<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/115595>115595</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            Unable to vectorize loop with connectivity table on result index
        </td>
    </tr>

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

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

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

<pre>
    Hi,

I try to find if there is already an issue on the subject, but I did not find any.

I work on loop like this (https://godbolt.org/z/fchd4e54e), where the final index on the result array `y` is given by the integer table `face_cell`  : 

```c++
void compute_local_extrema(
  const int                nface, 
  const int ncell, 
  const int*    __restrict face_cell,
  const double* __restrict x, 
        double* __restrict y) {
  int il, ir;

#pragma clang loop distribute(enable) vectorize(enable) interleave(enable)
  for (int iface=0; iface<nface; ++iface) {
    il = face_cell[iface];
    ir = face_cell[iface+nface];
    y[il] = std::max(y[il], x[ir]);
  }
}
```

It seems that the auto-vectorize does not work on the kind of loop, despite the fact that it seems working on GCC (https://godbolt.org/z/944q8M6T1 from 12.2 to 13.3).

Is there a way to finally vectorize this kind of loop ?
If it is possible, can you give me some advices on how to do this?

Thanks a lot for your help.

Bertrand M.
</pre>
<img width="1px" height="1px" alt="" src="http://email.email.llvm.org/o/eJyMVU1z4ygQ_TX40hWXBJIjHXSI7fLuHOY2e04haFlsEHgB2db8-i2Q5diz2apRqZQAr7sf_fHMvVdHg9iQckvK_YqPobeuaXFQoke9aq2cmj8VoTuS7Un2Nn-_QXATBAudMhJUB6FHh6A8cO2Qywm4AeX9iGBNPAQ_tn-jCITuoB0DfAOpJBgbZg_cTOtn_xfrPqKttvYEWn0ghF55ILTqQzh5wt4IPRB6OFrZWh3W1h0JPfwk9NCJXhZYFkhoHcNdErXIoVOGa1BG4nWh5dCPOgB3jk9ANtlENlm8xlGd0UA7JZAyAY_oIPBWY0R1XOC7QK0jGgh7g0fyZJPNryB0G9-0e7ZKgrDDaQz4rq3g-h2vweHACa1mCICwxocYD355TAwZb_NfpElEvjoi9C3avr879MEpEeCT-FLPBS7t2GqMFg_w66Pb-fkSNxFaA3ndLtBISyVOyhG2fUoOZSfHjwMHobk5zvWVKjpqx4CEVmh4ilDDGUWwTv183o3VcBr5-Wl7Cd1ZF5skMUg5Y_uMsO2y2M2JZFuYazNvP7MHUBoI2z9kq7wBy_39Ognn_gdHt-Yr_BQBmpT7ZOaDjG3M3gZ-JbS6H8a8XePCpUX94IK87m9pvP-zNNvT-ATwiIOH0POQWpiPwb7c8wnSok_Tt4xZxHzESbRdKknkINGfVLiNDhdh9qYW39FUmWO0_mO3-53JrIvin-r75kcOnbMD5HRNo4bkbM0IrZ8FwN8khcOFL0rDtZ4-m2IWhEfSQNjhZt5FnsrDyXqvUofsQHADkx3TbMOA4O2AwOVZCfTxFr29xEDSJs93X_P3R8_NhwcOOmqWddGTgx716Yn3Fl1w3Ej4vl7Jhsma1XyFTf7K8rzI80216hvBMs7KVrKylLxg2at83RS42VQVVqLKypVqaEaLPM_qrGK0qNY1y-Kq6_KWFy0VpMhw4EqvtT4PMcGrpLVNnpdlXa40b1H7pOiUGrzMSkwojQLvmmj00o5HT4pMKx_8p5uggsbmrzRUMRefyU75vajQR70wKII6qzDdJNGaRUiTuK5Gp5tfmkGFfmzXwg6EHmK425-Xk7Pz78IhkfSEHm63ODf03wAAAP__WhQI0Q">