This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: float conversion optimizations
- From: Jan Hubicka <jh at suse dot cz>
- To: Brad Lucier <lucier at math dot purdue dot edu>
- Cc: jh at suse dot cz, gcc-patches at gcc dot gnu dot org, rth at redhat dot com
- Date: Mon, 13 Jan 2003 18:17:34 +0100
- Subject: Re: float conversion optimizations
- References: <200301131546.h0DFk0I03499@banach.math.purdue.edu>
> I'm trying to work through your code at
>
> http://gcc.gnu.org/ml/gcc-patches/2003-01/msg00807.html
> http://gcc.gnu.org/ml/gcc-patches/2003-01/msg00809.html
> http://gcc.gnu.org/ml/gcc-patches/2003-01/msg00810.html
>
> The first two seem OK for IEEE arithmetic (because of the specific properties
> of the exponent and mantissa in float and double), but in avoiding a
> double rounding (meaning rounding a temporary result to double
> precision for the add and then rounding again to single precision for
> the store) it seems that the third "optimization" can change the
Where actually can the rounding happen? both operands fits in the
floats and are just extended. I am not quite sure but I seem to
remember that in such particular case the double rouding will be
equivalent to single rouding, but I may be wrong...
Thanks!
Honza
> results of the code and hence should only be activated if
> -funsafe-math-optimizations.
>
> Brad