Дослідження проблем швидкодії програмних додатків

Abstract

The article presents a study of methods for optimizing the performance of software applications aimed at identifying the most effective combination of internal and external factors that maximize the objective function. A generalized mathematical model is described, which includes the main factors affecting performance, such as computation time, the number of input/output operations, the number of computational operations, algorithm complexity, the volume of data processed, the use of parallelism, the architecture of hardware and software platforms, and code efficiency. The importance of using specialized libraries and tools to accelerate computational processes, which is critically important for achieving high performance in modern software systems, is emphasized. The developed approaches were implemented in software, allowing for the practical evaluation of the proposed methods. Software modules were created to analyze the impact of various factors on performance, considering the specifics of particular tasks and execution environments. The test results demonstrated significant potential for performance improvement through optimization at both the code level and the hardware architecture level. Particular attention is given to the study of memory management, addressing potential challenges that negatively impact performance. The necessity of using a caching system and avoiding duplication of immutable information is highlighted. The identified scenarios are independent of any specific implementation and can therefore be integrated into the developing recommendation system. The research has practical significance, offering comprehensive solutions for optimizing the performance of software systems that can be applied in industrial high-load environments. Further research will focus on expanding the functionality of the recommendation system, integrating more complex optimization models, and conducting large-scale computational experiments to validate the results under real-world conditions.