Towards to Load Balancing for Cooperative Multithreading Implementation of Process-Oriented Programs

In the modern world manufactures and industrial processes are controlled by microcomputers. The paradigm of process-oriented programming was proposed by IA&E SB RAS to help with this process of automation. This paradigm shows promising results due to innate accounting for every domain-specific issue. A typical implementation of a process-oriented program is based on cooperative multithreading, which allows the use of a stand-alone round-robin solution. However, along with the growing complexity of control programs, a number of issues arise: ensuring the reliability, a-priori estimating correspondence of calculation complexity of the algorithm to the platform, and introducing enhanced load balancing means. There are some existing algorithms to solve the load balancing problems but they require adaptation for the paradigm. And this work has to develop means and criteria to compare those algorithms. This comparison is important before adaptation because there is no use in spending time trying to adapt outright bad or suboptimal algorithms. As the result, this work presents the scheme of an experimental evaluation, which allows comparing different load balancing algorithms without directly comparing some implementations of them. In addition, this work presents the working adaptation of EDF for process-oriented languages to prove that the scheme is working by evaluating the algorithm through the scheme.