Question: why use TensorFlow BasicLSTMCell instead of other kind of LSTM cell in TF?

[yehenrytian]

Just curious, why use BasicLSTMCell instead of other "faster" LSTM cell in TF like:

tf.contrib.rnn.LSTMBlockCellor tf.contrib.rnn.LSTMBlockFusedCell` ?

I am not really a machine learning expert, just thought why not using a faster LSTM cell in TF. I am referring to this post online: https://returnn.readthedocs.io/en/latest/tf_lstm_benchmark.html