Vulkan Turns Two Years Old, What Do You Hope For Next?

[phoronix]

Phoronix: Vulkan Turns Two Years Old, What Do You Hope For Next?

This last week marked two years since the debut of Vulkan 1.0, you can see our our original launch article. My overworked memory missed realizing it by a few days, but it's been a pretty miraculous two years for this high-performance graphics and compute API...

http://www.phoronix.com/scan.php?page=news_item&px=Vulkan-Turned-Two-Years-Old

[kaprikawn]

It's been an excellent start and it is absolutely on track to realise the promise it had when released. The thing with graphics APIs is that they take a long time to gain traction due to various factors including userbase volume and the time it takes for devs to retool. Vulkan had an usual headstart with the former as most graphics cards supported it when it released.

And we're starting to see a tipping point with regards to devs retooling. With Feral in mind, we're past them dipping their toes in the water and well on our way to them using it exclusively I think. What I'd like to see is either a new game engine developed built with Vulkan in mind from the ground up, or one of the major vendors like Unreal really optimize their engine for Vulkan.

Yes, it'd be nice if Vulkan blew OpenGL out of the water, but we've got to reign in our expectations. It will take more time yet for Vulkan to fully realise its potential, but I have no doubt it will come. I think things will get better as the percentage of mobile device that support Vulkan increases. Unfortunately on mobile it takes a more recent version of Android to enable it, But as it becomes a realistic technology for mobile, Unity, Unreal and the like will start to optimize for Vulkan which benefit those who use those engines for Linux.

[OneTimeShot]

I'd like all GPU features - compute, audio, video, 2d and 3d - to be harmonised into a single "platform version" grouping, rather than having to track versions of OpenCL/AL/GL/MAX/Vulkan/SVG APIs separately.

Also, I prefer API stability over "new features every three weeks".

[Guest]

I'd like to see Wine have full, or at least DX9+ being able to be translated to Vulkan, in the hopes of it improving performance over the OpenGL translation that's done currently.

[humbug]

The Vulkan API itself seems to be great. Also looks like things work much better and cleaner after years of complicated openGL where different drivers did unpredictable things.

But we have a problem in the Linux gaming ecosystem where it seems that most games perform 20-30% slower than windows. I am not technically competent to explain why, what I hear is that the porting methodologies currently being used introduce new bottlenecks that the native windows code doesn't have.

Feral is releasing Rise of the Tomb Raider on Linux soon using Vulkan. I certainly hope they give us something that performs as good or better than the windows directX11 and directX12 versions on both AMD and Nvidia. On windows Doom Vulkan became a shining example of a great Vulkan implementation with great performance considering the graphical quality. We need more of this on Linux, and for that to happen the game has to be well optimized from top to bottom, not just the vulkan renderer.

Apart from that I hope that the big engines will roll out more mature, better performing vulkan renderers. Unreal, Unity, Crytek etc.

[humbug]

On the Windows side there have been a few prominent Vulkan titles like DOOM and Wolfenstein. But sadly not as many Vulkan games as I would have hoped for now two years in

Thing is the current development cycle for AAA games is so long... The games being released in the last couple years were well into heavy development before vulkan was even conceived. So I guess most devs don't like to switch graphics APIs in the middle of their development cycle.

[oooverclocker]

In fact, on GitHub is now around 1,800 projects referencing Vulkan while just 94 mention D3D12 or 38 as "Direct3D 12."

I wasn't aware of this. It sounds like a massive defeat for Direct3D 12. The Wikipedia page also shows that Vulkan wins because almost only the few Microsoft funded games use DirectX12.

But although DX12 is defeated, unfortunately Vulkan is not the biggest winner. The most companies still use OpenGL and older Direct3D versions.
Vulkan is still the most valuable opportunity for Linux to arise. And it's a statement of freedom.
So there must be more advertising for it and I am very happy for this article. Although Microsoft is insanely strong with all the impact they have through their operating system we can win when each software that runs on Windows does also run on Linux with at least about the same stability and performance.

[Tomin]

I hope people will stop calling it Vulcan.

[eydee]

Feral is releasing Rise of the Tomb Raider on Linux soon using Vulkan. I certainly hope they give us something that performs as good or better than the windows directX11 and directX12 versions on both AMD and Nvidia. On windows Doom Vulkan became a shining example of a great Vulkan implementation with great performance considering the graphical quality. We need more of this on Linux, and for that to happen the game has to be well optimized from top to bottom, not just the vulkan renderer.

There's Talos and SS Fusion. Of course not as known and big as DOOM, but they fulfill this, performing roughly the same in Windows and Linux when using Vulkan.

[humbug]

There's Talos and SS Fusion. Of course not as known and big as DOOM, but they fulfill this, performing roughly the same in Windows and Linux when using Vulkan.

Yep, Croteam's Vulkan renderer has been continually improving with patches since The Vulkan launch and now even though it is an unfinished beta for talos principle the performance now matches windows.

The problem that I referred to mostly afflicts ports done by companies like Feral and Aspyr.

whereas companies like Croteam with Talos, SS fusion and Valve with dota2 on the source 2 engine are achieving the same performance on windows and Linux I guess cause they do proper cross platform development. Still Maybe there is something for the porting companies to learn from them.