Radeon Linux OpenGL Driver Continues Giving Its Best Against Windows 10

[phoronix]

Phoronix: Radeon Linux OpenGL Driver Continues Giving Its Best Against Windows 10

With having around a Windows 10 installation this week for the latest Windows 10 WSL vs. