[Bug 103616] Increased difference from reference image in shaders

bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org
Wed Nov 8 08:00:53 UTC 2017


https://bugs.freedesktop.org/show_bug.cgi?id=103616

            Bug ID: 103616
           Summary: Increased difference from reference image in shaders
           Product: Mesa
           Version: 17.2
          Hardware: Other
                OS: All
            Status: NEW
          Severity: normal
          Priority: medium
         Component: Drivers/DRI/i965
          Assignee: intel-3d-bugs at lists.freedesktop.org
          Reporter: b7.10110111 at gmail.com
        QA Contact: intel-3d-bugs at lists.freedesktop.org

After commit 4009a9ead490ef1718e6fa83141aa086a43cd901 (found by git-bisect,
double-checked by git-revert) I get two failures of the integration_test in
precomputed_atmospheric_scattering demo [1]. The tests there compare reference
CPU-generated image with the one generated with GLSL code, checking that PSNR
is high enough. Before the above commit they all passed. Now the tests failing
are:

47 is not less than 46.7812
[ FAIL ] ModelTest RadianceSeparateTextures
46 is not less than 45.6379
[ FAIL ] ModelTest RadianceCombineTextures

I tested on Xeon E3-1226 v3 built-in GPU, as well as on Core i7-4765T built-in
GPU, with the exact same results.

I'm not sure if this is an expected decrease in precision, but it seems strange
to me: from the commit message I get that the code transformation introduced
should be equivalent. Apparently it's not, or there may be a bug somewhere.

[1]: https://github.com/ebruneton/precomputed_atmospheric_scattering

-- 
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20171108/8267ac26/attachment.html>


More information about the intel-3d-bugs mailing list