DXVK Reportedly Going Into "Maintenance Mode" Due To State Of Code-Base

[polarathene]

gaming in Wine + DXVK often works more reliably for me than native Windows gaming.

I've found it better than "native" Linux ports at times. Dying Light for example was running at 1FPS or something on my i5-6500 + GTX1070, switch to Proton and it runs waaaay better. The menu is possible to navigate and I could actually play the game.

[mphuZ]

Lol. That was the end of the tale.

So, Valve needs to hire him full-time so he can refactor it, and assign a sidekick to him he can delegate grunt-work to

No. This means that Valve made a mistake. It was necessary to conclude an agreement / truce on time with the developers of Wine, who assembled a whole team. And who just wouldn’t leave DXtoVK because of the routine work.

[polarathene]

perhaps Nvidia will have a reason to fix their driver. Even more specific, if the bug is only present on specific generation of nvidia cards, then it is more obvious where the problem is.

Does nvidia care that much for Linux gaming support? I thought their driver support was mostly intended for professional usage, where gaming was just a bonus?

[R41N3R]

DXVK is working extremely well, it is something that has not been achieved before and this in a short time frame. What is exhausting are probably the many different driver and linux versions, new introduced bugs with every new release, binary blobs, game workarounds and wine specifics. As a software developer you want to develop some stable code that is reliable in the long term, but this is quite impossible with the current platform, not sure if a CI system could be designed to test with every commit a broad game base. Development without the fear of break something fundamental is essential! I still wish wine would jump in to support D*VK as a stable foundation. WineD3D seems quite obsolete except for some corner cases.

Anyway, DXVK is a master piece and I'm very thankful for it!

[otikscypi]

The fact that he intend to spend some serious amount of time on refactoring is actually making me feel optimistic for the future of DXVK. Project was growing very fast, adding features in leaps, so eventual code-spaghetti was unavoidable. It's good that dev sees that and is ready to spend some time working on it.

What he needs is a nice CI to run a bunch of game tests automatically. One person testing all those games is not really sustainable.

I do not think that it would be a wise idea. Maintaining such CI suite would be extremely time consuming. Most games does not have API that let you automate movement or hooks to test if something happens. Otherwise we would have much more game benchmarks in PTS. I recall Michael complaining about this in at least one post.

[jrch2k8]

First of all, Thanks a million times to DXVK dev since this is by far hands down the best DX layer for wine(and Windows) and has made basically most games playable on Linux(note D9vk is a beast as well).

Basically i think Codeweavers and Valve should hire this guy full time, maybe he can fix the horrible mess wineD3D is, i don't means this as a disrespect to codeweavers developers but current wineD3D for any generation is a CPU hog barely using the GPU behemoth stacked on layers of bugs that on the best possible conditions qualifies as barely usable.

Now about DXVK regression i think his problem comes from try to support BLOB drivers with workarounds from AMD and nVidia, because both are randomly and heavily broken(depending the release) whereas RADV very rarely breaks and if it does it usually fixed in a blink of an eye.

If i was him i would probably clean of the codebase removing all workarounds for BLOB and only accept reports from RADV and ANV and use to Valve to put pressure on nVidia to fix their POS implementation and honestly simply give up support on AMDVLK(which i think he did) because honestly is a broken mess(not talking shit here, tested last week or something release and is broken on POLARIS again(put big freaking bold letters dripping blood AGAIN), Skyrim ENB simply destroy the driver to the point it hard reset my PC while RADV work through like a 60+ FPS boss while doing CAS with vkBasalt(<--- love this btw) )

[schmidtbag]

Well, this has kept my sanity in check. Not long ago, I was questioning some of DXVK's priorities, not because they were bad (any improvement is welcome) but rather there were more important things that needed more attention. People got angry at me, assuming I didn't like bug fixes or that there weren't more significant bugs to be fixed. Well, turns out, the lead developer of DXVK seems to agree with my sense of priority.


Anyway, I actually have had very smooth experience with DXVK and am grateful for its existence. The only recurring issue I encounter is an occasional shadow glitch, but it's nothing game-breaking.

[loganj]

Mystro256 that company might as well be valve. they can work with devs to make some automations for their games to test dxvk/d9vk. it doesn't even have to be the entire game. just some demos

[marlock]

The way I read this news is: he braved the waters swiftly by being a bit chaotic and now will be part of the team doing the better-planned rework under another name... its good news in my book, except that likely there will be a perceived slower pace in improvements for a while.

[artivision]

What are you talking about people, the bugs are introduced with new future levels of support and not with drivers. You have to equally share the development time between introduce new futures and debug the new problems. DXVK reached almost future parity with D3D11 in one year, this speed is unnatural. Time for debugging and then continue normally.