Here's the code.

void MathFunc()


{


int a;


a = 0;


a = a + 1;


a = a * 10;


a = a / 10;


a = a % 10;


}





void GfxFunc_FillScreen(int color)


{


int x, y;


while (y < ImageHeight(screen))


	{


	x = 0;


	while (x < ImageWidth(screen))


		{


		SetPixel(x, y, color, screen);


		x++;


		}


	y++;


	}


}

Here's the deal. I basically loop these functions 2000000 times, or two million.

I then unwrap the function and insert the actual code into the loop. For example.

void autoexec()


{


int c;


while (c < 2000000)


     {


     MathFunc();


     c++;


     }


}

void autoexec()


{


int c, a;


while (c < 2000000)


     {


     a = 0;


     a = a + 1;


     a = a * 10;


     a = a / 10;


     a = a % 10;


     c++;


     }


}

Finally, I track how long the loop takes for each piece of code, in a function and unwrapped.

MathFunc() called: 4004 centisecs
MathFunc() unwrapped: 3275 centisecs

FillScreen() called: 7364 centisecs
FillScreen() unwrapped: 7747 centisecs

(note I only looped the FillScreen code for 200 times. Otherwise I would have been here all day.)


...I understand the overhead difference with the math function, but why is the FillScreen code faster IN a function call?

It was the one I used for the demonstration, yes.

I was amazed to get the same result - the unwrapped fill screen took longer - not by much but still.
The only thing that springs to mind would be the extra
assignment operation y=0; that needs to be called after the unwrapped version.