Windows 10 vs. Ubuntu Linux Gaming Performance With NVIDIA GeForce GTX 1060/1080

[Michael]

Any Idea why Talos with Vulkan is still faster on Windows ?
Do we really compare the same driver version here ?
I thought the drivers were sharing the same code base, so the results should be the same don't you think ?

The latest drivers on each platform were used, but keep in mind that the code-base isn't 100% the same and that there are also kernel/scheduler differences, among many other differences, between the operating systems.

[spykes]

The latest drivers on each platform were used, but keep in mind that the code-base isn't 100% the same and that there are also kernel/scheduler differences, among many other differences, between the operating systems.

Could really the scheduler explain a 20 fps gap between Linux and Windows (in low quality) ? That's quite unlikely IMHO...
Anyway a deeper analysis on that difference would be really interesting. I'm pretty sure the dev from Croteam has a good idea of what's going on here.

[geearf]

I think it would be nice to include wine numbers when possible, at least when there's a huge disparity between Linux and Windows.

Another thing that would be great, would be to get some comment from Aspyr and Feral on these. As an AMD user on the OSS driver, I always expect FPS to be less than in Windows with D3D, but I'd expect Nvidia numbers to be about the same. A 40% is massive and shouldn't probably be blamed on the driver... I have read before porters saying they don't have as much money and time to optimize as much as the Windows devs, and that's acceptable for a small to medium difference, but almost half the speed...

[dado023]

I am not sure if it was mentioned, but comparison to Windows 7 would be awesome, since a lot of users are still on Win7.
Cheers

[Michael]

I am not sure if it was mentioned, but comparison to Windows 7 would be awesome, since a lot of users are still on Win7.
Cheers

Very unlikely considering so far today just one new premium subscriber, so already will be tough breaking even just on these tests.

[theriddick]

That would exclude DX12 only, the performance with win7 and win10 is much the same except for DX9 related vram limit bug which exists in newer windows versions.

[ngkaho1234]

Thanks for the interesting comparsion.
From the results(GPUTest and OpenArena) they show that the binary Nvidia driver performs the same on both platform.
Briefly speaking, two factor here matter (if we talk about scheduler difference or anything related to Linux kernel things go too complicated)
1. Considering that most of the games are ported with d3d call -> ogl wrapper
2. Renderer path optimization if these games are not ports.

[indepe]

Any update on Civ 6's multi-threading? I read in a bunch of places that it's single threaded, but I have a 4 core/8 thread Intel CPU, and I've seen it using 30%+ in System Monitor.

Edit - Meant Civ 6.

Interesting observation, given there was the impression that it is CPU bound.

So you are seeing 30% cpu usage on one core, and even less on the others?

[TheOne]

This benchmarks are depressing, seen how huge is the performance gap between dx11 and opengl, running some games on linux where performance is trash (like ark survivor evolved), while on windows is silky smooth makes me think that linux will never catch up as a gaming platform, or maybe the industry is just biased to the windows side, after all, after so many years now is that linux is been considered a gaming platforms thanks to steam so no effort was done before to properly optimize the linux software stack to run intensive games which is sad for enthusiasts that are trying the platform now. How many more years we need to have decent linux gaming performance!!! And when I say decent is equal or better than windows.

[theriddick]

There is more effort in optimization with DX11, opengl generally gets ignored when it comes to optimization and performance patches of games. Couple that with DX11 being easier to work with according to allot of devs. Its no surprise.,