Multiple Parallel Processing Research Articles

Numerical Techniques For Modeling Doppler Ultrasound Spectra Systems

Evaluation of blood-flow Doppler ultrasound spectral content is currently performed on clinical diagnosis. Since mean frequency and bandwidth spectral parameters are determinants on the quantification of stenotic degree, more precise estimators than the conventional Fourier transform should be seek. This paper summarizes studies led by the author in this field, as well as the strategies used to implement the methods in real-time. Regarding stationary and nonstationary characteristics of the blood-flow signal, different models were assessed. When autoregressive and autoregressive moving average models were compared with the traditional Fourier based methods in terms of their statistical performance while estimating both spectral parameters, the Modified Covariance model was identified by the cost/benefit criterion as the estimator presenting better performance. The performance of three time-frequency distributions and the Short Time Fourier Transform was also compared. The Choi–Williams distribution proved to be more accurate than the other methods. The identified spectral estimators were developed and optimized using high performance techniques. Homogeneous and heterogeneous architectures supporting multiple instruction multiple data parallel processing were essayed. Results obtained proved that real-time implementation of the blood-flow estimators is feasible, enhancing the usage of more complex spectral models on other ultrasonic systems.

NUMERICAL TECHNIQUES FOR MODELING DOPPLER ULTRASOUND SPECTRA SYSTEMS

Evaluation of blood-flow Doppler ultrasound spectral content is currently performed on clinical diagnosis. Since mean frequency and bandwidth spectral parameters are determinants on the quantification of stenotic degree, more precise estimators than the conventional Fourier transform should be seek. This paper summarizes studies led by the author in this field, as well as the strategies used to implement the methods in real-time. Regarding stationary and nonstationary characteristics of the blood-flow signal, different models were assessed. When autoregressive and autoregressive moving average models were compared with the traditional Fourier based methods in terms of their statistical performance while estimating both spectral parameters, the Modified Covariance model was identified by the cost/benefit criterion as the estimator presenting better performance. The performance of three time-frequency distributions and the Short Time Fourier Transform was also compared. The Choi–Williams distribution proved to be more accurate than the other methods. The identified spectral estimators were developed and optimized using high performance techniques. Homogeneous and heterogeneous architectures supporting multiple instruction multiple data parallel processing were essayed. Results obtained proved that real-time implementation of the blood-flow estimators is feasible, enhancing the usage of more complex spectral models on other ultrasonic systems.

Three-dimensional echocardiography in congenital heart disease.

Recent advances in transducer technology and image processing have led to the development of two techniques for three-dimensional (3-D) echocardiography: 1) 3-D reconstruction and 2) real-time 3-D (RT3-D) volumetric imaging. 3-D reconstruction creates a 3-D data set from a series of two-dimensional (2-D) images. RT3-D echocardiography uses a 2-D matrix phased array transducer with multiple parallel processing to produce real-time volumetric images of the heart. Both technologies produce novel views of congenital heart defects and offer improved quantification of ventricular volume, mass, and function. The main advantage of RT3-D imaging is its ability to capture 3-D data in real time. This avoids the motion artifact inherent with any reconstructive technique and permits analysis of events during a single cardiac cycle; however, at present, RT3-D imaging has poorer image quality and lacks the Doppler capability. Further development in both techniques will allow 3-D echocardiography to have more widespread clinical applicability.

Method and apparatus for controlling a digital computer using oral input

Oral input is used to control a digital computer. Associative searching techniques of tabular data structures are used in conjunction with rules and conventions derived from natural language to facilitate the use of oral input. The method is capable of being implemented in connection with conventional sequential computers, associative single-instruction multiple data computers and parallel processors.

On a methodology for discrete–continuous scheduling

Discrete–continuous problems of scheduling nonpreemptable jobs on parallel machines are considered. The problems arise e.g. when jobs are assigned to multiple parallel processors driven by a common electric, hydraulic or pneumatic power source. Existing models have assumed job processing rates as a function of the number of jobs currently being processed, or equivalently the number of machines currently in operation. In this paper a more general model is proposed in which processing rates of a job assigned to a machine depend on the amount of a continuous, i.e. continuously divisible resource (e.g. power) allotted to this job at a time. Thus the problem consists of two interrelated subproblems: (i) to sequence jobs on machines, and (ii) to allocate the continuous resource among jobs already sequenced. We provide a comprehensive analysis of the problem. This includes properties of optimal schedules, efficiently (in particular analytically) solvable cases, formulations of the possibly simplest mathematical programming problems for finding optimal schedules in the general case, heuristics and the worst-case analysis. Although our objective function in this paper is to minimize makespan of a set of independent jobs, the presented methodology can be applied to other criteria, precedence-related jobs, and many resource types (apart from, or instead of machines).

