This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: new alpha division routines
- From: Petter Reinholdtsen <pere at hungry dot com>
- To: libc-alpha at gcc dot gnu dot org
- Date: Sat, 27 Mar 2004 09:16:46 +0100
- Subject: Re: new alpha division routines
- Bcc: Petter Reinholdtsen <petter dot reinholdtsen at usit dot uio dot no>
- References: <20040327004755.GA10079@redhat.com>
[Richard Henderson]
> The new routines use the FPU to either get the complete result
> (32-bit routines; 64-bit routines with small arguments), or an
> estimate at the result from which we can refine the correct
> integral result.
I believe I have read somewhere that the alpha floating point
operations were not IEEE compatible, and that this was the reason
Tru64 Unix supplied the libieee library. Am I wrong or did this
change?
I guess my real question is if you are sure the results from the
calculations are the same with your change.