Is there a speed difference when building TensorFlow in C++ and Python?

I’m currently using a model in python that takes about 7 to 10 seconds to build. I’m researching to shorten the build time, and I saw an article saying that building in C++ can shorten the time. Is the method meaningful?

Hi @Script_K, C++ is a low level language that is focused on working directly with the hardware, and the compiled nature of C++ can sometimes be more handy than the interpreted nature of python. But a lot depends on what you are trying to achieve/build, if more insights are provided on this, we can get a better understanding. Thank You.

1 Like