A new solution based on multi-objective algorithm for multi-application mappings for Many-Core systems

In the current context of intelligent systems and big data, specifically concerning high-performance and efficient data processing for streams applications, single-core systems can no longer meet this demand. Increasing their operating frequencies would result in higher energy consumption for the chip as a whole. Thus, in this scenario, multicore and manycore chips appear to be a viable solution. However, distributing mono8 application or multi-application tasks on these devices is not trivial in practice. It is necessary to consider heat dissipation effects throughout the system while simultaneously ensuring fault tolerance and good load balancing. Although there is not yet a considerable number of works in the literature addressing the allocation of multi-applications in NoCs, there are many works focusing on mapping mono-applications. Therefore, this article proposes a multi-objective mapping model that focuses on performance metrics and heat distribution for multi-applications in manycores with NoCs, aiming to contribute in this direction. The results found are compared with the state-of-the-art algorithm in this scenario, showing promise for the field, with improvements of more than 30% in latency and 38% in fault tolerance when all applications are taken into account.