This comment [1] in Tensorflow's discussion group gives some valid points explaining some of the crazy numbers reported by CNTK.
[1] https://groups.google.com/a/tensorflow.org/d/msg/discuss/Dhy...
[1] https://groups.google.com/a/tensorflow.org/d/msg/discuss/Dhy...
I think Tensorflow is developed in a very principled manner focusing on ability to add more ops and platforms in future. While this has hurt TF usability in the short term I am bullish on its future.