I don't know how much this matters, but an issue similar to this had been raised here: https://github.com/tensorflow/tensorflow/issues/120 https://github.com/te

I don't know how much this matters, but an issue similar to this had been raised here: https://github.com/tensorflow/tensorflow/issues/120 https://github.com/tensorflow/tensorflow/issues/120 The responses seem to show that the way you implement things can make a big difference in runtime. Perhaps the scripts used for benchmarking can be further optimized? That said, the lack of in-place operations might be surprising (although it has been said that they are coming)