Question

please help me to identify which of these following is more optimized code?

for(int i=0;i<count;i++)
{
    switch(way)
    {
        case 1:
            doWork1(i);
            break;
        case 2:
            doWork2(i);
            break;
        case 3:
            doWork3(i);
            break;
    }
}

OR

switch(way)
{
    case 1:
        for(int i=0;i<count;i++)
        {
            doWork1(i);
        }
        break;
    case 2:
        for(int i=0;i<count;i++)
        {
            doWork2(i);
        }
        break;
    case 3:
        for(int i=0;i<count;i++)
        {
            doWork3(i);
        }
        break;
}

In the first case, there happens to be an overhead of always checking the switch case condition in every iteration. In second case, the overhead is not there. I feel the second case is much better. If anyone has any other workaround, please help me out in suggesting it.

Answer 1

A switch on low, contiguous values is insanely fast – this type of jump has highly optimised handling. Frankly, what you ask will make no difference whatsoever in the vast majority of cases – anything in doWork2(i); is going to swamp this; heck, the virtual-call itself might swamp it.

If it really, really, really matters (and I struggle to think of a real scenario here), then: measure it. In any scenario where it is noticeable, then only way to measure it will be with your actual, exact code – you can’t generalise pico-optimisations.

So:

1. it doesn’t matter
2. measure
3. it doesn’t matter