h2o much faster than neuralnet (in R)

I’m a novice to machine learning. I’ve been trying out different neural network implementations in R, including the neuralnet package and the deeplearning function of the h2o package. For neuralnet, the default setting is one hidden layer with one hidden neuron. With this setting, the model takes several minutes to fit to my data. In the h2o package, the default is two layers with 200 neurons each, and the model takes only a few seconds. How is this possible? Are these two entirely different types of neural networks, or is the h2o implementation just much more efficient?