can also significantly reduce the hardware development time as users can evaluate different ideas with high-level language without deep FPGA domain knowledge. Benchmarking of OpenCL-based framework is an effective way for analyzing the performance of system by studying the execution of the benchmark applications. CHO is a suite of benchmark applications that provides support for OpenCL [1]. The authors presented CHO as an OpenCL port of the CHStone benchmark. Using Altera OpenCL (AOCL) compiler to synthesize the benchmark applications, they listed the resource usage and performance of each kernel that can be successfully synthesized by the compiler. In this report, we evaluate the resource usage and performance of the CHO benchmark applications using the Intel FPGA SDK for OpenCL and Nallatech 385A FPGA board that features an Arria 10 FPGA device. The focus of the report is to have a better understanding of the resource usage and performance of the kernel implementations using Arria-10 FPGA devices compared to Stratix-5 FPGA devices. In addition, we also gain knowledge about the limitations of the current compiler when it fails to synthesize a benchmark application.

97 MATHEMATICS AND COMPUTING↗

Onboard Experiment Data Support Facility

An onboard array structure has been devised for end to end processing of data from multiple spaceborne sensors. The array constitutes sets of programmable pipeline processors whose elements perform each assigned function in 0.25 microseconds. This space shuttle computer system can handle data rates from a few bits to over 100 megabits per second.

Source record↗

A method for classifying multispectral remote sensing data using context

The paper describes a method of classifying multispectral remote sensing data using a context classifier. Because the computational requirements of the context classifier are very large, its implementation on parallel/pipelined multiprocessor systems is being investigated, so that the types of computations can be efficiently implemented. Special considerations necessary for such implementations are discussed, with particular reference to implementation on an array of Control Data Corporation Flexible Processors.

Swain, P. H.↗

Rain rate instrument for deployment at sea, phase 2

This report describes, in detail, the SBIR Phase 2 contracting effort provided for by NASA Contract Number NAS8-38481 in which a prototype Rain Rate Sensor was developed. FWG Model RP101A is a fully functional rain rate and droplet size analyzing instrument. The RP101A is a fully functional rain rate and droplet size analyzing instrument. The RP101A consists of a fiber optic probe containing a 32-fiber array connected to an electronic signal processor. When interfaced to an IBM compatible personal computer and configured with appropriate software, the RP101A is capable of measuring rain rates and particles ranging in size from around 300 microns up to 6 to 7 millimeters. FWG Associates, Inc. intends to develop a production model from the prototype and continue the effort under NASA's SBIR Phase 3 program.

Steele, Jimmy W.↗

Global synchronization algorithms for the Intel iPSC/860

In a distributed memory multicomputer that has no global clock, global processor synchronization can only be achieved through software. Global synchronization algorithms are used in tridiagonal systems solvers, CFD codes, sequence comparison algorithms, and sorting algorithms. They are also useful for event simulation, debugging, and for solving mutual exclusion problems. For the Intel iPSC/860 in particular, global synchronization can be used to ensure the most effective use of the communication network for operations such as the shift, where each processor in a one-dimensional array or ring concurrently sends a message to its right (or left) neighbor. Three global synchronization algorithms are considered for the iPSC/860: the gysnc() primitive provided by Intel, the PICL primitive sync0(), and a new recursive doubling synchronization (RDS) algorithm. The performance of these algorithms is compared to the performance predicted by communication models of both the long and forced message protocols. Measurements of the cost of shift operations preceded by global synchronization show that the RDS algorithm always synchronizes the nodes more precisely and costs only slightly more than the other two algorithms.

Seidel, Steven R.↗

Pre-Hardware Optimization and Implementation Of Fast Optics Closed Control Loop Algorithms

One of the main heritage tools used in scientific and engineering data spectrum analysis is the Fourier Integral Transform and its high performance digital equivalent - the Fast Fourier Transform (FFT). The FFT is particularly useful in two-dimensional (2-D) image processing (FFT2) within optical systems control. However, timing constraints of a fast optics closed control loop would require a supercomputer to run the software implementation of the FFT2 and its inverse, as well as other image processing representative algorithm, such as numerical image folding and fringe feature extraction. A laboratory supercomputer is not always available even for ground operations and is not feasible for a night project. However, the computationally intensive algorithms still warrant alternative implementation using reconfigurable computing technologies (RC) such as Digital Signal Processors (DSP) and Field Programmable Gate Arrays (FPGA), which provide low cost compact super-computing capabilities. We present a new RC hardware implementation and utilization architecture that significantly reduces the computational complexity of a few basic image-processing algorithm, such as FFT2, image folding and phase diversity for the NASA Solar Viewing Interferometer Prototype (SVIP) using a cluster of DSPs and FPGAs. The DSP cluster utilization architecture also assures avoidance of a single point of failure, while using commercially available hardware. This, combined with the control algorithms pre-hardware optimization, or the first time allows construction of image-based 800 Hertz (Hz) optics closed control loops on-board a spacecraft, based on the SVIP ground instrument. That spacecraft is the proposed Earth Atmosphere Solar Occultation Imager (EASI) to study greenhouse gases CO2, C2H, H2O, O3, O2, N2O from Lagrange-2 point in space. This paper provides an advanced insight into a new type of science capabilities for future space exploration missions based on on-board image processing for control and for robotics missions using vision sensors. It presents a top-level description of technologies required for the design and construction of SVIP and EASI and to advance the spatial-spectral imaging and large-scale space interferometry science and engineering.

