Am Freitag, 16. Mai 2008 20:57:04 schrieb Dan Kegel:
Anyone know how to get the d3d9 tests to pass? Are there a set of graphic cards that are known to work with them?
Nvidia gf6 and gf7 cards are confirmed. However, there's a rather strange bug in the instancing test. I do not see this failure, but Henri saw it, and it can be seen in some failures.
Right now, nobody who's reported test results under wine is seeing them pass:
http://test.winehq.org/data/2470b0b31605133ec046330dd79fdccaa7ba33fe/wine_xp... http://test.winehq.org/data/2470b0b31605133ec046330dd79fdccaa7ba33fe/wine_xp... http://test.winehq.org/data/2470b0b31605133ec046330dd79fdccaa7ba33fe/wine_xp... http://test.winehq.org/data/2470b0b31605133ec046330dd79fdccaa7ba33fe/wine_xp... http://test.winehq.org/data/2470b0b31605133ec046330dd79fdccaa7ba33fe/wine_xp... http://test.winehq.org/data/2470b0b31605133ec046330dd79fdccaa7ba33fe/wine_xp...
Those pass the test without failures. There are 27 TODOs, but well, they're TODOs. There is no way to get "0 skipped" on the visual test, because some tests are only executed on cards which do *not* have a specific feature, like the conditional NP2 test(skipped if unconditional non power of two support exists)
Let me check the others(from left to right, hoping you have the same order): -> I don't see why it writes "failed". Timeout maybe? -> That's the instancing failure -> success -> success -> success -> I guess the gf3 failures you reported. Can you resend your patch? I think we should get it in, unless Henri sends his -> ditto -> ditto -> ditto -> success -> gf3 -> success -> no idea :-( -> That one looks strange; It has the shademode failure, but the other failure looks odd -> I've no idea what's wrong here; It is an nvidia card, but things like fog and pixel shaders look very messed up. -> skip -> skip -> strange, no idea -> ATI driver. See remarks below -> broken GL setup? -> success -> skip -> No idea
Wrt the ATI driver: http://ati.cchtml.com/show_bug.cgi?id=1031 http://ati.cchtml.com/show_bug.cgi?id=1030 and others will probably kill the visual tests for a very long time. Matthew Tippet is happy to fix bugs which have a huge impact on gamers(once we get them isolated :-/), but he didn't sound as if he wants to invest any time in making our d3d9 test happy.