On Sunday, 97/01/19, Tom Almy wrote to Thomas Seeling about "Floating
point feature wi" as follows:

TA> I discovered that if you use the inline function optimization,
TA> trig functions break with large argument values. Try sin(1e30) --
TA> the result isn't between -1 and 1!

Hi Tom,

There was a discussion in the Usenet comp.lang.rexx Newsgroup a couple
of weeks ago about errors evaluating transcendental functions. This
specific problem was included.

The issue arises over the uncertainty in the argument exceeding the
full cycle from 0 to 2*pi, which means that any value in the range -1
to 1 could be returned as being valid. I suspect Intel has chosen to
ignore this in the hardware implementation of trig. functions, and the
FPU simply returns whatever turns up after its [probably] iterative
algorithm has run its course.

Regards

Dave


 * KWQ/2 1.2i * Windows: Revenge of the Nerds, Part 3.1

--- Maximus/2 3.01