Re: [PATCH v2 0/3] MR589: vkd3d-shader/dxil: Implement DX intrinsic SplitDouble.

[Giovanni Mascellani (＠giomasce)]

Giovanni Mascellani (@giomasce) commented about tests/hlsl/cast-64-bit.shader_test:
> uniform 4 int4 -20 8 0 0
>  draw quad
>  probe all rgba (-1.0, 2.0, 1073741824.0, 536870912.0)
> -
> -
>
> +[pixel shader todo]
> +uniform double2 d;
> -
>
> +float4 main() : sv_target
> +{
> - uint4 result;
> - asuint(d.x, result.x, result.y);
> - asuint(d.y, result.z, result.w);
> - return result;
>
> +}

I think this test should make sense for SM5, shouldn't it?

It would also be nice to return and compare results as `uint4`. Evan had written something for that in https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/416/diffs?commit_id=62..., but it hasn't been accepted yet. Maybe that patch could be cherry picked in your MR. If so, I would advise against using `%i`, see https://gitlab.winehq.org/wine/vkd3d/-/commit/9a27df3a8ca1c43a21668b44fd3c2a....