<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/140219>140219</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            [AMDGPU] Use reverse iteration in CGP
        </td>
    </tr>

    <tr>
      <th>Labels</th>
      <td>
            new issue
      </td>
    </tr>

    <tr>
      <th>Assignees</th>
      <td>
            Pierre-vh
      </td>
    </tr>

    <tr>
      <th>Reporter</th>
      <td>
          Pierre-vh
      </td>
    </tr>
</table>

<pre>
    #64591 has been fixed, so once the dust settles a bit (at least until the change propagates to downstream branches and we don't have any major issues) we can move forward with:

- Removing all of the i16 promotion logic from CGP, deleting the option and removing/reworking associated tests
- Use a reverse iteration order in the pass so it works like a normal combiner.
</pre>
<img width="1" height="1" alt="" src="http://email.email.llvm.org/o/eJxkkkFvnDAQhX-NuYx2BQZ24cAhTbQ5VYoq5QcMMAtOjGflGdjm31em2_ZQyZIl683n5-eHIm4KRJ2pvxlr3xzFSIdtNtaa-iXDVWeO3d_jrOfxqzO2PFV1W8CMAj1RgKv7SaOxzyAMHAYCnQnGVRSEVD0JIPROwdgGFTyhKKxBnd-Fw4xhIrhFvuGESgLKMPI9iEbCBfqIYZgTJIxwJxg5GHtWmHEjwPAFC35wBCeykhjbJs2AARbeCK4c7xhHuDudTflk8rQO8IMW3lyYAL0Hvu4-XHFKJhZWxwE8T26Aa-QFnl_f0uNG8qRpJon5tquSpfhgGXuJdOf4uXNFeHCoNIKSqOy3vgsBQqSNohA4pYg7heNIEVzYyTcUSUE6hQQT8O4zjQWOC3oYeOldoHjMxq4c27LFjLriXJ3LpizaNpu79lzVzbXPm6Zq-jovhmrEvr2eq7a0tixOmetsbuu8Lk55U5RFe8RTiXnb5FjlZ1vWvalyWtD5o_fbcuQ4ZXu4XVHltmgzjz15eXQm0P139I_OxC4NHfp1ElPl3onKP4w69XvZnr6_vL69m_plz-T_RFxIqWdr9N2sepP0dfZi7GVyOq_9ceDF2EviPrbDLfIHDWrs5U8RLg-7W2d_BQAA__-km_ay">