Hi!
It seems you need sometimes precise what kind of precision you
need ;-)
Try this site:
http://www.oonumerics.org/oon/ . But maybe someone
could help you a little more, if you wrote what are you really doing?
You need floating point numerics or fixed point? Maybe your algorithm
needs great precision, but input and output data have lower ones? So
maybe there is a better, less-precise-greedy algorithm?
These are sugestion for going around the problem. But - try the site I
linked above.
And give us your results ;-)
Best regards,
hbrt