Mission reliability of an automatic control system integrated with distributed intelligent built-in-test systems

This paper introduces distributed-centralized Built-in-Test (BIT) systems interfaced with an automatic control system with the purpose of improving mission reliability. By Using a block diagramming method, a complicated system is decomposed into mutually exclusive subsystems so that a distributed BIT is connected to each subsystem for multiple parallel processing of fault direction. The data produced by the distributed BITs is sent to a central control processor. We present a Markov process approach to analytically derive the mission reliability of an automatic control fault-tolerant system with distributed BITs. As diagnostic mistakes of the BIT, the false alarm and fault missing of BIT are considered with the malfunction of the BIT itself. Numerical examples are also prepared to evaluate the performance of distributed intelligent BITs, by comparing mission reliabilities corresponding to the variation of design parameters in a time domain.

Planning and Scheduling of Parallel Semicontinuous Processes. 2. Short-Term Scheduling

In part 1 of this series, it was seen how minimum run length constraints may complicate conventional multiperiod models. For short-term scheduling, these constraints, along with sequencing issues, become even more critical, and part 2 explores the application of continuous time formulations to this class of problems. Two mixed-integer linear programming (MIP) formulations are presented for the detailed short-term scheduling of a single-stage multiproduct facility with multiple parallel semicontinuous processors. Given a set of due dates, demands for products at these dates, and several operational and topological constraints, it is desired to determine the start and end times of individual product campaigns and machine assignments to minimize the inventory, transition, and shortage costs. The key feature of these formulations is that they can accommodate prespecified discrete time events such as due dates while employing a continuous representation of time. In addition to regular production, various outages such as tests and maintenance can be scheduled and a variety of transitions between operations can be accommodated. For purposes of illustration, a (relatively) small real plant scheduling problem is solved.

A multiple-tree search procedure for the resource-constrained project scheduling problem

For over three decades, researchers have sought effective solution procedures for PERT/CPM types of scheduling problems under conditions of limited resource availability. This paper presents a procedure for this problem which takes advantage of the emerging technology provided by multiple parallel processors to find and verify an optimal schedule for a project under conditions of multiple resource constraints. In our approach, multiple solutions trees are searched simultaneously in the quest for a minimum duration schedule. Global upper and lower bound information in common memory is shared among processors, enabling one or several processors to prune potentially significant portions of its search tree based upon bounds discovered by a processor using a different search tree. Computational experience is reported both for problems in which resources are available in constant amounts per period, as well as the much more difficult problem in which the resources available are allowed to vary over the schedule horizon (e.g., travel, sick leave, assignment to other tasks or projects, and so forth). The modular multiple-tree search procedure described in this paper is quite general, permitting most types of existing serial search strategies to be adapted to this approach where multiple processors are available.

Multiple brain systems generating the rat auditory evoked potential. I. Characterization of the auditory cortex response

The objectives of this study were to characterize the auditory cortex response in the rat and to examine its contributions to the auditory evoked potentials (AEPs) recorded from the dorsal and lateral skull. This was accomplished by simultaneously recording AEPs from the cortical surface and from skull screw electrodes in anesthetized animals. The initial positive-negative response (P17-N32) was largely restricted to the cortical region corresponding to area 41. More detailed examination of the AEP mapping revealed multiple subcomponents (P9, P14, P17, P19) underlying the initial positivity, with differing topographies. Stimulus-response properties further dissociated the multiple positive subcomponents Reversible local neurochemical suppression confirmed the auditory cortical origin of these AEPs. The auditory cortex-generated AEPs were refractory to barbiturate anesthesia which eliminated all dorsal skull AEPs, indicating that primary auditory cortical AEPs do not make a significant contribution to the dorsal skull-recorded (‘vertex’) AEPs. The findings raise issues regarding multiple parallel auditory processing systems and their associated AEPs.

