Considers DP (distributed/parallel) Cy-Clone, which is a further development of previous work in the area of designing highly predictable real-time systems. DP Cy-Clone systems contain only architectures that have wel...
详细信息
Considers DP (distributed/parallel) Cy-Clone, which is a further development of previous work in the area of designing highly predictable real-time systems. DP Cy-Clone systems contain only architectures that have well controlled time behavior describable in terms of time increments (dT). The issues related to designing time-critical distributed/parallel real-time systems and the motivation for the properties of DP Cy-Clone are presented first. The generic architecture of a node and a distributednetwork of nodes is presented, followed by the introduction of one possible approach to a bus-based concrete architecture. Means of network expansion via connecting multiple clusters are presented. Finally, some potential areas of application are described.< >
Continuing improvement in processing speed has made it feasible to solve the Reynolds-Averaged Navier- Stokes equations for simple three-dimensional flows on advanced workstations. Combining multiple workstations into...
详细信息
This paper presents a distributed system connected with an Ethernet network. All the processes and all the processors are identical. The only two resources that processes need are processors and communication among th...
详细信息
This paper presents a distributed system connected with an Ethernet network. All the processes and all the processors are identical. The only two resources that processes need are processors and communication among them. The goal is to obtain the utilization factor of the processors for the distribution and the interprocess communication. The utilization factor can be deduced from an equivalent M/G/1/k model (where k is the number of processes in a processor). We describe a mathematical model for the interprocessor communications (Ethernet network).< >
Describes the use of a fine grain parallel memory system using linear skewing schemes for storing multidimensional arrays of any size, and an addressing scheme based on the Chinese remainder theorem (CRT) that is comp...
详细信息
Describes the use of a fine grain parallel memory system using linear skewing schemes for storing multidimensional arrays of any size, and an addressing scheme based on the Chinese remainder theorem (CRT) that is compatible with the linear mapping. So, the number of memory banks and processing elements is fixed, whereas the size of the multidimensional arrays varies with the application of such a machine. We extend the well-known linear skewing scheme to multidimensional arrays which provide conflict-free access to all vectors of interest. We give the conditions over the range of the stored vector to enable a simple addressing scheme. The address generation is also discussed. To complete the parallel memory system, we present an interconnection network that is adapted to the mapping.< >
The vision system described serves as a sensor for an automated welding process. The chosen strategy of processing for obtaining the needed results in real-time will be explained. To improve the reliability of the sys...
详细信息
The vision system described serves as a sensor for an automated welding process. The chosen strategy of processing for obtaining the needed results in real-time will be explained. To improve the reliability of the system, fuzzy logic components have been applied to produce a decision variable that acts as an internal supervisor, judging the quality and reliability of the system's calculated output data. The vision system is used as one component of a series of sensors attached to an industrial robot. It has proven to be a successful way of controlling a production process by supplying both geometric and process information. It is implemented on a homogeneous transputer network.< >
The authors propose a novel VLSI sorting network implementing Leighton's column sort. The network is mech-based and modular;it consists of comparison-exchange processing elements (PEs), routing paths, and short br...
详细信息
In this paper, we investigate the use of the advanced fiber optic multiple-access networks in future high-performance commercial aircrafts to support real time on-board flight control, sensing, voice communications, a...
详细信息
ISBN:
(纸本)0819413364
In this paper, we investigate the use of the advanced fiber optic multiple-access networks in future high-performance commercial aircrafts to support real time on-board flight control, sensing, voice communications, and TV distributions. Novel tunable optical code division multiple-access (OCDMA) networks using prime-code sequences are designed, which can be quickly programmed to generate the required OCDMA sequences. Using the newly proposed OCDMA technique can efficiently build high-speed sensor data distribution networks which can guarantee parallel data communications and multiple connections to support real-time sensor data processing and distributed control. Due to the channel independence and protocol transparence of wavelength-division multiplexing (WDM), the WDM-based hybrid multiple- access techniques are very attractive to future avionics systems integration, which can achieve high communication efficiency and network utilization.
Mismatches between on-chip high performance CPU and data access times is the basic reason for the increasing gap between peak and sustained performance in distributed memory parallel computers. We propose the concept ...
详细信息
Mismatches between on-chip high performance CPU and data access times is the basic reason for the increasing gap between peak and sustained performance in distributed memory parallel computers. We propose the concept of balanced architectures, based on a network with a dynamic topology and communication patterns determined at compile time. The corresponding processing element is a cacheless CPU, which can achieve a 1 FLOP/clock cycle rate. network and PE features are presented. An example shows that balanced architectures keep efficiency when scaling.
Neural network (or paralleldistributedprocessing) models have been shown to have some potential for solving optimisation problems. Most formulations result in NP-complete problems and solutions rely on energy based ...
详细信息
Neural network (or paralleldistributedprocessing) models have been shown to have some potential for solving optimisation problems. Most formulations result in NP-complete problems and solutions rely on energy based models, so there is no guarantee that the network converges to a global optimal solution. In this paper, we propose a non-energy based neural shortest path networkbased on the principle of dynamic programming and least take all network. No problem of local minima exists and it guarantees to reach the optimal solution. The network can work purely in an asynchronous mode which greatly increases the computation speed.
We propose a general technique to study the perfor mance of random routing in a hypercube system. The proposed method is a based on association schemes in stead of using representation theory.
We propose a general technique to study the perfor mance of random routing in a hypercube system. The proposed method is a based on association schemes in stead of using representation theory.
暂无评论