------- Comment #1 from dberlin at gcc dot gnu dot org 2006-11-08 16:24 -------
(In reply to comment #0)
> In the testcase I posted, USE's initial_condition_in_loop_num is
> -5 and X's initial_condition_in_loop_num is 1. And, adding -5 - 1
> to perfectiv (aka Y in the above routine) is exactly what would
> cure this testcase.
>
> Do you agree with this?
>
I agree completely with your analysis. If you want to work up a fix to do
that, i'm happy to approve it (It should suffice to generate the initial
condition in the loop header and use that as the starting point for the new iv
variable).
> Not sure how hard is that though, other alternative would be to make
> can_convert_to_perfect_nest more conservative.
>
--
dberlin at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Last reconfirmed|0000-00-00 00:00:00 |2006-11-08 16:24:55
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29581
