[Bug tree-optimization/100173] telecom/viterb00data_1 has 16.92% regression compared O2 -ftree-vectorize -fvect-cost-model=very-cheap to O2 on CLX/ICX, 9% regression on znver3

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100173

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2026-07-23
     Ever confirmed|0                           |1
           Assignee|unassigned at gcc dot gnu.org      |pinskia at gcc dot 
gnu.org

--- Comment #6 from Drea Pinski <pinskia at gcc dot gnu.org> ---
In phiopt4 at `-O2  -march=znver3` we have:
````
  if (esMetric1_90 >= esMetric2_94)
    goto <bb 7>; [50.00%]
  else
    goto <bb 8>; [50.00%]

  <bb 7> [local count: 505302904]:
  # VUSE <.MEM_86>
  _96 = MEM[(short int *)pIn2_45 + -64B];
  goto <bb 9>; [100.00%]

  <bb 8> [local count: 505302904]:
  # VUSE <.MEM_86>
  _95 = MEM[(short int *)pIn2_45];

  <bb 9> [local count: 1010605808]:
  # _97 = PHI <_95(8), _96(7)>
  # cstore_98 = PHI <esMetric2_94(8), esMetric1_90(7)>
```

This should have factored out the load there too.

Let me take a look.