Performance optimisation in software for data acquisition systems

[1] High energy physic experiments conducted in modern accelerators or fusion reactors require to store and process enormous amount of information. For such systems it is extremely important to correctly design and implement software to maximally utilize computation power offered by available hardware. Overhead limitations introduced by factors like task scheduling, thread synchronization, dynamic memory allocation or incorrect hardware use must be limited to minimum. The paper presents several potential source of efficiency limitation that should be taken into consideration during development process of software for data acquisition systems in order to maximally utilize throughput of the offered hardware. Additionally, methods to minimize impacts of such problems and experimental results of proposed improvements are presented.