[zeroibis]

I am wondering if it is faster to use a bit more code and complexity to implement a series of switch statements instead of a single if. Below is a simple example:

new Float:arr[3]={0.0,0.1,0.2}

if(arr[ranposition]<=rand(0,1)
{
run
}

or

new position = ranposition
switch(position)
{
case 0:
{
new chance=ranint(0,100)
switch(case)
{
case 0...10
{
run
}
}
}
ect...
}

Basically you replace all ranges of values with switch statements. It would be a pain in the ass to code but would it actually run faster!?

[napalm00]

Switches are faster than if statements, so yes.

[zeroibis]

Yes under normal conditions however here it is a series of switch statements vs a single if statement. Would even a large number of switches still be faster than a single if!?

This I find to be more of a question of how big of a difference there really is between the two in sm.

[thetwistedpanda]

You should be able to use the profiler to determine which one is faster.

[napalm00]

Switches use a lookup table (or a hash list), which *should* make them faster in any condition. It's like a normal array vs a trie array, if you have just a few items you won't notice any difference, but if you have a lot, then using a switch is the best option.
That said, I believe using if/else statements is better for a cleaner code, but we're talking about performance here.

[zeroibis]

Well if anything I suppose it is time to play around with the profiler a bit then.

[Dr!fter]

I used this plugin to test the difference.

PHP Code:

//Yar
#include <sourcemod>

public OnPluginStart()
{
    new Float:time = GetEngineTime();
    PrintToServer("Start time for if loop is %f", time)
    for(new i = 0; i <= 100000; i++)
    {
        new random = GetRandomInt(0, 100);
        if(random <= 5)
        {
            PrintToChatAll("Yay!")
        }
    }
    PrintToServer("Total time for if was %f", GetEngineTime()-time)
    time = GetEngineTime();
    PrintToServer("Start time for switch loop is %f", time)
    for(new i = 0; i <= 100000; i++)
    {
        new random = GetRandomInt(0,100);
        switch(random)
        {
            /*case 0...5: //This dosnt work!
            {
                PrintToChatAll("Yay!")
            }*/
            case 0, 1, 2, 3, 4, 5:
            {
                PrintToChatAll("Yay!")
            }
        }
    }
    PrintToServer("Total time for switch was %f", GetEngineTime()-time)
} 


Output

Start time for if loop is 1.846865
Total time for if was 0.009515
Start time for switch loop is 1.856752
Total time for switch was 0.009230

As you can see no huge difference or even noticeable difference. This probably isnt the best method to test but. To be fair their is very little need to worry about if or switch.

[RedSword]

I do believe that when comparing using if or switch would in assembly would both do something like :

- put a var in a register
- put a constant in another
- compare the two

but whenever comparing more than once, the if would do :

- put a var in a register
- put a constant in another
- compare
- REPEAT

while switch would do :

-put a var in a register
-put a constant in another
-compare
-REPEAT from step 2 (keeps the same var)

However compilers does optimize it (I've no idea about pawn; but since it seems less important (then let's say Visual Studio compiler), I'd guess it is not optimized; I'd have to look at the compiler's code however ). Also, I've no idea about loops; as I'm talking here about consecutive comparison (if, elseif, etc.)