<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/129206>129206</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            [MLIR] [AMDGPU] refactor how we handle target features
        </td>
    </tr>

    <tr>
      <th>Labels</th>
      <td>
            mlir
      </td>
    </tr>

    <tr>
      <th>Assignees</th>
      <td>
      </td>
    </tr>

    <tr>
      <th>Reporter</th>
      <td>
          dhernandez0
      </td>
    </tr>
</table>

<pre>
    Currently, we hardcode hardware features in the code (see https://github.com/llvm/llvm-project/blob/15c49b9db3f60bdbd320271d5e97f118c00b95dd/mlir/lib/Dialect/AMDGPU/Transforms/EmulateAtomics.cpp#L173). We want to implement a better way of handling subtarget features. See this PR for discussion about possible implementations: https://github.com/llvm/llvm-project/pull/129029
</pre>
<img width="1" height="1" alt="" src="http://email.email.llvm.org/o/eJyUkk9r3DwQxj-NfBleI4-t9ergQ5J9XQoJhLSlZ8kar1VkyehPlvTTF--2DfSWkwRCzzz85qdSsmdPNDBxz8SpUiUvIQ5moeiVN_STVzqYt-GhxEg-uzeGD3AhWFQ0UzC3y0VFgplULpESWA95Ibi-MjwmIlhy3hJr7xiODMezzUvR9RRWhqNzr3-O_7YYftCUGY7aBc1wbMTUSS2NbucD10abFjn2jREk-7lpjhPnWgpjGI6rs3HPsfu_k1XuFnT3dPr0_I3h-DUqn-YQ18Rw_H8tTmW6y2G1U6qnbWPYPjZ9y1DW8J3gonyGHMCum6OVfAYFmnKmCBf1BmGGRXnjrD9DKjqreKb8l0ANX4ggLzbB8wvMIYKxaSop2eBB6VAybCElqx29D1DZBr8z-jisrTi3w0LJUVZmaI1spapoaPqO923TCqyWQQhsJt3K1ui-l9wII4Sautl0muv5QJUdkKPgiEd-RNlgrQ-6I6X4YeaTPIiGdZxWZV29z69DPFc2pUJDgxL5oXJKk0tXkRBv68BdqThc--pyTqzjzqac3hOyze4q39Pj5xcmTsDE_e-diRNEmtWUQ4QlXG7WeeMI_uFdleiGD1O7dk83bnv91wF_BQAA___Su_8G">