pow()

Cast the parameters to double / float?

What do you mean?
Thanks.

Bah forgot to log in, anyway that should work ^.

If you use only integers and don't need very big numbers, you can write power much faster than the pow function.
For example try this:

long long int_pow(int x, unsigned int exponent){    long long temp = x;    long long res = exponent & 1 ? x : 1;    exponent >>= 1;    while (exponent != 0)    {        temp *= temp;        if (exponent & 1)        {            res *= temp;        }        exponent >>= 1;    }    return res;}

This is 64-bit version which is relatively slow. If you know that you won't be needing numbers bigger than 2³¹, you should change return type and types of temporaries in the function to the long. It should speed it up on any 32 bit machine since it will not call 64-bit multiplication which is emulated on 32 bit machines.

Quote:

Original post by b2b3 If you use only integers and don't need very big numbers, you can write power much faster than the pow function. For example try this: long long int_pow(int x, unsigned int exponent){ long long temp = x; long long res = exponent & 1 ? x : 1; exponent >>= 1; while (exponent != 0) { temp *= temp; if (exponent & 1) { res *= temp; } exponent >>= 1; } return res;} This is 64-bit version which is relatively slow. If you know that you won't be needing numbers bigger than 231, you should change return type and types of temporaries in the function to the long. It should speed it up on any 32 bit machine since it will not call 64-bit multiplication which is emulated on 32 bit machines.

But how useful is it to someone who's simply asking how to use pow()?
I must have missed the part about "help, I've profiled my program, and it turns out the pow() function is my main bottleneck. Help me optimize it" [wink]

Yes I think for just a few calls for pow it's a bit too much to introduce a new function to do it.

Quote:

Original post by Spoonbender But how useful is it to someone who's simply asking how to use pow()? I must have missed the part about "help, I've profiled my program, and it turns out the pow() function is my main bottleneck. Help me optimize it" [wink]

I didn't mean to push him into optimizing his program (alhough from my post it may sound like that [smile]). That would be really unnecessary in most cases.
For me it just looks little weird to use floating-point pow when working with integers [wink]. You may get incorrect results in some cases. With doubles it is not that big problem, since you have 52 bits for mantissa, but for floats it may be problem if he needs bigger numbers.