In this paper we explore the possibility of using GP GPU technology (General Purpose Graphical Processing Unit) in the context of integrative biology. For more than a decade, 3D cellular automata represent a promising approach to handling multi-scale modeling of organs. However, the computing time of such huge automata has limited the experiments. Current GP GPUs now allow the execution of hundreds of threads with a regular PC hosting a device card. This capability can be exploited in the case of cellular automata where each cell has to compute the same algorithm. We have implemented two algorithms to compare different memory usage. The performances show very significant speedup even when compared to the latest CPU processors. The interconnection of GP GPU boards and servers will be considered to build a local grid of hybrid machines.