On Wednesday 31 May 2006 18:29, Hans-Werner Hilse wrote:
> Hi,
>
> On Thu, 1 Jun 2006 00:51:47 +0930
>
> Raymond Lewis Rebbeck <[EMAIL PROTECTED]> wrote:
> > On Thursday, 1 June 2006 0:38, Hemmann, Volker Armin wrote:
> > > > > Also -ftracer is not in the safe cflags list, personally I would
> > > > > not use it but if you believe it benefits you then go ahead.
> > > >
> > > > Another handy tip. Can't remember why I had it (did the "research"
> > > > when I reinstalled it 32bit). I'll remove it too.
> > >
> > > ftracer is harmless.
> > > From man gcc:
> > > -ftracer
> > > Perform tail duplication to enlarge superblock size. This
> > > trans- formation simplifies the control flow of the function allowing
> > > other optimizations to do better job.
> >
> > If it was harmless and beneficial it'd already be included in an -O?
> > level.
>
> Probably. And it seems to be only of interest when using
> -fsched2-use-superblocks or -fsched2-use-traces. The man page entry for
> the further (included in the latter) option says: "This option is
> experimental, as not all machine descriptions used by GCC model the CPU
> closely enough to avoid unreliable results from the algorithm."
fsched2-use-traces
Use -fsched2-use-superblocks algorithm when scheduling after regis-
ter allocation and additionally perform code duplication in order
to increase the size of superblocks using tracer pass. See
-ftracer for details on trace formation.
This mode should produce faster but significantly longer programs.
Also without -fbranch-probabilities the traces constructed may not
match the reality and hurt the performance. This only makes sense
when scheduling after register allocation, i.e. with -fsched-
ule-insns2 or at -O2 or higher.
where does it say 'experimental'?
And the entry for ftracer is above.
--
gentoo-user@gentoo.org mailing list
