David Noon wrote in a message to Mike Bilow:

 DN> If Intel is using a look-up table for its trig functions, it 
 DN> has some ropey values in the sine/cosine table if it can yield 
 DN> a value > +1 or < -1. It looks a lot more like a Taylor 
 DN> polynomial of fixed order has been chosen and whatever junk 
 DN> comes out the other end is assumed to be right.

If you set up for IEEE compliance, you should get NAN.
 
-- Mike


---