[Bug tree-optimization/28794] New: missed optimization with non COND_EXPR and vrp and comparisions

[Bug tree-optimization/28794] New: missed optimization with non COND_EXPR and vrp and comparisions

[pinskia at gcc dot gnu dot org]

int f(int x, int y)
{
  int t;
  for (t = 0; t < 50; t++)
    g(t>0);
}
void f1(int x, int y)
{
  int t;
  for (t = 0; t < 50; t++)
    g(t!=0);
}

--------------
The above two functions should produce the same code with f1 being better than
f.

If we change it to:
void f2(int x, int y)
{
  int t;
  for (t = 0; t < 50; t++)
  {
    int tt;
    if (t>0)
      tt = 1;
   else
      tt = 0;
   g(tt);
  }
}

-----
We get f1 so we are only folding comparisions in a COND_EXPR which is wrong, we
should also be doing them in MODIFY_EXPRs too.


-- 
           Summary: missed optimization with non COND_EXPR and vrp and
                    comparisions
           Product: gcc
           Version: 4.2.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: enhancement
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pinskia at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

[Bug tree-optimization/28794] missed optimization with non COND_EXPR and vrp and comparisions

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2006-08-21 23:31 -------
I found this while trying to figure out how to get VRP to optimize:
a_1 != 0 into a_1 if the range of a_1 is [0,1] (well with a NOP_EXPR).
If I do it inside simplify_cond_using_ranges, I miss all the MODIFY_EXPRs.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

[Bug tree-optimization/28794] missed optimization with non-COND_EXPR and vrp and comparisions

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2006-08-21 23:52 -------
For x86, there is no difference in the code gen for f and f1/f2, but for PPC32,
there is:
for f1/f2:
        addic %r0,%r31,-1
        subfe %r3,%r0,%r31

For f:
        srawi %r3,%r31,31
        subf %r3,%r31,%r3
        srwi %r3,%r3,31


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|missed optimization with non|missed optimization with
                   |COND_EXPR and vrp and       |non-COND_EXPR and vrp and
                   |comparisions                |comparisions


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

[Bug tree-optimization/28794] missed optimization with non-COND_EXPR and vrp and comparisions

[rguenth at gcc dot gnu dot org]

------- Comment #3 from rguenth at gcc dot gnu dot org  2006-08-22 08:10 -------
Confirmed.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rguenth at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2006-08-22 08:10:03
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

[Bug tree-optimization/28794] missed optimization with non-COND_EXPR and vrp and comparisions

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED
   Target Milestone|---                         |14.0

--- Comment #9 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Fixed finally.

[Bug tree-optimization/28794] missed optimization with non-COND_EXPR and vrp and comparisions

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

--- Comment #8 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The trunk branch has been updated by Andrew Pinski <pinskia@gcc.gnu.org>:

https://gcc.gnu.org/g:aadc5c07feb0ab08729ab25d0d896b55860ad9e6

commit r14-3084-gaadc5c07feb0ab08729ab25d0d896b55860ad9e6
Author: Andrew Pinski <apinski@marvell.com>
Date:   Mon Aug 7 00:05:21 2023 -0700

    VR-VALUES [PR28794]: optimize compare assignments also

    This patch fixes the oldish (2006) bug where VRP was not
    optimizing the comparison for assignments while handling
    them for GIMPLE_COND only.
    It just happens to also solves PR 103281 due to allowing
    to optimize `c < 1` to `c == 0` and then we get
    `(c == 0) == c` (which was handled by r14-2501-g285c9d04).

    OK? Bootstrapped and tested on x86_64-linux-gnu with no
    regressions.

            PR tree-optimization/103281
            PR tree-optimization/28794

    gcc/ChangeLog:

            * vr-values.cc
(simplify_using_ranges::simplify_cond_using_ranges_1): Split out
            majority to ...
            (simplify_using_ranges::simplify_compare_using_ranges_1): Here.
            (simplify_using_ranges::simplify_casted_cond): Rename to ...
            (simplify_using_ranges::simplify_casted_compare): This
            and change arguments to take op0 and op1.
            (simplify_using_ranges::simplify_compare_assign_using_ranges_1):
New method.
            (simplify_using_ranges::simplify): For tcc_comparison assignments
call
            simplify_compare_assign_using_ranges_1.
            * vr-values.h (simplify_using_ranges): Add
            new methods, simplify_compare_using_ranges_1 and
simplify_compare_assign_using_ranges_1.
            Rename simplify_casted_cond and simplify_casted_compare and
            update argument types.

    gcc/testsuite/ChangeLog:

            * gcc.dg/tree-ssa/pr103281-1.c: New test.
            * gcc.dg/tree-ssa/vrp-compare-1.c: New test.

[Bug tree-optimization/28794] missed optimization with non-COND_EXPR and vrp and comparisions

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Created attachment 55700
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=55700&action=edit
Patch, still needs testcase

[Bug tree-optimization/28794] missed optimization with non-COND_EXPR and vrp and comparisions

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I have a patch now.

[Bug tree-optimization/28794] missed optimization with non-COND_EXPR and vrp and comparisions

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |pinskia at gcc dot gnu.org

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I am going to finally optimize this.

My plan is to change simplify_cond_using_ranges_1 to take 3 reference arguments
and return true if a change to them.

[Bug tree-optimization/28794] missed optimization with non-COND_EXPR and vrp and comparisions

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=28794

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2020-01-17 00:00:00         |2021-6-8

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note really the loops at -O3 should really be optimized to:
void f3(int x, int y)
{
  int t;
  g(0);
  for (t = 1; t < 50; t++)
  {
   g(1);
  }
}

That is peel off the first iteration.  I don't know why -O3 -fno-ssa-phiopt
-fno-tree-vrp does not do that either. I will file that as a different bug
because this is about converting the comparison only in COND_EXPR gimple.