| 1 |
On Friday 09 June 2006 19:11, Vladimir G. Ivanovic wrote: |
| 2 |
> On Fri, 2006-06-09 at 15:36 +0200, Hemmann, Volker Armin wrote: |
| 3 |
> > > Haven't noticed any problems on my laptop or a workstation. |
| 4 |
> > |
| 5 |
> > because they were filtered out. |
| 6 |
> > |
| 7 |
> > DO NEVER USE FFAST-MATH! |
| 8 |
> > |
| 9 |
> > Apps that can use it safely, usually set it in their makefile. |
| 10 |
> > |
| 11 |
> > Everything else is prone to break. |
| 12 |
> > |
| 13 |
> > And that is the reason, that almost all ebuilds have 'filter-flags' |
| 14 |
> > and 'strip-flags'. |
| 15 |
> > |
| 16 |
> > Again: NEVER USE IT. |
| 17 |
> > |
| 18 |
> > And: NEVER TELL ANYBODY TO USE IT. |
| 19 |
> |
| 20 |
> If -ffast-math is filtered or stripped out, there is no harm in leaving |
| 21 |
> it in CFLAGS, right? |
| 22 |
|
| 23 |
no, because a lot of other flags are filtered too! |
| 24 |
Some ebuilds filter ALL flags, only because some ricers had to use ffast-math |
| 25 |
and other 'controversial' flags. |
| 26 |
|
| 27 |
And some day, you will hit a package, that will built a damaged binary doing |
| 28 |
the wrong stuff, that does not filter out, because nobody knew about it. And |
| 29 |
then you are f*. |
| 30 |
|
| 31 |
> |
| 32 |
> But, on my system, only 14 packages filter out -ffast-math: gnubg |
| 33 |
> postgresql pgcluster libpq zoom mpfr gmp octave openoffice gsl goffice |
| 34 |
> rrdtool xv gimp. None strip it out. So, the huge majority of the |
| 35 |
> packages on my system are compiled with -ffast-math, unless I've made a |
| 36 |
> mistake in grepping for "fast-math" in ebuilds that contain |
| 37 |
> "filter-flags". |
| 38 |
|
| 39 |
and look for 'strip-flags' |
| 40 |
|
| 41 |
|
| 42 |
> http://gcc.gnu.org/ml/gcc/2004-03/msg01511.html |
| 43 |
> For me, it's very simple: If -ffast-math leads to answers that are |
| 44 |
> less accurate (in the verification-against-observation-sense) or unphysical |
| 45 |
> (against theoretical limit analysis), then I'll inspect my Fortran code and |
| 46 |
> repair the formulation (either a single expression or the layout of loop |
| 47 |
> code) that is responsible for the mayhem. |
| 48 |
> |
| 49 |
> Under no circumstances I will give up using -ffast-math. |
| 50 |
> |
| 51 |
> |
| 52 |
> http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/320 |
| 53 |
>35.pdf -ffast-math is recommended by AMD. |
| 54 |
|
| 55 |
FOR SPECIAL APPS. AND THIS DOCUMENT IS MEANT FOR PROGRAMMERS! PEOPLE WHO KNOW |
| 56 |
THEIR CODE! |
| 57 |
|
| 58 |
> |
| 59 |
> These opinions seem to contradict your advice. Could you be more |
| 60 |
> specific about why -ffast-math should not be used? |
| 61 |
|
| 62 |
no, they don't contradict me. You did not understand them. |
| 63 |
|
| 64 |
Nowhere does AMD tells endusers to build their complete system with that flag. |
| 65 |
-- |
| 66 |
gentoo-amd64@g.o mailing list |
Archives