I maintain a program written in Python that is faster than the program written in C that it replaces. The C version can do a lot more operations, but it amounts to enumerating 2^N alternatives when you could enumerate N alternatives instead.
Certainly my version would be even faster if I implemented it in C, but the gains of going from exponential to linear completely dominate the language difference.
So you're saying two different programs implementing two different algorithms perform differently and that lets you draw a conclusion about how the underlying language/compliers/interpreters behave?
I must have been unclear. C is faster. The Python program would be faster if I reimplemented it in C. However, Python makes it so much easier to transform the problem into a linear form that it’s a bigger win to use Python than to continue maintaining the C version.
And that’s my point: raw execution speed is only helpful when you’re executing the right thing. Don’t discount how much easier it can be to implement the right thing in Python.
Certainly my version would be even faster if I implemented it in C, but the gains of going from exponential to linear completely dominate the language difference.