Set Classification in Set@l Language for Architecture-Independent Programming of High-Performance Computer Systems

Abstract

Traditional programming languages for parallel computer systems do not separate the description of an algorithm from the details of its hardware implementation efficiently. As a result, the porting of the same algorithm between different computational architectures and configurations requires a significant code modification. To reduce the time and complexity of porting, we proposed an architecture-independent Set@l programming language based on the aspect-oriented programming paradigm and set-theoretical code view. In contrast to conventional tools for parallel programming, Set@l operates by sets, subsets, attributes and relations between them. Various aspects of realization transform an architecture-independent source code according to the certain architecture and configuration of a computer system. Set@l provides the porting of parallel applications without the changing of a source code, the modification of an information graph or a computing structure with regard to computer system’s features or user’s preferences, and the indefinite description of calculations. This paper treats the essential issue of set classification in the Set@l programming language. We propose three basic criteria of set typing: by the parallelism of collection’s elements, by the definiteness, and by the processing method. We yield examples of programs in Set@l that demonstrate the techniques and specificities of various set types’ usage.