<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/136838>136838</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            Auto-vectorizer generates different code for the main loop and for the trailing section
        </td>
    </tr>

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

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

    <tr>
      <th>Reporter</th>
      <td>
          dmenendez-gruposantander
      </td>
    </tr>
</table>

<pre>
    Repro and code available in godbolt: https://godbolt.org/z/c936s31hn

Function `g()` in the example is expected to be auto-vectorized, and produce the same results as function f`()`, which is not vectorized. This is true for the code generated for the "main" part of the loop: I can see it in the generated assembly and in the test results.

However, the code generated for the **trailing** part of the loop does not match the code generated for the "main" part of the loop: the trailing section fails to fuse a multiplication and an addition.

The example program shows this effect:
 - it compares output of `f()` and `g()` for 1 element
 - it compares the output of `g()` for 5 equal input elements, showing that the 5-th result (coming out of the trailing part of the loop) is different than the rest.

For reference, the godbolt link compares against gcc, which generates the expected result.

```
$ clang --version
Ubuntu clang version 19.1.7 (++20250114103320+cd708029e0b2-1~exp1~20250114103432.75)
Target: x86_64-pc-linux-gnu
Thread model: posix
InstalledDir: /usr/lib/llvm-19/bin
```

Source of installation: https://apt.llvm.org/jammy/dists/llvm-toolchain-jammy-19/main/
</pre>
<img width="1" height="1" alt="" src="http://email.email.llvm.org/o/eJykVUuPqzgT_TXOpkRkDHmwYEF_V-i725k765GxK-A7xmb8SKd7cX_7yIZ0-iHNZqQoEi5Tdc6pUwX3Xo0GsSWHJ3L4tuMxTNa1ckaDRuJrMbq4WM9N4Eai2w1WvrS_4eIscCNBWInAr1xpPmgEZWC0crA6kKqDKYTFk6ojrCes3wJ760bC-lfCetFUR1-VkyG0I7TroxFBWQPkSEfCzoQ15EhTzjAh4I3PSyrhAW8LioASgoUBgcdgiyuKYJ16RUnY_zK0xVkZBeaXPZ8RHPqogwfu4XIvdSFH-lYqvfk8KTGlIsYGeCTdw49J-XQeXES4WJfzZvojGnQ84bkfE8ZmrgxhDBbuAthLPtbWLkmX7yC4AY8IKtzpPZJw73Ee9EsmsUUD-nDHv1_V-r99xiu6hPlfkXSEdcFxpZUZ16cvmEBaXAnPPIjpPzDLWLdi4HHTmCvtU68u0SNwmKMOatFK8BxONLkBLqVKzxu9H-9avjg7Oj6Dn-yzh5AagZcLimQyQjsoko7Czgt36MHGsMSMjBzp5eGjVOejsxKpElDjjCZ8TZTIfEj26dUD4N-Ra1AmXdnS-NSQBDQpECYecppDEaatf0DYWdg5hW18E_BNtC-qsia5TqrLBR2alI6vnnDowyZWbx04zBcE3h2xzRtoZf56kOIjV8YHGIV42P3eZ79N2jZeK-CtRpqP9Uc7wmoQmpsRiuKKziubRviPIZoQt8B2DGWzL_cnyMo9EfbEKDvQsqxLWlWMEvYk5ImeKWuQDqwof-FtKX-9u1RXbH86JNVp94O7EfNmuZ2Pfx7rYhGFVibeitHE7BmHXMJsJep0a7Fe3QjtvhsfuNYovymXzgnro3eE9VoN6V9f56JsCOsHZT4zpd3vNjqBqSdqzZNt-3W_8SXsU6ptwf3k8_xCWC-VT6ZYqwRrtZi4MkUOr1XXgep3sq1kUzV8h215qg81ZafmvJva5lLWlJ5OR3qoztVhaMoam3Mtqkry4Yz1TrVZsJpVJaXH6rynrJF4Kht6PJwYbQ6kpjhzpd_w7ZT3EduyOp6r807zAbXPXwDGDD5DjhLG0gfBtRn5EEdPaqoTm0eaoILGtvuwgN07Oz1sm_fJfYskxuvmSUN5P_28OHbR6fbTN0SFKQ57YedN0buwi7M_00JgfcaeBN_IXVv2TwAAAP__x71DCQ">