(06-26-2014, 12:59 PM)Tino Wrote: [ -> ]amozzz: I think his doing an excelent job, optimizations an improved resource management are alwas usefull, and even fixing old bugs in the way is great. he only need more time to refine his code and learn more abut dolphin core functionality.
Agreed.
I look forward to a day when these are properly in master. Maybe you could offer advice to each other and be the ultimate optimization team.
After all you guys are making great progress
(06-26-2014, 12:59 PM)Tino Wrote: [ -> ]amozzz: I think his doing an excelent job, optimizations an improved resource management are always usefull, and even fixing old bugs in the way is great. he only need more time to refine his code and learn more about dolphin core functionality.
our job is complementary , i already applied his first push request to my fork, to allow more people to test his changes. the only way to really test changes is to give everyone access to a working copy.
But in your recent builds his optimizations are not applied to DX9 backend, right?
(06-26-2014, 11:01 AM)Tino Wrote: [ -> ]is generated by the latest version or it was there already? a specific place or everywhere?
It was there already. Happens everywhere.
Look at the grass in the background:
![[Image: FIFnuOS.gif]](http://i.imgur.com/FIFnuOS.gif)
Look at the carpet over the stairs:
![[Image: yTjzu00.gif]](http://i.imgur.com/yTjzu00.gif)
I think many games have this. I see it in Xenoblade Chronicles, too. Actually I don't even know anymore if that didn't happen on the real console...
(06-26-2014, 08:33 PM)StripTheSoul Wrote: [ -> ]I think many games have this. I see it in Xenoblade Chronicles, too. Actually I don't even know anymore if that didn't happen on the real console...
It doesn't happend to me with master revision.
that is called z-fighting it happends when rounding errors in depth causes that poligons that are to close jump betwen frames. will take a look.
StripTheSoul : galop1n's optimization that where make public in the PR are merged but they only affect dx11.
just and update. i'm figthing against a nasty bug in nvidia dx9 shader compiler, it seems that dx9 is really looking bad on nvidia, and i can't find why dx11 compiler works fine but dx9 not.
Amd and microsoft reference rasterizer are free from this bug. will update when i have more news.
I tried using the non-beta NVIDIA driver and had the same problem, just to let you know.
already know, tested diferent version and the error seems to be there from ages, is only present on the dx9 compiler, the same shader on dx11 is matematically correct. it seems to be some kind of optimization but is matematically inacurate. intel, amd and the reference rasterizer works.
Updated with a new version. hopefully issues with nvidia cards should be fixed.