| Re: [AD] [test scaled rotate 5] failure |
[ Thread Index |
Date Index
| More lists.liballeg.org/allegro-developers Archives
]
On Wed, 2010-10-06 at 22:49 +0900, John Murphy wrote:
> Correct: r13797; [sw] test succeeded:
>
>
> >test_driver -s C:\allegro\allegro.git\tests\test_bitmaps.ini "test
> scaled rotate 5"
> OK test scaled rotate 5 [sw] - by signature
> FAIL test scaled rotate 5 [hw] - RMS error is 41.2741
> hash_hw=a85e5275
> sig_hw=KWnQdRKMNKTkQYVHIKKQhnYCKINKKWnObRKMKKTjQXXKJKKQhnYKJJKKKWnXYRIKKKTjQYYJKKKQgnYSO
>
>
> total tests: 2
> passed tests: 1
> failed tests: 1
>
>
> http://dl.dropbox.com/u/508947/test_scaled_rotate_5_hw.png
> http://dl.dropbox.com/u/508947/test_scaled_rotate_5_sw.png
>
>
I guess Intel is using float instead of double in their polygon
rasterizer :X If that's true, not much we can do and we may as well have
the test fail on Intel cards. I think Pavel said something similar
happens on ATI cards.
--
Elias Pschernig <elias.pschernig@xxxxxxxxxx>