[Bug tree-optimization/49542] New: [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[Bug tree-optimization/49542] New: [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

           Summary: [4.7 regression] 2 failures in the vectorization
                    testsuite in 64-bit mode
           Product: gcc
           Version: 4.7.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: ebotcazou@gcc.gnu.org
            Target: sparc*-*-*


2 failures have recently appeared in the vectorization testsuite in 64-bit
mode:

FAIL: gcc.dg/vect/pr33804.c scan-tree-dump-times vect "vectorized 1 loops" 1
FAIL: gcc.dg/vect/pr33804.c scan-tree-dump-times vect "vectorizing stmts using
S
LP" 1
FAIL: gcc.dg/vect/pr33804.c -flto scan-tree-dump-times vect "vectorized 1
loops"
 1
FAIL: gcc.dg/vect/pr33804.c -flto scan-tree-dump-times vect "vectorizing stmts
u
sing SLP" 1

FAIL: gcc.dg/vect/slp-multitypes-3.c scan-tree-dump-times vect "vectorized 1
loo
ps" 1
FAIL: gcc.dg/vect/slp-multitypes-3.c scan-tree-dump-times vect "vectorizing
stmt
s using SLP" 2
FAIL: gcc.dg/vect/slp-multitypes-3.c -flto scan-tree-dump-times vect
"vectorized
 1 loops" 1
FAIL: gcc.dg/vect/slp-multitypes-3.c -flto scan-tree-dump-times vect
"vectorizin
g stmts using SLP" 2

Both tests are already XFAILed in 32-bit mode (ilp32 tag).

[Bug tree-optimization/49542] [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.7.0

[Bug tree-optimization/49542] [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |irar at gcc dot gnu.org

--- Comment #1 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2011-07-04 15:18:57 UTC ---
Still present.  Ira, should they be XFAILed in 64-bit mode as well?

[Bug tree-optimization/49542] [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[irar at il dot ibm.com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

Ira Rosen <irar at il dot ibm.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |irar at il dot ibm.com

--- Comment #2 from Ira Rosen <irar at il dot ibm.com> 2011-07-05 06:15:23 UTC ---
I don't know. Why do they fail? 
You wrote (here http://gcc.gnu.org/ml/gcc-patches/2009-11/msg00508.html)
> They pass on SPARC 64-bit because the operations are 
> transformed to use the word mode.
It doesn't work anymore?

Thanks,
Ira

[Bug tree-optimization/49542] [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2011-07-05 15:42:55 UTC ---
> You wrote (here http://gcc.gnu.org/ml/gcc-patches/2009-11/msg00508.html)
> > They pass on SPARC 64-bit because the operations are 
> > transformed to use the word mode.
> It doesn't work anymore?

No, for pr33804.c the word mode trick isn't invoked anymore because in:

  if (icode == CODE_FOR_nothing)
    {
      if (vect_print_dump_info (REPORT_DETAILS))
        fprintf (vect_dump, "op not supported by target.");
      /* Check only during analysis.  */
      if (GET_MODE_SIZE (vec_mode) != UNITS_PER_WORD
          || (vf < vect_min_worthwhile_factor (code)
              && !vec_stmt))
        return false;
      if (vect_print_dump_info (REPORT_DETAILS))
        fprintf (vect_dump, "proceeding using word mode.");
    }

(gdb) p debug_gimple_stmt(stmt)
D.2015_14 = D.2014_13 + D.2012_9;
(gdb)  p vf
$9 = 2
(gdb) call vect_min_worthwhile_factor (code)
$10 = 4
(gdb) p vec_stmt
$11 = (gimple *) 0x0

vf is 8 on the 4.6 branch instead.  Likewise for slp-multitypes-3.c.

[Bug tree-optimization/49542] [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[irar at il dot ibm.com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

--- Comment #4 from Ira Rosen <irar at il dot ibm.com> 2011-07-06 06:20:06 UTC ---
In that case they should fail. The vf in 4.7 is correct. In 4.6 it is updated
after stmts analysis, which may cause inconsistency. This patch
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02210.html fixed this by moving vf
update before the analysis.

[Bug testsuite/49542] [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011.07.06 08:35:19
          Component|tree-optimization           |testsuite
     Ever Confirmed|0                           |1

--- Comment #5 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2011-07-06 08:35:19 UTC ---
Thanks for the diagnosis; I'm going to adjust the testcases.

[Bug testsuite/49542] [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

--- Comment #6 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2011-07-06 08:46:45 UTC ---
Author: ebotcazou
Date: Wed Jul  6 08:46:41 2011
New Revision: 175905

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=175905
Log:
    PR testsuite/49542
    * gcc.dg/vect/pr33804.c: XFAIL if vect_no_align unconditionally.
    * gcc.dg/vect/slp-multitypes-3.c: XFAIL on SPARC unconditionally.

Modified:
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/vect/pr33804.c
    trunk/gcc/testsuite/gcc.dg/vect/slp-multitypes-3.c

[Bug testsuite/49542] [4.7 regression] 2 failures in the vectorization testsuite in 64-bit mode

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49542

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #7 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2011-07-06 08:48:59 UTC ---
The vectorizer testsuite is clean again on SPARC.