This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [RFC] Mixed results with O(1) PHI arg lookup
- From: Dorit Naishlos <DORIT at il dot ibm dot com>
- To: law at redhat dot com
- Cc: gcc at gcc dot gnu dot org, Kazu Hirata <kazu at cs dot umass dot edu>, stevenb at suse dot de
- Date: Thu, 11 Nov 2004 12:02:09 +0200
- Subject: Re: [RFC] Mixed results with O(1) PHI arg lookup
> > The code that directly writes PHI_ARG_EDGE in the vectorizer will go
away
> > once this patch is installed:
> > http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00283.html
> Is there any way we can break out those changes?
I'll try to prepare a patch for this relative to the old (current) peeling
scheme.
dorit
Jeffrey A Law
<law@redhat.com> To: Dorit Naishlos/Haifa/IBM@IBMIL
cc: Kazu Hirata <kazu@cs.umass.edu>, gcc@gcc.gnu.org, stevenb@suse.de
09/11/2004 18:57 Subject: Re: [RFC] Mixed results with O(1) PHI arg lookup
Please respond to
law
On Sun, 2004-11-07 at 15:15 +0200, Dorit Naishlos wrote:
>
>
>
> > The patch survived a bootstrap, but it should not pass testsuite
> > because I have not fixed a part of tree_vectorizer.c that directly
> > writes PHI_ARG_EDGE.
>
> The code that directly writes PHI_ARG_EDGE in the vectorizer will go away
> once this patch is installed:
> http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00283.html
Is there any way we can break out those changes? I'm certainly not
qualified to review that entire changeset, but I might be inclined
to take the time to understand a more localized change.
jeff