This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
RE: Sine and Cosine Accuracy
- From: "Menezes, Evandro" <evandro dot menezes at AMD dot com>
- To: "Scott Robert Ladd" <scott dot ladd at coyotegulch dot com>
- Cc: gcc at gcc dot gnu dot org
- Date: Fri, 27 May 2005 16:44:35 -0500
- Subject: RE: Sine and Cosine Accuracy
Scott,
> I still maintain that hardware fsin and fcos are valid and
> valuable for certain classes of applications,
I agree. I've just been trying to demonstrate that your test doesn't check sin and cos accuracies, but that sin^2 + cos^2 == 1. If I had a sin that always returned 1.0 and a cos that always returned 0.0 they would pass your test. :-)
> and that we
> need better options and documentation -- both of which I'm
> more than happy to work on. I look forward to your future comments.
By all means, there are many holes in GCC documentation and you've probably tripped at one.
Yet, I think that enabling x87 transcendentals on x86 only with -funsafe-math-optimizations makes sense, because they're anything but safe: "(a) assume that arguments and results are valid and (b) may violate IEEE or ANSI standards."
And it doesn't make sense to enable them on x86_64 because they're not more optimal than the SSE routines.
Regards,
--
_______________________________________________________
Evandro Menezes AMD Austin, TX