Z fighting in test suite

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Z fighting in test suite

nophead
opencsgtest_issue1215 and throwntogethertest_issue1215 fail for me because the expected results have Z fighting and my machine generates a different Z fighting pattern.

opencsgtest_issue1165 has a very thin feature that causes Z fighting on my machine.

How are these tests expected to work? Is Z fighting expected to be deterministic?




_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org
Reply | Threaded
Open this post in threaded view
|

Re: Z fighting in test suite

kintel
Administrator

> On Apr 27, 2017, at 05:12, nop head <[hidden email]> wrote:
>
> opencsgtest_issue1215 and throwntogethertest_issue1215 fail for me because the expected results have Z fighting and my machine generates a different Z fighting pattern.
>
> opencsgtest_issue1165 has a very thin feature that causes Z fighting on my machine.
>
> How are these tests expected to work? Is Z fighting expected to be deterministic?
>
Z-fighting should be deterministic across GPUs from the same vendor where Z buffer size is the same.
Across GPUs from different vendors it may change.
..and it definitely changes across drivers when software-rendering is involved

We disable a bunch of tests, including the two you mention, when running tests on the CI, see https://github.com/openscad/openscad/blob/master/scripts/travis-ci.sh
We may want to tag these tests using a named group in ctest to make enabling/disabling them more portable.

 -Marius


_______________________________________________
OpenSCAD mailing list
[hidden email]
http://lists.openscad.org/mailman/listinfo/discuss_lists.openscad.org