Estimation of software structures dimension influence on data processing time increasing

The object of the study is the phenomenon of an extreme increase in the time of program code execution at certain sizes of data processed by it. The problem to be solved was to verify the general nature of the phenomenon for different equipment. The evolution of modern computing technology, its RAM often takes place in an extensive way – by increasing the number of structural elements. Problems can manifest themselves in the fact that periodic processes in the code begin to exhibit a resonance effect, which leads to different indicators of data processing time, the sizes of which are multiples and non-multiples of the block structures. The work is studied the influence of the dimensionality of data blocks on the speed of execution of the cycle that iterates them. The tools of differential regression analysis are used. Experiments were carried out on equipment with different architecture, type and amount of RAM, running different operating systems. In all of them resonant effects were revealed. It led to differences in the average code execution time by 1.6–3.6 times, and the time of memory access operations increased up to 136 times. Special attention was drawn to the fact that the increase in operating time was found for structures whose size is a power of two multiple (2N), specifically for the values 512 and 1024. These dimensions are present in many types of tasks, in particular, cryptographic purposes or stream-based data processing. Following the recommendations given in the paper can help identify time delays in applications, and improve the performance of applications by eliminating them