Key words
Fortran, compilers, parallel programming, free software, proprietary software, co-arrays, Autocrossparallelization.
articles
Authors:
Alekseev Е., Boltacheva N., Demin P. Published in
№ 1(73)
27 february 2018 year
Rubric: Software development technologies
Rubric: Software development technologies
New technologies for developing high-performance and parallel applications on modern Fortran
The article is devoted to the use of new parallelism technologies, which appeared in modern Fortran
standards (Fortran-2003, Fortran-2008). The authors describe conveyor operations and subroutines.
Examples of speeding up programs on Fortran using conveyor operations are given. New technologies
of parallelism are considered for the first time: coarrays and auto-parallelization. On the example
of iterative algorithms for solving systems of linear algebraic equations, the analysis of the speed of
parallel programs is performed using all concurrency technologies supported by Fortran compilers.
A comparative analysis of the performance of free and proprietary compilers is carried out. The recommendations
on the use of various Fortran compilers for solving computational problems are presented.
The features of coding on modern Fortran are described with the use of various compilers.
Provided reasoned recommendations on the use of Fortran as a system of mathematical modeling.
Literature sources are presented, in which the graphical conclusion of research results in cross-platform
programs on Fortran is described.
The choice of Fortran as the language of teaching programming for students of mathematical, physical,
biological, chemical and technical specialties is justified.