Kizhner, Semion↗

Single-Scale Retinex Using Digital Signal Processors

The Retinex is an image enhancement algorithm that improves the brightness, contrast and sharpness of an image. It performs a non-linear spatial/spectral transform that provides simultaneous dynamic range compression and color constancy. It has been used for a wide variety of applications ranging from aviation safety to general purpose photography. Many potential applications require the use of Retinex processing at video frame rates. This is difficult to achieve with general purpose processors because the algorithm contains a large number of complex computations and data transfers. In addition, many of these applications also constrain the potential architectures to embedded processors to save power, weight and cost. Thus we have focused on digital signal processors (DSPs) and field programmable gate arrays (FPGAs) as potential solutions for real-time Retinex processing. In previous efforts we attained a 21 (full) frame per second (fps) processing rate for the single-scale monochromatic Retinex with a TMS320C6711 DSP operating at 150 MHz. This was achieved after several significant code improvements and optimizations. Since then we have migrated our design to the slightly more powerful TMS320C6713 DSP and the fixed point TMS320DM642 DSP. In this paper we briefly discuss the Retinex algorithm, the performance of the algorithm executing on the TMS320C6713 and the TMS320DM642, and compare the results with the TMS320C6711.

Hines, Glenn↗

Method and apparatus for ultra-high-sensitivity, incremental and absolute optical encoding

An absolute optical linear or rotary encoder which encodes the motion of an object (3) with increased resolution and encoding range and decreased sensitivity to damage to the scale includes a scale (5), which moves with the object and is illuminated by a light source (11). The scale carries a pattern (9) which is imaged by a microscope optical system (13) on a CCD array (17) in a camera head (15). The pattern includes both fiducial markings (31) which are identical for each period of the pattern and code areas (33) which include binary codings of numbers identifying the individual periods of the pattern. The image of the pattern formed on the CCD array is analyzed by an image processor (23) to locate the fiducial marking, decode the information encoded in the code area, and thereby determine the position of the object.

Leviton, Douglas B.↗

Method and apparatus for optical encoding with compressible imaging

The present invention presents an optical encoder with increased conversion rates. Improvement in the conversion rate is a result of combining changes in the pattern recognition encoder's scale pattern with an image sensor readout technique which takes full advantage of those changes, and lends itself to operation by modern, high-speed, ultra-compact microprocessors and digital signal processors (DSP) or field programmable gate array (FPGA) logic elements which can process encoder scale images at the highest speeds. Through these improvements, all three components of conversion time (reciprocal conversion rate)--namely exposure time, image readout time, and image processing time--are minimized.

Leviton, Douglas B.↗

Absolute Position Encoders With Vertical Image Binning

Improved optoelectronic patternrecognition encoders that measure rotary and linear 1-dimensional positions at conversion rates (numbers of readings per unit time) exceeding 20 kHz have been invented. Heretofore, optoelectronic pattern-recognition absoluteposition encoders have been limited to conversion rates <15 Hz -- too low for emerging industrial applications in which conversion rates ranging from 1 kHz to as much as 100 kHz are required. The high conversion rates of the improved encoders are made possible, in part, by use of vertically compressible or binnable (as described below) scale patterns in combination with modified readout sequences of the image sensors [charge-coupled devices (CCDs)] used to read the scale patterns. The modified readout sequences and the processing of the images thus read out are amenable to implementation by use of modern, high-speed, ultra-compact microprocessors and digital signal processors or field-programmable gate arrays. This combination of improvements makes it possible to greatly increase conversion rates through substantial reductions in all three components of conversion time: exposure time, image-readout time, and image-processing time.

