NVIDIA Vulkan Driver Shows Impressive Mad Max Linux Performance

[ramrod]

We know Vulkan is better than opengl.

But what we - at least I - want to know is when it is better than directx (I supose sometimes is and others not) under WINE and under native MS WOS, and if you want to work a bit more with MS WOS under Xen an KVM GPU pass through VMs.

And that is the battle, because if Vulkan beats directx, and even better Vulkan under GNU/Linux beats also Vulkan under MS WOS, then gamer users will switch to GNU/Linux, and gamer users are a small, but big enough computer market share.

I found this video showing the performance of the same cards except the 1080 ti on windows. The performance looks fairly close to the benchmark results in this article.

https://www.youtube.com/watch?v=jwQ7DpXA-eI

[Danny3]

Finally a developed driver that could show how good Vulkan really is.
I hope to see it used in games for Windows too, since I will never upgrade from Windows 7 to the crappy W8 or spyware W10.
I will definitely look to buy, in the future, only games who use Vulkan so I can play them on both Windows 7 or Linux.

[gamerk2]

Yeah, 480 is faster than a 1070, Fury almost even with 1080 Ti. What is wrong here?

Its the scheduler. As I've been noting for YEARS now, the scheduler is sucking down CPU performance in certain situations, gaming being the most obvious. The switch from OGL to Vulkan is freeing up enough CPU time where the GPU is able to do its job, but OGL can put up similar performance numbers if the scheduler is fixed.

[juno]

Is it just me, or does the nvidia OpenGL performance look rather low compared to radeonsi?

Yeah, 480 is faster than a 1070, Fury almost even with 1080 Ti. What is wrong here?

[mitcoes]

We know Vulkan is better than opengl.

But what we - at least I - want to know is when it is better than directx (I supose sometimes is and others not) under WINE and under native MS WOS, and if you want to work a bit more with MS WOS under Xen an KVM GPU pass through VMs.

And that is the battle, because if Vulkan beats directx, and even better Vulkan under GNU/Linux beats also Vulkan under MS WOS, then gamer users will switch to GNU/Linux, and gamer users are a small, but big enough computer market share.

[darkbasic]

Unfortunately with Mad Max I don't believe there is an equivalent benchmark mode on Windows to Feral's benchmark mode with this being a new addition to the vulkan beta. But if there is, happy to tesf.

Kudos to Feral for adding the benchmark mode then.

[andre30correia]

Am I the only one that thinks it's strange that the 'normal' and 'very high' presets have almost exactly equal performance?

I saw this on the radeonsi test as well. I haven't tried the game my self, but is there so little difference between the settings?

most likely the difference is minimal from normal to very high, or cpu is bootleneck the performance, Vulkan is young the drivers are young and the engines are young we will need two years at least

[Avenger]

Am I the only one that thinks it's strange that the 'normal' and 'very high' presets have almost exactly equal performance?

I saw this on the radeonsi test as well. I haven't tried the game my self, but is there so little difference between the settings?

[bug77]

Michael
Could You please do also a test-run with the performance governor for the CPU? I'm pretty certain that this would give even more impressive numbers, since powersave tends to downclock the CPU under Vulkan, because of the lighter load on each individual CPU core/thread.

PS: Somebody should give Crysis (1) [2007] the Vulkan treatment - just so Intel IvyBridge users can claim: "Yes, it can run Crysis!"

Tbh, if the CPU is underused, it should be left to throttle.

[VikingGe]

As the GPU is maxed out, I see no reason for a Windows version to be faster. At best the performance would be more or less equivalent.

Actually it's Vulkan that has the slightly higher performance potential when correctly using renderpasses and out of order rasterization - the latter is an AMD extension, however.