Hacker News new | ask | show | jobs
by sgillen 805 days ago
I like to joke that to make python fast you need to avoid using python. A ton of python stuff is C/C++ or Fortran wrapped in python, numpy, pandas, pytorch etc is what I have experience with but this applies in all domains. A huge part of learning to write fast python is to first learn these libraries, and second learn about how data is shepherded back and forth. Your experience with C/C++ gives you a good background to understand this!

That being said it's totally possible to write pure python that is much faster than other ways of doing pure python, learn how its all implemented if you are interested, and profile like crazy! Good luck!