Designing and Implementing Library Parallel Processing of Large Data

The ever-increasing data volume for analysis requires to seek new approaches for its processing and to modernize already existing. Now the parallel computation and computation optimization technologies at the processor level are considered the most effective. The article describes a developed system library combining both approaches. There is some information on its comparative testing with programs using other tools clearly illustrating its effectiveness. Also, the library testing process identified an algorithmic problem causing the program to slow down in case of a large heterogeneity of the original data. Doubling the number of parallel threads helped the authors to decide the arisen problem.