Improving performance for emergent environments parameter tuning and simulation in games using GPU

Computer games that handle realistic environments are becoming more popular in the game market. Games that make use of natural environments such as the spreading of fire or the flow of water need to be very carefully designed. In order to produce a desired effect of fire or water, a designer needs to try and test map properties several times. There has been an effort to use genetic algorithm to find values of each map property that allow a required scenario to take place in cellular automaton-based emergent environment maps. However, the calculation was extremely slow, especially in large maps. This paper presents an application of graphical processing units (GPUs) to perform the simulations in emergent environment maps, and to perform the genetic algorithm. Test results show that up to overall 11.79x speedup can be obtained from this change.