In this study, the author researches the numerical results of the comparison of synchronization speed of regular neural networks with unidirectional and bidirectional coupling. Each neuron is linearly coupled with the others and is represented by a reaction-diffusion system of FitzHugh-Nagumo type. The result shows that the necessary coupling strength for the synchronization in two cases decreases when the coupling number of neurons increases. In other words, the bigger the coupling number of neurons in the regular networks is, the easier the synchronization occurs. Moreover, synchronizing the regular networks with bidirectional coupling is easier than the one with unidirectional coupling over the same given number of neurons.