摘要:Distributed systems and parallel computing have been use for many different purposes. This fact is related to its high performance and flexible architecture. In this context, areas such as engineering, computing and medicine make use this feature to get results in less time. For instance, in medicine there are many research that associating its computing power for processing genome, complex exams and others. In this paper, we present a new concept to evaluating exams that uses artificial neural networks on graphics processing unit (GPU). We intend investigate which conditions there are performance gains and what characteristics is more relevant for its use.