Multiple brain systems generating the rat auditory evoked potential. II. Dissociation of auditory cortex and non-lemniscal generator systems

This study addressed the issue of multiple parallel auditory processing systems and their relationship to the skull-recorded auditory evoked potentials (AEPs) in the unanesthetized, unrestrained rat. In the preceding paper ( Brain Res., 602 (1993) 240–250) it has been shown that auditory cortex activity does not contribute significantly to the vertex maximal AEPs recorded from the dorsal skull of the rat. In the present study, mapping of the AEP skull distribution revealed two sets of components: one set maximal at the dorsal skull vertex, and another set maximal at the lateral skull over auditory cortex. Barbiturate anesthesia eliminated the dorsal skull maximal components (and later components at the lateral skull), but not the early (P7–P11, N15) lateral skull components generated in auditory cortex. Bilateral auditory cortex ablation eliminated the lateral skull maximal AEP components, but not the dorsal skull maximal components. These findings support extensive parallel processing of auditory inputs (reflected by the dorsal AEPs) in the absence of primary auditorycortex. Ablation of primary auditory cortex did result in a modulation of the dorsal skull AEPs, indicative of an interaction between the geniculocortical system and the parallel system which generates the dorsal AEPs.

Stochastically Minimizing Total Delay of Jobs Subject to Random Deadlines

This paper analyzes a scheduling system where a fixed number of nonpreemptive jobs is to be processed on multiple parallel processors with different processing speeds. Each processor has an exponential processing time distribution and the processors are ordered in ascending order of their mean processing times. Each job has its own deadline that is exponentially distributed with rate ß1, independent of the deadlines of other jobs and also independent of job processing times. A job departs the system as soon as either its processing completes or its deadline occurs. We show that there exists a simple threshold strategy that slochastically minimizes the total delay of all jobs. The policy depends on distributions of processing times and deadlines, but is independent of the rate of deadlines. When the rate of the deadline distribution is 0 (no deadlines), the total delay reduces to the flowtime (the sum of completion times of all jobs). If each job has its own probability of being correctly processed, then an extension of this policy stochastically maximizes the total number of correctly processed, nontardy jobs. We discuss possible generalizations and limitations of this result.

Graphics processing on a graphics supercomputer

A description is given of the Titan Graphics Supercomputer. The primary design philosophy was to have as little redundant hardware as possible and to make as much of the hardware available to compiled application code as possible. This led to a design with multiple parallel processors, each with an integer unit and a vector floating-point unit. The system provides 3D graphics and image processing, using the main processors for most computations. The graphics subsystem consists of only a frame buffer with rasterization (vector and triangle pixel drawing) support. The Titan architecture provides a very good balance of computation and graphics but to make it more competitive in graphics or imaging-intensive applications, a special-purpose accelerator also fits into the architecture. >

Using multiple CPUs for problem solving: Experiences in multitasking on the CRAY X-MP/48

The integration of vector computers into multiprocessor configurations allows the use of multiple high-speed processors in parallel for one program. There are two aspects which are considered to be important for an efficient use of multiprocessor configurations. First, the flexibility, speed and user friendliness of the available synchronization and communication primitives, and second, the user problems in detecting data dependencies and in translating programs correctly into the parallel form required by the system. This paper is intended to give an overview of our experiences in multitasking using up to four CPUs of a CRAY X-MP/48. The results gained by macrotasking and microtasking will be compared for program kernels and real-life application programs. Special attention is paid to the difficulties of using more than two CPUs in parallel.

Parallel Machine Scheduling: Processing Rates Dependent on Number of Jobs in Operation

We treat the class of n job − m machine scheduling problems with job processing times dependent on the number of jobs being simultaneously processed in the system at any point in time. Such systems occur when jobs are assigned to multiple parallel processors driven by a common power source. In situations typical of hydraulic and pneumatic power sources the level of power delivered to each processor is inversely proportional to the number of processors simultaneously at work. Aside from the variable processing rate assumptions, the remaining assumptions on the structure of the system conform to those of the standard identical parallel processor problem without job preemption. Although the standard n job − m machine problem is NP-hard with respect to a makespan measure, this is not the case when our seemingly complicating variable processing rate function, typical of hydraulic power sources, is included. In this case our main results are exceedingly simple. The makespan is found to be independent of the job-machine assignment, and the flowtime is minimized using one out of m processors. When other job processing rate functions or constant switch-over times are considered, then solving the problem becomes increasingly difficult, and the use of multiple processors is recommended. This paper concludes with a description of a real-world problem (scheduling the refueling of navy boats) which motivated this research.

