The failing test is "LIT test 2 (degenerate case: 0 ^ 0 -> 1)". The z component of output vector should contain 1.0 value and gives us 0.0. The attached patch fixes it, but it looks like not all cards are affected. As it was discussed on IRC it is probably not a bug, but intentional change in HW to conform to DirectX which requires 0.0 result in this case. This bug was created to collect some test results from different GPUs to track down which GPUs conform to DirectX/OpenGL in this case. Currently checked cards OpenGL conformant: RV350 DirectX conformant: RV380 RV570
Created attachment 26003 [details] [review] LIT 0^0 fix
Just FYI: I'm thinking of adding a -pedantic flag to glean and to make this sub-test (and some others) depend on that flag. The 0^0 case is unlikely to come up very often in practice and it's hard to justify the cost of all the extra instructions for LIT just to satisfy this corner case.
(In reply to comment #2) > Just FYI: I'm thinking of adding a -pedantic flag to glean and to make this > sub-test (and some others) depend on that flag. > > The 0^0 case is unlikely to come up very often in practice and it's hard to > justify the cost of all the extra instructions for LIT just to satisfy this > corner case. > I agree, I was also wondering if this new path should be enabled by default or only in full conformance mode.
Mass version move, cvs -> git
The classic r300 driver has been abandoned long ago. It was replaced by the Gallium driver r300g. If you have issues with r300g please file a new bug report with component Drivers/Gallium/r300 Thanks.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.