Dynamic Reconfigurable Computer With A Dynamic Genetic Algorithm

Abstract

The performance of the human brain is incredible and motivated us to create a system which behaves similarly. The flexibility of performance is the main function we emulate. We will create a system model of heterogeneous dynamic multi-core reconfigurable computers with a genetic algorithm. We introduce several concepts to find the next configuration with the genetic algorithm. Before we create our model, we introduce concepts of heterogeneous static multi-core reconfigurable computers to verify the optimization capability of the genetic algorithm. From the simulation results, we conclude that the genetic algorithm can optimize the configuration candidate of static systems. Then, the dynamic reconfigurable systems are generated, simulated, and observed. We conclude that there might exist potential performance improvement of our systems compared to homogeneous multi-core systems. Our evaluation model is fully parameterized, and will be available to the research community. We suggest future applications and improvements for static and dynamic systems.