Status of Vectorized Monte Carlo for Particle Transport Analysis

The conventional particle transport Monte Carlo algorithm is ill suited for modem vector supercomputers because the random nature of the particle transport process in the history based algorithm in hibits construction of vectors. An alterna tive, event-based algorithm is suitable for vectorization and has been used recently to achieve impressive gains in perfor mance on vector supercomputers. This re view describes the event-based algorithm and several variations of it Implementa tions of this algorithm for applications in particle transport are described, and their relative merits are discussed. The imple mentation of Monte Carlo methods on multiple vector parallel processors is con sidered, as is the potential of massively parallel processors for Monte Carlo par ticle transport simulations.

Single‐ and multiple‐processor models for minimizing completion time variance

AbstractThis article concerns the scheduling of n jobs around a common due date, so as to minimize the average total earliness plus total lateness of the jobs. Optimality conditions for the problem are developed, based on its equivalence to an easy scheduling problem. It seems that this problem inherently has a huge number of optimal solutions and an algorithm is developed to find many of them. The model is extended to allow for the availability of multiple parallel processors and an efficient algorithm is developed for that problem. In this more general case also, the algorithm permits great flexibility in finding an optimal schedule.

Performance evaluation of the binary tree access mechanism in mimd type parallel computers

AbstractOne of the most serious problems in a highly multiple parallel processing system involves memory access contention. In the hierarchically structured access mechanism, the number of simultaneous access paths increases with degree of parallelism. Consequently, it is considered as a configuration suited to a highly multiple parallel system. This paper uses a queue model to analyze the hierarchical routing bus (H‐R bus), which is an access mechanism with a hierarchical structure. The paper further makes a quantitative evaluation of the locality in data access. Based on the derived theoretical expressions and measured results obtained by executing several test programs on the H‐R bus‐connected parallel computer, the H‐R bus performance is evaluated. The results obtained in this paper are as follows: (1) Low external data access rates (i.e., external data access time/processing time) indicate that the H‐R bus‐connected multiprocessor system operates with high efficiency even under high multiplicity. (2) Under sufficient access locality, even high external data access rates do not unduly impair efficiency. Although the above properties have qualitatively been known before, this paper derives theoretical expressions for the properties, supported by experiments.

Multiprocessor Implementation of Adaptive Digital Filters

Adaptive filters, employing the transversal filter structure and the least mean square (LMS) adaptation algorithm, or its variations, have found wide application in data transmission equalization, echo cancellation, prediction, spectral estimation, on-line system identification, and antenna arrays. Recently, in response to requirements of fast start-up, or fast tracking of temporal variations, fast recursive least squares (FRLS) adaptation algorithms for both transversal and lattice filter structures have been proposed. These algorithms offer faster convergence than is possible with the LMS/ transversal adaptive filters, at the price of a five-to-tenfold increase in the number of multiplications, divisions, and additions. Here we discuss architectures and implementations of the LMS/transversal, fast-converging FRLS filter, and lattice filter algorithms which minimize the required hardware speed. We show how each of these algorithms can be partitioned so as to be realizable with an architecture based on multiple parallel processors.

A Discussion of Digital Processing in Synthetic Aperture Radar

This is a summary paper describing the processing of synthetic aperture radar (SAR) data using digital correlation algorithms. Fundamental SAR theory as it applies to the various SAR modes, namely, strip mapping, spotlight mapping, and Doppler beam sharpened mapping, is described and a baseline design applicable to all SAR modes is presented. Digital processor design is developed, starting with a simple single filter mechanization and proceeding through more complex processing algorithms. Prefilter design is discussed, as is the more advanced processing algorithms, namely, multiple parallel prefilters, two-stage correlation, and FFT processing. The primary processor tradeoff is increased functional complexity versus reduced arithmetic and memory requirements. For high-resolution applications, the arithmetic requirements can be reduced by an order of magnitude or more by implementing the more advanced processing algorithms.