This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: -ffast-math and gfortran
- From: howarth at bromo dot msbb dot uc dot edu (Jack Howarth)
- To: howarth at bromo dot msbb dot uc dot edu, jvdelisle at verizon dot net
- Cc: fortran at gcc dot gnu dot org, fxcoudert at gmail dot com, sgk at troutmask dot apl dot washington dot edu, toon at moene dot indiv dot nluug dot nl
- Date: Sun, 4 Jun 2006 17:24:32 -0400 (EDT)
- Subject: Re: -ffast-math and gfortran
Jerry,
The point isn't so much what sort of speed increase can be
obtained but rather what sort of support problems will be created
by leaving the feature in place to bitrot. To that extent I agree
with Steve. However since folks like Toon seem to be using this
feature regularly, I still wonder what percentage of the problems
seen are inappropriate application of -ffast-math to source code
that can't support it versus actually breakage in the -ffast-math
option. Also, are the same level of problems to be seen from using
-ffast-math in gcc/g++? If so, then maybe this should be discussed
on the gcc list. Either the feature should be testcased and
debugged or removed.
Jack