The method of Wolf, basically provides the calculation of the maximum exponent of Lyapunov by the evolution of the distance between two points closes from two nearby paths. This enables the quantification of the chaos through the sign of the exponent, is also an important tool as an indicator of the predictability of a time series. The algorithm presented by Alan Wolf is not very efficient because of a search performed by brute force, therefore, the execution times turn out to be very high, here is where lies the importance of the parallelization of this method. In the present work, two parallel implementations of the method of Wolf shared memory platforms, described the first using the API second based for GPU, and OpenMP using NVIDIA CUDA platform. The results obtained in terms of execution time decreased approximately above 90%, for all patients in study for parallel OpenMP program. In the case of the parallel program on GPU, achieved a decrease of runtimes between approximately 15% and 65% (depending on the case study), in comparison to the parallel OpenMP version; and even more, a considerable decrease of time in comparison to the sequential version.