[Bug tree-optimization/124571] New: `ABS<(float)(us1 - us2)> < 1.0f` can be optimized to us1 == us2

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

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124571

            Bug ID: 124571
           Summary: `ABS<(float)(us1 - us2)> < 1.0f` can  be optimized to
                    us1 == us2
           Product: gcc
           Version: 16.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: enhancement
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: pinskia at gcc dot gnu.org
            Blocks: 26163
  Target Milestone: ---

```
int f(unsigned short a, unsigned short b)
{
  float af = a;
  float bf = b;
  float s = __builtin_fabsf(a - b);
  return s < 1.0f;
}
```

Can be turned into just:
```
int f(unsigned short a, unsigned short b)
{
  return a == b;
}
```

Note there is 2 missed optimizations here (for GCC).
One is pushing the abs back over the conversion from float to int.
The second is converting `(float)us1 < 1` into `us1 < 1`.

This shows up in spec 2017 in imagick_r.


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=26163
[Bug 26163] [meta-bug] missed optimization in SPEC (2k17, 2k and 2k6 and 95)