Core i7 8700K vs. Ryzen 7 2700X With Rise of The Tomb Raider On Linux

[Kendji]

Their pretty tight, I'm interested in seing a windows vs Linux comparison. If linux, with vulkan, can beat Windows.

[humbug]

That's seems to be a good implementation of vulkan

Nope, even when using Nvidia who have the fastest Linux Vulkan driver it's still 15-20% slower than the windows dx11 and dx12 version. Not a great advertisement for Vulkan...

[eydee]

There's no way to know if that's true considering that this case is a GPU bottleneck.
To find out, match the clock speed of one intel and one ryzen processor and run multiple benchmarks which have a CPU bottleneck.

1080p low settings on a Vega 64 is bottleneck really? In what universe?

[ramrod]

There is a reason why many CPU comparisons is done with the most powerful GPUs and/or games running on resolutions even smaller than 1080p.

Personally I ignore those kinds of test since they're testing hardware that is so expensive I would never even consider buying it, while also using settings I would not enjoy.

[theriddick]

Nope, even when using Nvidia who have the fastest Linux Vulkan driver it's still 15-20% slower than the windows dx11 and dx12 version. Not a great advertisement for Vulkan...

I thought NVIDIA's Vulkan driver was on par with windows vulkan driver? I do know AMD does better at DX12 and Vulkan in certain apps under Windows.
The real big draw for Vulkan is the cross platform ease of use and also the OPTION to improve performance if you put the effort in (wrappers generally won't bring out the best performance, i.e. Feral). Good example of decent use of Vulkan is Croteam (Talos).

In saying that wrappers for DX11 to Vulkan do bring out pretty decent performance, for example I get 45-60(50+ avg) FPS in Fallout4 at 4k with godrays and some performance tweaks under DXVK. Pretty impressive for a game that is terribly optimized, you can look 10deg left and get 30fps, then 10deg right and get 60fps.. lol

[pal666]

Nope, even when using Nvidia who have the fastest Linux Vulkan driver it's still 15-20% slower than the windows dx11 and dx12 version. Not a great advertisement for Vulkan...

nvidia is different driver
no matter how fast is it, it does not prevent windows rottr + amd windows driver from being slower than linux rottr + radv(i don't think it is the case, but not because of nvidia results)

[Unshra]

Seeing this comparison made me happy. For a long, while I felt my red build was underperforming and while I still think that's true it's nice to see that it did the similar to both the Intel and second gen Ryzen on the QHD very high pass. Scoring 91.72, 67.82 and 65.05 respectively. With that said it's a Ryzen 7 1700 clocked to 4.9 at 1.3v using a stock spire cooler crammed into mini-ITX (Phanteks Evolv Shift) case with a Vega 64 lc.

So I'm happy to see the numbers are not that off, I use to think I was 20fps too low compared to similar Ryzen builds and had thought about getting the second gen Ryzen to see if that would help any but seeing this has me holding off until Ryzen 2 and Navi. So thank you very much for the comparison.

[riklaunim]

Would it be more intersting? Sure. But there is a reason why nobody does that because even though it shows that there is indeed a difference in these virtually synthetic conditions (720p low), end customer only needs to know that CPU choice doesn't matter for tested games.

Benchmark title is "Core i7 8700K vs. Ryzen 7 2700X" where as the benchmark doesn't test any differences between the CPUs. You put a high end CPU with mid level GPU and you can say both high end CPUs aren't the bottleneck in this game. that's all. Only one of benchmarks shows some sign of CPU differences. And audience interested in performance of such high end CPUs would also use high end GPU or use non-GPU related workloads.

Aside of quantity of benchmarks their quality does matter a lot.

Personally I ignore those kinds of test since they're testing hardware that is so expensive I would never even consider buying it, while also using settings I would not enjoy.

Putting like a 1080Ti there isn't there to benchmark 1080Ti but to show differences between benchmarked CPUs. What you see in this benchmark are results capped by weak GPU for which it doesn't matter if it's a mid or high end CPU. With RX580 it could turn out even that Core i3 is as good as i7 8700k and so on...

[humbug]

I thought NVIDIA's Vulkan driver was on par with windows vulkan driver? I do know AMD does better at DX12 and Vulkan in certain apps under Windows.

Nvidia's is.
AMD's amdvlk and RadV still need more performance tuning, but they are headed in the right direction. Yep on windows AMD always easily wins at Vulkan and dx12 when comparing same tier GPUs.
I.e.
r9 290 > gtx 970
rx 580 > gtx 1060
vega64 > gtx 1080
etc

[humbug]

nvidia is different driver
no matter how fast is it, it does not prevent windows rottr + amd windows driver from being slower than linux rottr + radv(i don't think it is the case, but not because of nvidia results)

It's less of a driver issue now and more a question of how much the application is optimized. Let's face it we do have good drivers now on Linux (both AMD and Nvidia).

But we need to be able to port applications without taking a performance hit.