This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
RE: Tree SSA measurement in CSiBE
- From: "S. Bosscher" <S dot Bosscher at student dot tudelft dot nl>
- To: 'Gábor Lóki ' <loki at inf dot u-szeged dot hu>,"'s dot bosscher at student dot tudelft dot nl '" <s dot bosscher at student dot tudelft dot nl>
- Cc: 'Arpad Beszedes ' <beszedes at inf dot u-szeged dot hu>,"'gcc at gcc dot gnu dot org '" <gcc at gcc dot gnu dot org>,"'dnovillo at redhat dot com '" <dnovillo at redhat dot com>
- Date: Thu, 15 Jan 2004 15:12:23 +0100
- Subject: RE: Tree SSA measurement in CSiBE
> > > > Are you compiling with checking disabled?
> > >
> > > We use only the -Os or -O2 switches. Is this ok?
> >
> > Depends on how you configured the compiler ;-)
> We don't use --disable-checking flag for configure,
> neither for mainline nor for tree-ssa.
If you are interested in comparing compile times, then you should
disable checking. Checking is far more expensive on tree-ssa,
because there are more trees to check (so tree checking becomes
expensive), and the extra control flow and SSA checks, especially
the latter, are very expensive too.
So comparing compile times with checking enabled is not meaningful.
> There'll be a measurement document soon which describes the whole
> config/build process about how to create these results.
That will be interesting to read. Thanks.
Gr.
Steven