This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PR 10776
> > On Sun, Jan 11, 2004 at 12:02:26PM +0100, Jan Hubicka wrote:
> > > There is yet another quadratic loop in reload itself when looking for
> > > equal registers, but I didn't find any testcase for it, perhaps it fires
> > > for you?
> > > What settings do you use?
> >
> > ./configure on i686-linux. Run ./cc1 -O1 test.c. I think, given how
> > we're configuring now, direct invoccation of cc1 implies i386 code
> > generation. Not sure.
>
> I see, I used ./xgcc -B ./. The code generated for i686 differs from
> i386, but for ./cc1plus -O1 ~/pr10776.cpp I get similar timings.
> I am attaching the timmings and the precise testcase I use.
> (I had to edit it little bit in order to compile)
>
> If you have different testcase, can you please forward it to me? If it
> reproduces the quadratic behaviour of find_equiv_reg, I would like to
> look into it. Bugzilla has one very old testcase for this too, but it
> no longer compile on new C++.
>
> I will also double check the patch on clean tree now.
Oops, stupid me, I forgot the patch to limit reload search in the tree
too, so I didn't hit the quadratic behaviour.
I will send patch for that shortly.
Honza