Re: _Float16 handling changed in gcc 13?

[Jonathan Wakely via Gcc-help <gcc-help@xxxxxxxxxxx>]

On Thu, 17 Oct 2024 at 15:51, David Aldrich via Gcc-help <
gcc-help@xxxxxxxxxxx> wrote:

> Hi
>
> I have code that compiles correctly with gcc 12 and gcc 14, but not with
> gcc 13:
>
> SIMD.h: In static member function ‘static std::complex<_Float16>
> SIMD::reduce_add_pch(const __m256h&)’:
> /SIMD.h:893:44: error: no matching function for call to
> ‘std::complex<_Float16>::complex(float&, float&)’
>   893 |         return std::complex<_Float16>(re,im);
>       |                                            ^
>
> We are using C++20.
>
> I know that the behaviour of _Float16 changed in gcc 13 but I don't
> know how to fix this error.
>
> Any suggestions please?
>

There's no implicit conversion from float to the narrowing _Float16, but it
shoudl work with explicit casts:

return std::complex<_Float16>(static_cast<_Float16>(re),
static_cast<_Float16>(im));