http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36928
Thomas Koenig tkoenig at gcc dot gnu.org changed:
What|Removed |Added
Status|NEW |RESOLVED
--- Comment #13 from tkoenig at gcc dot gnu dot org 2010-05-31 20:22
---
Subject: Bug 36928
Author: tkoenig
Date: Mon May 31 20:22:10 2010
New Revision: 160085
URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=160085
Log:
2010-05-31 Thomas Koenig tkoe...@gcc.gnu.org
PR
--- Comment #14 from tkoenig at gcc dot gnu dot org 2010-06-01 05:30
---
Fixed for the constant case.
Still to do, but much harder: A case like
a(n,m,2) = a(n+1,m,2).
Unassigning myself for now.
--
tkoenig at gcc dot gnu dot org changed:
What|Removed
--- Comment #10 from tkoenig at gcc dot gnu dot org 2010-05-22 10:11
---
I've gotten a bit further with this.
For x(la:ua:sa) = x(lb,ub,sb), there can be no collision
if abs(la-lb) mod gcd(sa, sb) == 0
where gcd is the greatest common divisor.
This will at least fix the test case
--- Comment #11 from dominiq at lps dot ens dot fr 2010-05-22 10:50 ---
For x(la:ua:sa) = x(lb,ub,sb), there can be no collision
if abs(la-lb) mod gcd(sa, sb) == 0
where gcd is the greatest common divisor.
You probably mean if abs(la-lb) mod gcd(sa, sb) != 0 (assuming
--- Comment #12 from tkoenig at gcc dot gnu dot org 2010-05-22 21:20
---
(In reply to comment #11)
You probably mean if abs(la-lb) mod gcd(sa, sb) != 0 (assuming
x(lb:ub:sb);-).
Yes, I had this reversed when I wrote this.
Note that if I am not mistaken, this result extends to
--- Comment #8 from tkoenig at gcc dot gnu dot org 2010-05-16 09:00 ---
Richard, what do you think of this?
Does it make sense to do the dependency analysis in the
front end?
Does Graphite (about which I know next to nothing, I admit) have
the necessary infrastructure to detect the
--- Comment #9 from rguenth at gcc dot gnu dot org 2010-05-16 10:30 ---
It makes sense to do this in the frontend. The worst thing is when the
frontend
creates array temporaries - those are really really hard to get rid of in the
middle-end.
There are basically two (or maybe two and a
--- Comment #3 from tkoenig at gcc dot gnu dot org 2010-05-15 15:55 ---
I'm working on this (designing an algorithm so far).
It is an interesting problem.
--
tkoenig at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #4 from dominiq at lps dot ens dot fr 2010-05-15 17:05 ---
It is an interesting problem.
Should not it be handled by the middle-end, possibly with the help of graphite?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36928
--- Comment #5 from tkoenig at gcc dot gnu dot org 2010-05-15 17:30 ---
(In reply to comment #4)
It is an interesting problem.
Should not it be handled by the middle-end, possibly with the help of
graphite?
If we can improve dependency analysis in the front end, where we
--- Comment #6 from dominiq at lps dot ens dot fr 2010-05-15 18:01 ---
If we can improve dependency analysis in the front end, where we (potentially)
have some more information, it should not hurt the middle-end.
OTOH, if somebody is already doing work in this direction, it would
--- Comment #7 from jvdelisle at gcc dot gnu dot org 2010-05-15 18:13
---
I suggest you give a chat to Richard Guenther who seems quite up on the
optimisation aspects of things. If we are creating a temporary in the front
end I think that would be the least optimal approach.
--
--- Comment #2 from dfranke at gcc dot gnu dot org 2010-05-09 20:28 ---
Still valid with: gcc version 4.6.0 20100509 (experimental) (GCC)
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36928
--
jv244 at cam dot ac dot uk changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Last
--- Comment #1 from burnus at gcc dot gnu dot org 2008-07-25 08:42 ---
See also PR 36915. This is one of the items where the middle-end array
expressions would help (see links in PR 36915). The ME array expr. support is
planned for 4.5.
--
16 matches
Mail list logo