This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [lno] [patch] bring over vectorizer changes from mainline
- From: Daniel Berlin <dberlin at dberlin dot org>
- To: Dorit Naishlos <DORIT at il dot ibm dot com>
- Cc: gcc-patches at gcc dot gnu dot org, Mark Mitchell <mark at codesourcery dot com>
- Date: Sun, 03 Oct 2004 10:58:01 -0400
- Subject: Re: [lno] [patch] bring over vectorizer changes from mainline
- References: <OFC6AA4286.56C8469E-ONC2256F22.002F96FB-C2256F22.004C6E17@il.ibm.com>
On Sun, 2004-10-03 at 15:54 +0200, Dorit Naishlos wrote:
> > The pointer alignment pass on lno needs to be rewritten to use tree-ssa-
> > propagate, which exists on mainline but not lno.
> > I can rewrite it just for mainline if you planned on submitting it for
> > 4.0, i can't remember whether that was the case or not.
>
> Unfortunately I didn't explicitly mention it on the vectorizer patch list
> for 4.0, although it is used by the vectorizer in lno. I guess it's too
> late now?
>
Dunno from Mark's perspective.
The pass has been working fine for many moons, and isn't a
transformation itself (so it's very low risk).
I said it "needs" to be moved to tree-ssa-propagate, but this is just a
cleanup, of course. It was always planned, I just wrote tree-ssa-align
before diego had finished the propagation engine stuff, and moved ccp to
it.
If it's not moved to tree-ssa-propagate, it just means there will be
some duplicated code, as there is now.
In other words, it could be submitted as is, with the cleanup scheduled
for 4.1, if we wanted to do that, and Mark felt that was better in some
way.
Or Mark just might not want it at all at this point.
> dorit
>
--