How is the OMD so fast

Sums and subtractions. Really.

If you put the lens construction aside, all the work that the camera processor has to do is to read a pixel, read it's neighbour and compute the difference. The bigger the difference the more probability it is in focus. Simplistically you can them rack the lens north and south until you find the biggest value. 
The faster you can rack the lens (and read the sensor) the more measures of those differences you have and the faster you can find out the maximum value and thus ideal focus distance.
That's how Olympus worked CDAF along most of the Pen line until the E-P3 which finally was revised to work a little bit more intelligently and thus radically focus faster.

Note: I said simplistically before because there are other issues that need to be accounted with, like: image noise, camera shake, changing lighting and size of focus area. But in the end the general idea is that simple.

--

Duarte Bruno