[Bug tree-optimization/56826] New: Run-fail after r197189.

[Bug tree-optimization/56826] New: Run-fail after r197189.

[ysrumyan at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56826

             Bug #: 56826
           Summary: Run-fail after r197189.
    Classification: Unclassified
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: ysrumyan@gmail.com


We got RTF for 435.gromacs from spec2006 suite. I attached simple reproducer
which if failed if it is compiled with -O3 -funroll-loops options but passed at
-O2:

gcc -O3 -funroll-loops -ffast-math -march=corei7 t.c t_main.c -o tt2
./tt2
Test failed
gcc -O2 -ffast-math -march=corei7 t.c t_main.c -o tt1  
./tt1
Test passed

[Bug tree-optimization/56826] Run-fail after r197189.

[ysrumyan at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56826

--- Comment #1 from Yuri Rumyantsev <ysrumyan at gmail dot com> 2013-04-03 09:48:59 UTC ---
Created attachment 29789
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29789
test-case to reproduce

To reproduce tha failure the test must be compiled on x86 corei7 with the
following options:
-O3 -funroll-loops -ffast-math -march=corei7

[Bug tree-optimization/56826] [4.9 Regression] Run-fail after r197189.

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56826

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2013-04-04
         AssignedTo|unassigned at gcc dot       |rguenth at gcc dot gnu.org
                   |gnu.org                     |
   Target Milestone|---                         |4.9.0
            Summary|Run-fail after r197189.     |[4.9 Regression] Run-fail
                   |                            |after r197189.
     Ever Confirmed|0                           |1

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> 2013-04-04 11:06:58 UTC ---
Confirmed, mine.

[Bug tree-optimization/56826] [4.9 Regression] Run-fail after r197189.

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56826

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> 2013-04-04 11:12:56 UTC ---
Testcase:

extern void abort (void);

typedef struct {
    int a[3];
    int num;
} t1;
t1 B[100];
int A[300];

void __attribute__((noinline,noclone))
bar (int *A, t1 *B, int n)
{
  int i;
  int *a = A;
  for (i=0; i<n; i++, a+=3)
    {
      a[0] = B[i].a[0];
      a[1] = B[i].a[1];
      a[2] = B[i].a[2];
    }
}

int main()
{
  int i;
  for (i=0; i<100; i++)
    {
      B[i].num = i;
      B[i].a[0] = i * 3;
      B[i].a[1] = i * 3 + 1;
      B[i].a[2] = i * 3 + 2;
      __asm__ volatile ("");
    }
  bar (&A[0], &B[0], 100);
  for (i=0; i<300; i++)
    if (A[i] != i)
      abort ();
  return 0;
}

[Bug tree-optimization/56826] [4.9 Regression] Run-fail after r197189.

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56826

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> 2013-04-04 11:41:03 UTC ---
Ok, the bug is that 'ncopies' in vect_build_slp_tree is not computed correctly.
I can trivially incrementally improve it but the ultimate fix is to delay
the checks done for vectorizing SLP loads / stores to after the whole SLP
tree is built and thus all types participating in the SLP are known and we
can compute the final unroll factor (in vect_make_slp_decision).

Thus, I'll push the trivial improvement for now, the real fix will follow
a SLP tree build re-org I am planning for next week anyway.

[Bug tree-optimization/56826] [4.9 Regression] Run-fail after r197189.

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56826

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

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

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> 2013-04-04 15:07:00 UTC ---
Author: rguenth
Date: Thu Apr  4 15:06:44 2013
New Revision: 197486

URL: http://gcc.gnu.org/viewcvs?rev=197486&root=gcc&view=rev
Log:
2013-04-04  Richard Biener  <rguenther@suse.de>

    PR tree-optimization/56826
    * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
    more accurately.

    * gcc.dg/vect/pr56826.c: New testcase.
    * gcc.dg/vect/O3-pr36098.c: Adjust.

Added:
    trunk/gcc/testsuite/gcc.dg/vect/pr56826.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/testsuite/ChangeLog
    trunk/gcc/testsuite/gcc.dg/vect/O3-pr36098.c
    trunk/gcc/tree-vect-slp.c

[Bug tree-optimization/56826] [4.9 Regression] Run-fail after r197189.

[Dmitriy.Poterukha at uvoteam dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56826

creeon <Dmitriy.Poterukha at uvoteam dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |Dmitriy.Poterukha at uvoteam dot c
                   |                            |om

--- Comment #6 from creeon <Dmitriy.Poterukha at uvoteam dot com> ---
Most wanted essay writing servicec in U.S.
https://writingservice.essayhave.com/