Floating point trouble with x86's extended precision
Richard Henderson
rth@redhat.com
Thu Aug 21 13:16:00 GMT 2003
On Wed, Aug 20, 2003 at 11:27:02PM +0200, Gabriel Dos Reis wrote:
> Somes compilers (e.g. VC++) seem to manage to get around those
> problems on x86. GCC should be able to do something.
IIRC, VC++ just sets rounding precision to 64. So you have
excess precision if you try to do stuff with "float" but
don't with "double".
> | 3) Set the FP reg rounding precision to 64 bits. This has the flaw
> | that you can no longer perform XFmode operations. This is only a
> | partial fix, in that we still have excess precision problems for
> | SFmode operations.
>
> Agreed. But wouldn't it be better than current situation?
No, not better, just different. You anger a different set of folk.
r~
More information about the Gcc
mailing list