[Bug tree-optimization/98137] New: Could use SLP to vectorize if split_constant_offset were smarter

[Bug tree-optimization/98137] New: Could use SLP to vectorize if split_constant_offset were smarter

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98137

            Bug ID: 98137
           Summary: Could use SLP to vectorize if split_constant_offset
                    were smarter
           Product: gcc
           Version: 11.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rguenth at gcc dot gnu.org
  Target Milestone: ---

void
gemm_m10_n9_k17_ldA20_ldB20_ldC10_beta0_alignedA1_alignedC1_pfsigonly(const
double* __restrict__ A, const double* __restrict__ B, double* __restrict__ C,
const double* A_prefetch, const double* B_prefetch, const double* C_prefetch) {

  unsigned int l_m = 0;
  unsigned int l_n = 0;
  unsigned int l_k = 0;

  for ( l_n = 0; l_n < 9; l_n++ ) {
    for ( l_m = 0; l_m < 10; l_m++ ) { C[(l_n*10)+l_m] = 0.0; }

    for ( l_k = 0; l_k < 17; l_k++ ) {
      for ( l_m = 0; l_m < 10; l_m++ ) {
        C[(l_n*10)+l_m] += A[(l_k*20)+l_m] * B[(l_n*20)+l_k];
      }
    }
  }
}

is nicely vectorized with BB SLP when you make l_{m,n,k} signed but when
unsigned as above then split_constant_offset gives up when seeing

C + ((unsigned long)(_286 + 1) * 8)

but we even have nice range-info:

  # RANGE [0, 80] NONZERO 126
  _286 = l_n_189 * 10;
  # RANGE [0, 80] NONZERO 126
  _288 = (long unsigned int) _286;
  # RANGE [0, 640] NONZERO 1008
  _289 = _288 * 8;
  # PT = null { D.2428 } (nonlocal, restrict)
  _290 = C_37(D) + _289;

^^ C + ((unsigned long)(_286) * 8)

  # RANGE [1, 81] NONZERO 127
  _296 = _286 + 1;
  # RANGE [1, 81] NONZERO 127
  _297 = (long unsigned int) _296;
  # RANGE [8, 648] NONZERO 1016
  _298 = _297 * 8;
  # PT = { D.2428 } (nonlocal, restrict)
  _299 = C_37(D) + _298;

^^ C + ((unsigned long)(_286 + 1) * 8

giving up means DR group analysis doesn't relate them and we do not consider
SLP vectorization.

[Bug tree-optimization/98137] Could use SLP to vectorize if split_constant_offset were smarter

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98137

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |rguenth at gcc dot gnu.org
   Last reconfirmed|                            |2020-12-04
             Status|UNCONFIRMED                 |ASSIGNED
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Ah, so the issue is that

                /* Split the unconverted operand and try to prove that
                   wrapping isn't a problem.  */
                tree tmp_var, tmp_off;
                split_constant_offset (op0, &tmp_var, &tmp_off, cache, limit);

                /* See whether we have an SSA_NAME whose range is known
                   to be [A, B].  */
                if (TREE_CODE (tmp_var) != SSA_NAME)
                  return false;

we end up with tmp_var as MULT_EXPR (l_n_189 * 10).  We can fix that by
retaining the original code when splitting the constant offset returns zero.
While this makes us match up (_286 + 1) * 8 and (_286 + 2) * 8 it fails
to catch (_286 + 0) * 8 because that now is no longer expanded.

But we can use determine_value_range instead of get_value_range to also handle
expression trees.  That fixes the testcase.

[Bug tree-optimization/98137] Could use SLP to vectorize if split_constant_offset were smarter

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98137

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Created attachment 49677
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=49677&action=edit
patch

I am testing the attached simple patch.

[Bug tree-optimization/98137] Could use SLP to vectorize if split_constant_offset were smarter

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98137

--- Comment #3 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Richard Biener <rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:7b4ea2827d2003c8ffc76cd478f8974360cbd78f

commit r11-5809-g7b4ea2827d2003c8ffc76cd478f8974360cbd78f
Author: Richard Biener <rguenther@suse.de>
Date:   Fri Dec 4 11:13:48 2020 +0100

    tree-optimization/98137 - enhance split_constant_offset range handling

    split_constant_offset currently gives up looking at ranges when
    dealing with possibly wrapping operations for looking through
    conversions when the downstream analysis does not yield a SSA name.
    That's overly conservative and we have a nice helper that can
    deal with arbitrary expresssions.  Use that.  This helps data
    reference group analysis so the testcase is fully SLP vectorized,
    making use of the whole-function "BB" vectorization capabilities
    we now have.

    2020-12-04  Richard Biener  <rguenther@suse.de>

            PR tree-optimization/98137
            * tree-data-ref.c (split_constant_offset_1): Use
            determine_value_range instead of get_range_info to handle
            arbitrary expressions.

            * gcc.dg/vect/bb-slp-pr98137.c: New testcase.

[Bug tree-optimization/98137] Could use SLP to vectorize if split_constant_offset were smarter

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98137

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

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

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.

[Bug tree-optimization/98137] Could use SLP to vectorize if split_constant_offset were smarter

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98137

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

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

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
*** Bug 88767 has been marked as a duplicate of this bug. ***