This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [RFC] Redundant zero store elimination (PR31150)
- From: "Diego Novillo" <dnovillo at google dot com>
- To: "Jakub Jelinek" <jakub at redhat dot com>
- Cc: "Richard Guenther" <rguenther at suse dot de>, "Daniel Berlin" <dberlin at dberlin dot org>, gcc-patches at gcc dot gnu dot org
- Date: Sat, 29 Sep 2007 14:43:40 -0400
- Subject: Re: [RFC] Redundant zero store elimination (PR31150)
- References: <20070907160146.GS2063@devserv.devel.redhat.com>
On 9/7/07, Jakub Jelinek <jakub@redhat.com> wrote:
> Also I'd appreciate any comments about the optimization itself...
I think this optimization should/could be done generically as part of
DSE. As with the patch that was extending DCE to do some of DSE's
tasks, I think it is misguided for us to do these one-of passes just
because the pass that's supposed to take care of this is
limited/broken.
I would much rather spend the energy implementing a proper DSE pass.
I am going to spend some time working on a new DSE formulation, but
this will not happen overnight.
Unless this fixes a horrendous performance problem, I would rather not
apply another hack on top of DSE. Having said that, I am willing to
not block this patch if there is enough support for it. But I cannot
bring myself to supporting this approach. Sorry.