[PATCH 4/5] d3d/tests: Table fog does not abs the fog coordinate (v2).

Thu Nov 27 03:33:44 CST 2014

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am 2014-11-27 10:27, schrieb Henri Verbeet:
> On 26 November 2014 at 19:08, Stefan Dösinger <stefan at codeweavers.com> wrote:
>> +    /* Needed to make AMD drivers happy. Yeah, it is not supposed to
>> +     * have an effect on RHW draws. */
>> +    static const D3DMATRIX identity =
>> +    {{{
>> +        1.0f, 0.0f, 0.0f, 0.0f,
>> +        0.0f, 1.0f, 0.0f, 0.0f,
>> +        0.0f, 0.0f, 1.0f, 0.0f,
>> +        0.0f, 0.0f, 0.0f, 1.0f
>> +    }}};
> Is this the same projection matrix issue as in
> fog_test()/fog_with_shader_tests()/etc.?
I can't look into the AMD drivers, so I don't know for sure, but I think this is different. fog_test, fog_interpolation_test want a redundant application of the projection matrix. In this case there's a SetTransform(D3DTS_PROJECTION) already, but the z = 0.0 part breaks AMD drivers in RHW draws.

Why are you asking? Do you have some idea what is causing this behavior?