This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Memory partitioning considered harmful (!?)
On Mon, 29 Jan 2007, Daniel Berlin wrote:
> On 1/29/07, Richard Guenther <rguenther@suse.de> wrote:
> > On Mon, 29 Jan 2007, Daniel Berlin wrote:
> >
> > > On 1/29/07, Richard Guenther <rguenther@suse.de> wrote:
> >
> > > There are certainly better partitioning heuristics than what we do
> > > now, and i know diego has said he welcomes any tuning of the
> > > heuristics we have to get better results.
> > > ways to partition than we do now.
> >
> > Btw. can someone do an exemplary "fixing" of lim / lsm to "look at" MPTs
> > to disambiguate the loads/stores again? I recall this was possible
> > somehow.
>
> While possible, you are probably better off making the partitioning do
> what you want :)
Hmm, like "don't partition vars used as induction variables or in loops"?
At least the proposed patch, while a big hammer, makes it "work". (I put
it on the C++ tester, so we'll see if it fixes tramp3d, too)
Richard.
--
Richard Guenther <rguenther@suse.de>
Novell / SUSE Labs