Parallel computers classification

Unfortunately, Flynn s classification, although commonly used, is quite restrictive when discussing parallel-architecture computers. There have been several attempts to formulate more detailed classification schemes for the great variety of parallel computers now available. None of these efforts have been entirely successful, and none appear to be in general use. A discussion of representative machines from some of the more common classes follows. 

During the early part of this decade, there was a vivid debate between proponents of MIMD (multiple instruction stream, multiple data stream) and SIMD (single instruction stream, multiple data stream) type parallel computers. This taxonomy, which stem from Flynn s classification [25,26] of possible computer architectures, can now be said to be mostly of historical and academic interest. 

As an example consider the 1990 U.S. Census classification task. Free text samples from 22 million citizens were classified by assigning industry and occupation codes, based on job descriptions given in questionnaires [14]. An expert system called AIOCS was used. The rules were laboriously derived from a training data base consisting of 132 247 cases classified by human experts. The time required for the development of this system was equivalent to 192 month/person (16 years/person). Still, accuracy is rather low (Table 1). An alternative memory based system called PACE was developed in just 4 months/person and efficiently implemented on CM-2 parallel computer. 

Parallel computers may also be distinguished based on whether they are allocated for capacity or capability computing, and this distinction is related to the commodity versus custom classification. Capacity computers are designed to obtain good throughput and high cost-effectiveness for running many small- to modest-sized jobs. Capability computers are reserved for... 

This chapter provides an overview of parallel compufer architectures, including the traditional Flynn classification scheme and a discussion of computation nodes and the networks connecting them. We also present an overall system view of a parallel computer, describing the hierarchical nature of parallel archifecfure, machine reliability, and the distinction between commodity and custom computers. 

The development of vector and parallel computers has greatly influenced methods for solving linear systems, for such computers greatly speed up many matrix and vector computations. For instance, the addition of two n-dimensional vectors or of two nxn matrices or multiplication of such a vector or of such a matrix by a constant requires n or arithmetic operations, but all of them can be performed in one parallel step if n or processors are available. Such additional power dramatically increased the previous ability to solve large linear systems in a reasonable amount of time. This development also required revision of the previous classification of known algorithms in order to choose algorithms most suitable for new computers. For instance, Jordan s version of Gaussian elimina-... 

We describe in this section various terms and concepts pertaining to parallel computing, including the classification of parallel computers, issues relevant to development of parallel algorithms, and measures for assessing the performance of parallel programs. 

A variety of parallel computer architectures exist, differing in their performance characteristics and imposing different requirements for efficient execution of parallel programs. It is therefore useful to categorize parallel computers based on certain important architectural features, and we discuss in the following some of the more commonly used classification schemes. 

Parallel computers may be classified based on the flow of instructions and data as proposed by Flynn. Flynn s classification scheme divides computers into four classes ... 

The parallel stmcture in the NSC allows for rapid computations of output signals. Although training takes some time, it can be done once on a representative set of data. When training has been completed the classification process is fast and easy to implement in a realtime application. 

A currently popular approach to classification and pattern-recognition problems involves neural networks. Neural networks are mainly used as (non-)linear approximations to multivariable functions or as classifiers (Ripley, 1993). Principally, the technique is intended to mimic the computational properties of the brain, which is highly parallel in its operation. Artificial neural networks (Figure 3.10) consist of units with some of the properties of real neurons. 

Cappadonna PA, Costa A, Fichera S (2013) Makespan minimization of unrelated parallel machines with limited human resources. Procedia CIRP 12 450 55 Edis EB, Oguz C, Ozkarahan I (2013) Parallel machine scheduling with additional resources notation, classification, models and solution methods. Eur J Oper Res 230(3) 449—463 Kaplan S, Rabadi G (2012) Exact and heuristic algorithms for the aerial refueling parallel machine scheduling problem with due date-to-deadline window and ready times. Comput Ind Eng 62 (l) 276-285... 

E. M. Rasmussen, G. M. Downs, and P. Willett,/. Comput, Chem., 9, 378 (1988). Automatic Classification of Chemical Structure Databases Using a Highly Parallel Array Processor. 

---