Leviton, Douglas B.↗

Advanced Signal Conditioners for Data-Acquisition Systems

Signal conditioners embodying advanced concepts in analog and digital electronic circuitry and software have been developed for use in data-acquisition systems that are required to be compact and lightweight, to utilize electric energy efficiently, and to operate with high reliability, high accuracy, and high power efficiency, without intervention by human technicians. These signal conditioners were originally intended for use aboard spacecraft. There are also numerous potential terrestrial uses - especially in the fields of aeronautics and medicine, wherein it is necessary to monitor critical functions. Going beyond the usual analog and digital signal-processing functions of prior signal conditioners, the new signal conditioner performs the following additional functions: It continuously diagnoses its own electronic circuitry, so that it can detect failures and repair itself (as described below) within seconds. It continuously calibrates itself on the basis of a highly accurate and stable voltage reference, so that it can continue to generate accurate measurement data, even under extreme environmental conditions. It repairs itself in the sense that it contains a micro-controller that reroutes signals among redundant components as needed to maintain the ability to perform accurate and stable measurements. It detects deterioration of components, predicts future failures, and/or detects imminent failures by means of a real-time analysis in which, among other things, data on its present state are continuously compared with locally stored historical data. It minimizes unnecessary consumption of electric energy. The design architecture divides the signal conditioner into three main sections: an analog signal section, a digital module, and a power-management section. The design of the analog signal section does not follow the traditional approach of ensuring reliability through total redundancy of hardware: Instead, following an approach called spare parts tool box, the reliability of each component is assessed in terms of such considerations as risks of damage, mean times between failures, and the effects of certain failures on the performance of the signal conditioner as a whole system. Then, fewer or more spares are assigned for each affected component, pursuant to the results of this analysis, in order to obtain the required degree of reliability of the signal conditioner as a whole system. The digital module comprises one or more processors and field-programmable gate arrays, the number of each depending on the results of the aforementioned analysis. The digital module provides redundant control, monitoring, and processing of several analog signals. It is designed to minimize unnecessary consumption of electric energy, including, when possible, going into a low-power "sleep" mode that is implemented in firmware. The digital module communicates with external equipment via a personal-computer serial port. The digital module monitors the "health" of the rest of the signal conditioner by processing defined measurements and/or trends. It automatically makes adjustments to respond to channel failures, compensate for effects of temperature, and maintain calibration.

Lucena, Angel↗

Spacecraft on-board SAR image generation for EOS-type missions

Spacecraft on-board synthetic aperture radar (SAR) image generation is an extremely difficult problem because of the requirements for high computational rates (usually on the order of Giga-operations per second), high reliability (some missions last up to 10 years), and low power dissipation and mass (typically less than 500 watts and 100 Kilograms). Recently, a JPL study was performed to assess the feasibility of on-board SAR image generation for EOS-type missions. This paper summarizes the results of that study. Specifically, it proposes a processor architecture using a VLSI time-domain parallel array for azimuth correlation. Using available space qualifiable technology to implement the proposed architecture, an on-board SAR processor having acceptable power and mass characteristics appears feasible for EOS-type applications.

Liu, K. Y.↗

Analog hardware for learning neural networks

This is a recurrent or feedforward analog neural network processor having a multi-level neuron array and a synaptic matrix for storing weighted analog values of synaptic connection strengths which is characterized by temporarily changing one connection strength at a time to determine its effect on system output relative to the desired target. That connection strength is then adjusted based on the effect, whereby the processor is taught the correct response to training examples connection by connection.

Eberhardt, Silvio P.↗

Redundant disk arrays: Reliable, parallel secondary storage

During the past decade, advances in processor and memory technology have given rise to increases in computational performance that far outstrip increases in the performance of secondary storage technology. Coupled with emerging small-disk technology, disk arrays provide the cost, volume, and capacity of current disk subsystems, by leveraging parallelism, many times their performance. Unfortunately, arrays of small disks may have much higher failure rates than the single large disks they replace. Redundant arrays of inexpensive disks (RAID) use simple redundancy schemes to provide high data reliability. The data encoding, performance, and reliability of redundant disk arrays are investigated. Organizing redundant data into a disk array is treated as a coding problem. Among alternatives examined, codes as simple as parity are shown to effectively correct single, self-identifying disk failures.

Gibson, Garth Alan↗

NbN A/D Conversion of IR Focal Plane Sensor Signal at 10 K

We are implementing a 12 bit SFQ counting ADC with parallel-to-serial readout using our established 10 K NbN capability. This circuit provides a key element of the analog signal processor (ASP) used in large infrared focal plane arrays. The circuit processes the signal data stream from a Si:As BIB detector array. A 10 mega samples per second (MSPS) pixel data stream flows from the chip at a 120 megabit bit rate in a format that is compatible with other superconductive time dependent processor (TDP) circuits being developed. We will discuss our planned ASP demonstration, the circuit design, and test results.

conversion↗