This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: gcc on Intel x86 and failure of -ffloat-store option
- To: beebe at math dot utah dot edu, gcc-bugs at gcc dot gnu dot org
- Subject: Re: gcc on Intel x86 and failure of -ffloat-store option
- From: N8TM at aol dot com
- Date: Wed, 8 Dec 1999 21:32:09 EST
In a message dated 12/8/99 2:07:53 PM Pacific Standard Time,
beebe@math.utah.edu writes:
> -ffloat-store is ignored when optimization is
> turned on
It's not ignored, but the recent compiler versions often hoist the storage
outside of the loop. There are examples where -O2 causes some (but not all)
loop variables to be subjected to rounding to 53 bits each time, while adding
-ffloat-store causes the rounding to be postponed until after the loop. Not
the expected behavior.
Tim
tprince@computer.org