This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Serious bug
- To: "Alexandre Oliva" <oliva at dcc dot unicamp dot br>, "kabatek at chemie dot uni-halle dot de" <kabatek at chemie dot uni-halle dot de>
- Subject: Re: Serious bug
- From: "Geert Bosch" <bosch at gnat dot com>
- Date: Fri, 25 Sep 1998 11:09:37 +0200
- Cc: "egcs at cygnus dot com" <egcs at cygnus dot com>
- Reply-To: "Geert Bosch" <bosch at gnat dot com>
On 24 Sep 1998 11:14:13 +-300, Alexandre Oliva wrote:
Complain to Intel. They have put more precision in FP registers than
in doubles stored in memory, that's why you see the difference.
You can set the FPU to any precision you want by changing its rounding mode.
This FPU precisely follows IEEE 756 standards in this respect and provides,
single, double and double extended precisions.
It is the failure of compiler writers/OS vendors if you have no
portable way of selecting the precision you want. For your application
it might be best to explicitly set the rounding mode of the FPU to
round to double, using an assembly language insertion.
Regards,
Geert