CODEWITHSUNDEEP
pythonoptimization
Saqib Ali
Saqib Ali

Reputation: 12585

How should I optimize these if/elseif branches for speed?

Suppose I have the following python code:

a = b = 0

if f(a) and (g(b) == 1):
    # Do Something #1
    pass

elif f(a) and (g(b) == 2):
    # Do Something #1
    pass

You can assume that the functions f() and g() take up some non-negligible amount of CPU resource.

I want to optimize this code for computational efficiency. Should I change it to the following?

a = b = 0

if f(a):
    x = g(b)
    if x == 1:
        # Do Something #1
        pass
    elif x == 2:
        # Do Something #2
        pass

Is there any way to automate this optimization? I hate doing it by hand throughout my entire codebase.

EDIT: The functions f() and g() are completely deterministic. IE: they always return the same results if given the same input parameters. And they have no side effects outside their own stack context.

Upvotes: 1

Views: 76

Answers (1)

Antti Haapala -- Слава Україні
Antti Haapala -- Слава Україні

Reputation: 133929

Using a single if for the f(a) is definitely more efficient. For optimal efficiency you should store the result of expensive common sub-expressions in a variable:

if f(a):
    gb = g(b)
    if gb == 1:
        ...
    if gb == 2:
        ...

This will execute these functions as seldom as possible

If g(b) can also return a value that does not match any if, then you should profile - it could be beneficial to calculate g(b) first and see if it returned any valid value, and only then calculate f(a) (if these are not dependent of course):

gb = g(b)
if gb in (1, 2) and f(a):
    if gb == 1:
        ...
    if gb == 2:

There is no way that any standard Python compiler could statically optimize these unlike in C/C++ where the function contents can be inlined.

Upvotes: 4

Related Questions