This email address is being protected from spambots. You need JavaScript enabled to view it.
 
+7 (4912) 72-03-73
 
Интернет-портал РГРТУ: https://rsreu.ru

UDC 004.432.4

ARCHITECTURE-INDEPENDENT PROGRAM OF FAST FOURIER TRANSFORM IN SET@L PROGRAMMING LANGUAGE

I. I. Levin, Dr. Sc. (Tech.), full professor, professor, Head of the IMS Department, ICTIS, SFedU, Taganrog, Russia; This email address is being protected from spambots. You need JavaScript enabled to view it.
A. I. Dordopulo, Ph.D. (Tech.), Head of MAS Division, SRC SC & NC, Taganrog, Russia; This email address is being protected from spambots. You need JavaScript enabled to view it.
I. V. Pisarenko, junior researcher, SRC SC & NC, Taganrog, Russia; This email address is being protected from spambots. You need JavaScript enabled to view it.
A. K. Melnikov, Ph.D. (Tech.), senior researcher, «Inform Invest Group» C. C., Moscow, Russia; This email address is being protected from spambots. You need JavaScript enabled to view it.

Conventional techniques of parallel programming do not divide the descriptions of an algorithm and the features of its implementation on a high-performance computer system. As a result, the porting of same parallel application between different computational architectures requires a significant code revision. To solve the problem, we proposed an architecture-independent Set@l programming language based on set theory and aspect-oriented paradigm. Using multiple combinations of collections, attributes, set partitions and relations, one can specify various parallelization and modification methods for an algorithm in a unified aspectoriented Set@l program. In this paper, the aforementioned feature of the Set@l programming language is illustrated by the example of the rotation coefficients’ calculation for the fast Fourier transform. We show that the procedural program describes only two boundary cases with minimum and maximum memory costs. In contrast to the procedural approach, the fast Fourier transform code in Set@l defines additional implementations of the algorithm with intermediate memory sizes and provides the continuity of calculation model. The proposed method of parallel programming in the Set@l language allows us to synthesize the set of solutions which are invisible from the standpoint of the procedural programming paradigm and switch between them according to architecture and configuration parameters of the computer system used for calculations.

Key words: architecture-independent programming, Set@l programming language, aspect-oriented methodology, fast Fourier transform.

 Download