This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] gfortran.dg/vect/vect-5: don't xfail lp64 revised
- From: Dorit Nuzman <DORIT at il dot ibm dot com>
- To: howarth at bromo dot msbb dot uc dot edu (Jack Howarth)
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Tue, 19 Sep 2006 16:11:30 +0300
- Subject: Re: [patch] gfortran.dg/vect/vect-5: don't xfail lp64 revised
howarth@bromo.msbb.uc.edu (Jack Howarth) wrote on 19/09/2006 15:06:34:
> Dorit,
> I've added back in the '"vectorized 2 loops" 1 "vect"'
> test line and verified the revised patch works on both Darwin PPC
> at -m64 and x86_64. Okay for trunk?
This sounds like the right thing to do to me (I can't approve patches
though).
Since with this patch we'll get a failure on this line on ia64, we'll need
to open a PR when this patch goes in, or figure out before then why ia64
fails vectorization and check for the right condition. Can someone send the
output of -fdump-tree-vect-details on ia64 for this testcase?
thanks,
dorit
> Jack
>
>
> 2006-09-19 Jack Howarth <howarth@bromo.med.uc.edu>
>
> * gfortran.dg/vect/vect-5.f90: Don't xfail lp64.
>
> Index: gcc/testsuite/gfortran.dg/vect/vect-5.f90
> ===================================================================
> --- gcc/testsuite/gfortran.dg/vect/vect-5.f90 (revision 117054)
> +++ gcc/testsuite/gfortran.dg/vect/vect-5.f90 (working copy)
> @@ -35,9 +35,9 @@
> stop
> end
>
> -! { dg-final { scan-tree-dump-times "vectorized 2 loops" 1 "vect" {
> xfail { lp64 } } } }
> -! { dg-final { scan-tree-dump-times "Alignment of access forced
> using peeling" 1 "vect" { xfail { vect_no_align || lp64 } } } }
> -! { dg-final { scan-tree-dump-times "Vectorizing an unaligned
> access" 2 "vect" { xfail { vect_no_align || lp64 } } } }
> +! { dg-final { scan-tree-dump-times "vectorized 2 loops" 1 "vect" } }
> +! { dg-final { scan-tree-dump-times "Alignment of access forced
> using peeling" 1 "vect" { xfail { vect_no_align } } } }
> +! { dg-final { scan-tree-dump-times "Vectorizing an unaligned
> access" 2 "vect" { xfail { vect_no_align } } } }
> ! { dg-final { scan-tree-dump-times "Alignment of access forced
> using versioning." 3 "vect" { target { ilp32 && vect_no_align } } } }
>
> ! We also expect to vectorize one loop for lp64 targets that support