This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: [ballen@uwm.edu] libc/2269: triginometric argument reductionerror in libm
- To: Ulrich Drepper <drepper at cygnus dot com>
- Subject: Re: [ballen@uwm.edu] libc/2269: triginometric argument reductionerror in libm
- From: Bruce Allen <ballen at gravity dot phys dot uwm dot edu>
- Date: Sun, 20 May 2001 15:02:59 -0500 (CDT)
- cc: Geoff Keating <geoffk at redhat dot com>, aj at suse dot de, libc-alpha at sources dot redhat dot com, ballen at uwm dot edu, moshier at mediaone dot net
On 20 May 2001, Ulrich Drepper wrote:
> Geoff Keating <geoffk@geoffk.org> writes:
>
> > Someone needs to fix libc/sysdeps/i386/fpu/s_sin.S so that it does
> > proper range reduction, I think. It works fine on non-x86 chips.
>
> It's not worth doing this. What is the result of such a large
> argument good for? It has absolutely no meaning.
Please go back and read my original posting.
(1) The sin and other trig functions have exact values for all normalized
IEEE floats. The IEEE standard requires that the functions return the
closest IEEE value to this correct value. The current library is
returning values that are grossly wrong.
(2) There are certainly scientific codes that depend upon getting this
right. It's why (as you can see from what I sent out before) every major
vendor gets it right and has been getting it right for almost twenty
years.
(3) If you suffer from the common misconception that IEEE floats are
somehow "approximate" please go back and read the standard again.
Bruce Allen