[wladimir]

Interesting story, with a very good speedup. Though I personally wouldn't be this patient, and would write a simple function like this in cython or even the C API directly (especially as the Python code drifts further from idiomatic with each step...).

[pwang]

Writing it in Cython would have been my solution. No futzing around with 8 or 9 iterations. It would just be fast, and the code would still look clean (unlike the obfuscated garbage they ended up with in the blog post).