The strangest one I've seen is the fast sqrt() function in SGI OpenGL source code: there's a magic step where a right bit shift is applied to a 32bit IEEE float.
(oops.. just clicked your link, looks like we're related)
The strangest one I've seen is the fast sqrt() function in SGI OpenGL source code: there's a magic step where a right bit shift is applied to a 32bit IEEE float.
(oops.. just clicked your link, looks like we're related)