On 5 June 2016 at 23:56, Stefan Dösinger stefandoesinger@gmx.at wrote:
They're more broken on Nvidia + Win10 than they used to be on Nvidia + Win7. I can't find a way to detect the problem without putting the driver into a broken state.
The annoying part of this is that we lose 4 bit precision tests. X4R4G4B4 doesn't seem to exist everywhere.
I don't suppose there's any chance of getting the driver fixed. Would something along the lines of adapter_is_warp() work here perhaps? With recent Windows versions, I'm getting the impression that there may come a point where we'll have a maximum version to run the ddraw tests on.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
Am 2016-06-07 um 16:18 schrieb Henri Verbeet:
I don't suppose there's any chance of getting the driver fixed. Would something along the lines of adapter_is_warp() work here perhaps? With recent Windows versions, I'm getting the impression that there may come a point where we'll have a maximum version to run the ddraw tests on.
WARP isn't so much a problem here - it only returns junk data. The Nvidia driver is the bigger problem here because the test gets stuck forever.
Are you suggesting to try to get Nvidia to fix their drivers? I'd rather not spend time on this. Detecting the nvidia driver in advance and skipping this test based on that similar to adapter_is_warp is something I'd be open to though.
On 7 June 2016 at 21:29, Stefan Dösinger stefandoesinger@gmail.com wrote:
Are you suggesting to try to get Nvidia to fix their drivers? I'd
Ideally, but it's one of those low-probability things.
rather not spend time on this. Detecting the nvidia driver in advance and skipping this test based on that similar to adapter_is_warp is something I'd be open to though.
Yeah, I think that's better than disabling the test everywhere, if only marginally. You'll probably want to take "winetest_platform" into account in addition to the driver name.
-
Henri Verbeet -
Stefan Dösinger