Early Radeon Vulkan Windows vs. AMDGPU PRO Linux Benchmarks

[phoronix]

Phoronix: Early Radeon Vulkan Windows vs. AMDGPU PRO Linux Benchmarks

On Friday I posted Some Early Windows 10 vs. Ubuntu Linux Vulkan Tests With NVIDIA Graphics while today the tables have turned to show The Talos Principle on Windows 10 and Ubuntu 16.04 Linux under AMD Radeon graphics...

http://www.phoronix.com/scan.php?page=news_item&px=Radeon-VLK-Windows-Linux

[Space Beer]

Vulkan on Windows ended up being much faster than Vulkan on Linux with the current AMD/RTG drivers

I wouldn't say it's much faster, difference is not that big. The best thing is - Vulkan is much faster than openGL

[Kemosabe]

I wouldn't say it's much faster, difference is not that big. The best thing is - Vulkan is much faster than openGL [/FONT][/COLOR]

You get two test scenarios and 3 possibe results. Please assign them to each other correctly:

N: nVidia Vulkan Linux vs Windows
A: AMD Vulkan Linux vs Windows

1: Windows > Linux
2: Linux > Windows
3: Linux ~ Windows

I am very sure everybody would have guessed: A1.
There is nothing more to say.

[duby229]

This is a good comparison, it actually shows how much Vulkan improves the situation for AMD on linux. Obviously it means we have to wait to get optimized titles with Vulkan renderers. So for now it means the status quo on OpenGL performance and competitive on Vulkan performance.

[jf33]

Michael, your judgement seems to be biased. Given the benchmark results, you can't say OpenGL had the "slight upperhand" vs D3D11 on Nvidia and at the same time call AMD's Vulkan "much faster" on Windows than on Linux.

[blackout23]

The Talos Principle developer thinks it's a bad idea to benchmark highly GPU bound scenarios (4K) at this point.

http://steamcommunity.com/app/257510...48158145024457

• Vulkan (and OpenGL, for that matter) are much slower at higher resolutions and visual qualities than DXs. So, as you become more and more GPU-bound, VLK/OGL will get slower and slower. Both us and IHVs are still in dark of what exactly is going on there, but currently shader compilers are no.1 suspects in this case. Will let you know how this unfolds... in the meantime, don't benchmark GPU-bound scenarios, just... don't!.

800x600 would be better.

[duby229]

The Talos Principle developer thinks it's a bad idea to benchmark highly GPU bound scenarios (4K) at the point.

http://steamcommunity.com/app/257510...48158145024457



800x600 would be better.

My opinion is screw that! Benchmark what you play at and if they are trying to hide something, we will figure out what it is on our own.

EDIT: Why do I feel like it's 1998 all over again? The whole point in Vulkan is that there is no point in benchmarking the API.

[lvlark]

Micheal, your judgement seems to be biased. Given the benchmark results, you can't say OpenGL had the "slight upperhand" vs D3D11 on Nvidia and at the same time call AMD's Vulkan "much faster" on Windows than on Linux.

Yep, this. It's a small difference across the board, and even smaller for the high-end card.

Quite happy with these initial results. For one, Vulkan outperforms D3D11 (albeit not with a huge margin) and OpenGL (at least on AMD). For another, AMD Linux performance isn't as far behind as it used to be.

[PublicNuisance]

I'm excited for Vulkan overall but especially for what it could do for AMD gaming performance on Linux. Everything I see just gets me more pumped for future drivers and games.

[lvlark]

My opinion is screw that! Benchmark what you play at and if they are trying to hide something, we will figure out what it is on our own.

My opinion is - take it into consideration when you do benchmarks. Realise there's alot of things that could change the outcome once Vulkan usage has matured.