Batching and preparing our datasets (to make them run fast)

[kumar8074]

In lecture number 211, after applying a series of methods like map, batch and prefetch at time stamp 12:44 our data is transformed into PrefetchDataSet. But now with latest version of TensorFlow and colab the default datatype without even applying the above transformations is PrefetchDataSet. So do we still need to apply the above transformations for faster computation ? does applying these transformations will still result in faster computation ?