Maybe everyone knows this already, but I just thought of it for my first time...

Say you've got a 64bit processor, but you're working with two 32bit numbers... if they're stored in a single 64bit variable, and you're sure they won't overflow, you can do two additions simultaneously! Or 4 16bit additions, or 8 8bit!

So, what could this be used for? How often would that really be useful?

I dunno... a screen is significantly smaller than 65536x65536 pixels... 

....

Presently I'm using an 8bitter, I have a function that I would like to return two TRUE/FALSE values. I want to keep running sums of those two values from each time it's called.

//Returns:
//0x01 if button A pressed
//0x10 if button B pressed
//0x11 if both are pressed
uint8_t getButtons(void);

main()
{
   uint8_t countsCombined=0;

   for(i=0; i<15; i++)
      countsCombined+=getButtons();

   printf("A presses = %d\n"
          "B presses = %d\n",
          countsCombined&0x0f,
          countsCombined>>4);
}

But, holy moly, this seems a little cheesy on an 8bitter, but just think what could be done on a 16bitter, or 64bitter!

Maybe you're designing PONG on a 16bit computer in a 256x256 window, the ball moves two pixels up, one left:

//Upper byte is X, lower byte is Y

uint16_t ballPosition = 0x0000;

uint16_t ballStep = 0x0102;

while(wallNotHit)
{
   ballPosition += ballStep;
}