Created attachment 128941 [details] screenshot Very similar (but visually not exactly identical) glitches exists without nine, but there are workaround in wine: [HKEY_CURRENT_USER\Software\Wine\Direct3D] CheckFloatConstants="enabled" which helps with it. This key does not make a difference with nine. There are no such problem with nouveau (750ti) here, os ubuntu-16.04 with oibaf and commendsarnex ppa.
Experienced the same problem on stock mesa from https://launchpad.net/~inferrna/+archive/ubuntu/featured-mesa/ (nine enabled) and wine from commendsarnex ppa.
Could you try the patch from this bugreport : https://bugs.freedesktop.org/show_bug.cgi?id=94459 If the bug is still present, please make an d3d9 apitrace of the game that shows the bug. The smaller the file the better. Upload it on the Ixit Nine ftp or some filesharing service (like google drive).
I am not sure about compiling 32 bit mesa on 64 bit system, but will try.
patch from 94459 does not help, d3d9 apitrace (witcher.trace.7z - 170M): https://drive.google.com/file/d/11yogHC1q45emz3hOQOkJ70wW_xzhjOP8/view?usp=sharing
Tested on llvmpipe and nvc0 (card NVD9): The bug isn't visible. Tested on radeonsi (card AMD TAHITI): The bug is visible. csmt_force makes no difference. Mark as AMD bug. Tested with mesa 17.3.0.
Glitches disappeared for me after graphics card upgrade to radeon 560. It very possible means that amdgpu driver is not affected by this problem. http://storage6.static.itmages.com/i/18/0402/h_1522696416_2964411_ae68ba0d5e.png
It's not driver specific bug. I should have followed the hint that there is a registry entry to workaround this problem with wined3d. It turns out that this registry entry is specifically for this bug. The analysis and possible solutions could be found here: https://bugs.winehq.org/show_bug.cgi?id=34052#c42 Also, this bug is also tracked at: https://github.com/iXit/Mesa-3D/issues/312
This patch seems to fix rendering on Evergreen: https://bugs.freedesktop.org/page.cgi?id=splinter.html&bug=91808&attachment=139438 Now, it remains to find out why it is reproducible on Tahiti.
The fix for r600/Evergreen has been committed: https://cgit.freedesktop.org/mesa/mesa/commit/?id=ce027ac5c798b39582288e5d7d9973b3cdda591e The Tahiti issue remains under investigation.
so it may be closed? p.s. thanks a lot to all participated
(In reply to Roman Elshin from comment #10) > so it may be closed? > p.s. thanks a lot to all participated If it works OK for you, you are free to close it. You may also mark it as duplicate of https://bugs.freedesktop.org/show_bug.cgi?id=91808 Since both issues seems to be fixed by same change. About the problem still existing on Tahiti, the issue might be some other (similar) bug. I've been told that you can't just port that fix to radeonsi/gcn as they work differently. But it is not problem to open a new issue for it.
*** This bug has been marked as a duplicate of bug 91808 ***
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.