Murom, Vladimir, Russian Federation
In the paper there is considered an algorithm developed on the basis of modified PRAM-model for efficiency increase of parallel computations on specialized computer modules. By means of the efficiency assessment method there were carried out comparative experimental inves-tigations of the algorithm developed. The assessment results of the algorithm for the parallel computation efficiency increase on special computer modules show efficiency increase not less than 2-4 times depending on the number of flows under investigation.
GPGPU, PRAM-model, parallel computations, heterogeneous computer systems, graphics processors
1. Voevodin, V.V. Parallel computations: new concepts in science and education / V.V. Voevodin, Vl.V. Voevodin // Modern Problems in Calculus Mathematics and Mathematic Modeling in 2 Vol. / Vol.1. Calculus Mathematics. - Science, 2005. - pp. 2-15.
2. Graham, R.L. Bounds on Multiprocessing Timing Anomalies / R.L. Graham // SIAM Journal on Applied Mathematics. - 1969. - Vol. 17, No. - № 2. - Р. 416-429.
3. Kolpakov, А.А. Aspects of assessment of computation efficiency increase at paralleling processors of computer systems / А.А. Kolpakov, Yu.А. Kropo-tov // Methods and Devices for Information Transfer and Processing. - 2011. - №1(13). - pp.124-127.
4. Kolpakov, А.А. Theoretical assessment of computer system efficiency increase using some computer devices / А.А. Kolpakov // In the World of Scientific Discoveries. - 2012. - №1. - pp. 206-209.
5. Kolpakov, А.А. Genetic algorithms optimization using computations on graphic processors by example of problem of null bit vectors / А.А. Kolpakov // Information Systems and Technologies. - 2013. - №2(76). - pp. 22-28.
