<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/68771>68771</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            CUDA code compile error: use of overloaded operator '=' is ambiguous
        </td>
    </tr>

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

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

    <tr>
      <th>Reporter</th>
      <td>
          zhiweij1
      </td>
    </tr>
</table>

<pre>
    ```
#include "cuda_fp16.h"

inline __device__ float4 hmma_fp32(__half2* array, __half a) {
    array[123] = {a, a};
}
```
https://godbolt.org/z/9Po5KosvK

This code can be complied with nvcc: https://godbolt.org/z/zTTos1zKv
</pre>
<img width="1px" height="1px" alt="" src="http://email.email.llvm.org/o/eJyEUU2PozAM_TXmYk0FTijlwKEziMtc9jB7RiExJaNAqiR0NP31K1qk_bisFCVW_N6zn61itJeFuYHyFco2U2uafGjuk_1i-1lkgzffDRzz_eQt5GcgYRftVsMIRHo1qh-vxfEwAdGOeNx2cXZh7HvDN6u573F0XiWJ0zxvFEFAp76flBsJ6IwqBPUN9IbPP1RANUL1-lRDxB1RvhYkoGwRRLvl1cZRULUgduwWP4O_O59SukYQZ6AOqLt4M3iXDj5cgLo7UFf_8OW7j7f3P218TDai9oZRqwUHRu3nq7Ns8MumCZeb1iDO-D_t-8eHj8X9_ZaZRpha1CrjpjjWVZ7XhayzqTkJdRq1lCxFfSqPupJSGso1DcKQGiizDeUkirwo8opKWR9GOZZaMstKDKaQFcicZ2XdwbnbvNXObIwrN8dTVRWZUwO7-Ng00cJf-EhuSyvbLDQb52VYLxFk7mxM8bdKsslx8_azPe-D8PPVOkYOwYfN_BoZ_Yj-xsF5Zdigv3JQyQcEqkC0QBXaiGoe7GX1a8zW4Jp_RmbTtA4H7Wegbiu9Py_X4D9ZJ6Du0XAE6h6GfgUAAP__8F7Tgw">