I tried to look for a more recent pre-compiled build from the master branch for testing, but the site I used to use only goes as far as 17th Jan

We offer builds of every commit here: https://git.froggi.es/doitsujin/dxvk/-/jobs

Thanks for reporting, I'll take a look and see if I can repro it at all.

@Joshua-Ashton

I tested the builds after 1.5.1 and the issue became present as of commit - https://git.froggi.es/doitsujin/dxvk/commit/caae5d18edd3ae03bfcbdb19e1d9eb849e82e459

im having the same issue but im on windows with AMD GPU.
Ill take a look at the builds

@Joshua-Ashton

#1091 d3d9-view-fix 73910895

this is what broke it

So I tried to repro the issue on GRID on a machine with Intel Iris Plus Graphics 655 with mesa-19.3.4-2 + linux-5.5.3-arch1-1 and I did not have any issues running it on 1.5.4 (the version shipped in Proton) or on master.

I also tried to repro on Windows using AMD but it also worked fine for me there.

I'm not really sure where to go from here given I can't reproduce the issue at all.

Can you try to update your mesa? It's the only thing I can really think of.

As for what you bisected it down to -- I don't see how that can be the cause, if that was broken it would be broken everywhere.

Only place im getting the issue is on an emulator CXBX-R. so i didnt report it plus im on windows and your focus isnt on windows

I develop the d3d9 frontend on Windows 🐸

Are you using the latest AMD Drivers by any chance? Adrenalin 2020 Edition 20.2.1 Optional.

Yes

There's a chance the CXBX-R issue is separate... Can you provide an apitrace of that?

So I tried to repro the issue on GRID on a machine with Intel Iris Plus Graphics 655 with mesa-19.3.4-2 + linux-5.5.3-arch1-1 and I did not have any issues running it on 1.5.4 (the version shipped in Proton) or on master.

I also tried to repro on Windows using AMD but it also worked fine for me there.

I'm not really sure where to go from here given I can't reproduce the issue at all.

I'll try updating the Kernel and Mesa tomorrow and re-test. Will keep the ticket open for now.

Thanks for checking it out :)

There's a chance the CXBX-R issue is separate... Can you provide an apitrace of that?

not sure how to do that any instructions to do so?

Follow https://github.com/Joshua-Ashton/d9vk/wiki/Making-a-Trace

cant really do that since there will be 2 d3d9.dlls in the folder

trace it without dxvk then

doesnt repro without it

does it repro if you trace it on native d3d9 then play back the trace with dxvk?

that's kinda what i am asking

uh how do you do that? heres a trace that came from just dropping the file in
cxbx.zip

Well first you'd trace it like the guide, then put dxvk's d3d9.dll next to d3dretrace.exe and play back the thing you recorded

yep i can repro it that way

Can you send me the trace you recorded?

its in my previous comment

Hm, I am not able to reproduce the flickering on my end by replaying the trace

must be a Polaris issue then. the RX 480 is old by now and probably lacking support for somthing.
I would report it to AMD but this seems like a isolated issue and may not get attention.

I Have the same problem

My Specs
CPU:Intel Core i3 5005U
GPU:Intel HD5500
Driver:Mesa 19.3.4
Wine version:5.0
DXVK version:1.5.3,1.5.4
Kernel version:5.4.18

Game:
Wasteland 2

D3D9 log:
DXVK 1.5.3
https://drive.google.com/open?id=15QyY6sDuZN1K9KFc4P99_h3qJEikyGv0
DXVK 1.5.4
https://drive.google.com/open?id=1I8do3Rr65ygUNyXXfb8OuKB_towdxe-i

I ran further tests regarding the flicker:

Mesa 19.3.4 + LLVM 9.0.1 + Kernel 5.5.4 + DXVK 1.5.4
Mesa 20.0.0 + LLVM 9.0.1 + Kernel 5.5.4 + DXVK 1.5.4

Issue still persists. At this point I thought I was doing something wrong with my setup, but DXVK with DX11 works perfectly and even OpenGL mode with the Iris driver.

I'm beginning to think this may be a specific issue related to Intel Gen8 hardware in particular (which would not be the first time). This coming weekend I shall clone my OS over to an i3-7100 system and re-test, hopefully it will be enough to reveal the source of the issue.

Will keep you posted.

As mentioned a couple of days ago, I've cloned my test environment to a system with Kaby Lake GPU (Intel G4560 and i3-7100) and the flickering didn't occur.

This means the issue is most likely due to an Intel ANV bug. Going to file a ticket in the right place.

the latest nightly has fixed the issue on my end