Cython actually shows the cost of interpretation quite well. Even without added type declarations it speeds up a quicksort of a list of integers by anywhere from 40-50%.
https://github.com/psoberoi/stepanov-conversations-course/tr...
I certainly don't see Cython as anything but a quick way to get efficient C code in to Python module though... you typically get the worst of both worlds with compilation + dynamic typing.
Why are you not using existing, efficient sort implementations?
The github repo is from Alex Stepanovs course at A9, which is largely focused on data structures and algorithms. The point of the benchmarks is to compare 'FORTRAN like' code in many languages and to demonstrate that languages don't really matter if you pick the right DS&As.
That quicksort algorithm is about as efficient as it gets btw.
Well yes, Hack is a "new language" in a similar way TypeScript is, but Hack is built to interoperate with regular PHP code. It's a superset, similar to how Objective-C was/is with C, only even more integrated with it's subset language :)
Regardless, it's damned nice (especially with TypeScript on the front end: static optional typing everywhere!)
Re:Why Python is Slow: Looking Under the Hood
by maxerickson on Friday May 09, 2014 @02:52PM
(#7722641)
There is support in Cython for doing the compilation transparently:
http://docs.cython.org/src/tutorial/cython_tutorial.html#pyx...
I assume by "new PHP implementation", you're referring to Hack[2] which is a different language, although heavily based on PHP. (I believe Hack is a superset?)
[1] http://cython.org/ [2] http://hacklang.org/