CODEWITHSUNDEEP
c++performanceif-statement
Dr.Kameleon
Dr.Kameleon

Reputation: 22820

Nested Ifs VS 2 separate IFs - performance-wise?

Version A :

if ((A)&&(B)) doSth;
if (B) doSthElse;

Version B :

if (B)
{
    if (A) doSth;
    doSthElse;
}

Which of these two is preferable, performance-wise?

Notes :

Upvotes: 0

Views: 309

Answers (3)

jstine
jstine

Reputation: 3350

If your A and B values are booleans then I'll suggest a third version:

if (A & B) doSth;
if (B) doSthElse;

This version uses bitwise AND to create a single comparison out of multiple booleans. This can also be applied to the ternary operator solution posted in another answer.

This can be beneficial because it removes a branch for every && substitution. Most of the time, AND'ing several booleans together is cheaper than doing a branch on the value of each one. This applies to any cpu architecture that has relatively expensive branching, which is anything with either out-of-order execution, or a long instruction execution pipeline (which accounts for almost everything).

Important note: On x86, the conditional executors and branch predictors are good enough that if boolean A is well-predicted (eg, rarely changes and prediction rate is above 99.6%) then its actually more efficient to use && form and shortcut past the rest of the conditional. ARM and PowerPC architectures however almost always benefit from fewer branches.

Upvotes: 0

user1252446
user1252446

Reputation:

Assume A and B are both simple boolean values, We need take into account the likelihood of what the condition resolves into:

  1. short-cut evaluation: if (A) is likely to resolves to false than (B), write (A && B), otherwise (B && A).

  2. branch predictability: use conditions that more predictable enclose big blocks. for example, if B is predictable, then 2nd form is preferred.

  3. try to translate non-predictable conditional assignment into (? :), for example, prefer

    x = c ? a : b; // data dependency

    to 

    if (c) x = a; // control flow dependency
else   x = b;

if c is not predictable. In this case, you want to substitute a control flow dependency by a data dependency, which can be compiled into a conditional move. It is net gain when the control dependency is not predictable.

Upvotes: 2

Luchian Grigore
Luchian Grigore

Reputation: 258638

Depends on what A and B are. If B is a complicated function, the second one only evaluates it once, whereas the first one evaluates it twice (if A holds, of course).

For trivial cases (i.e. both bools) it won't matter.

And, of course, you could profile, I doubt this will be a bottleneck though.

Upvotes: 3

Related Questions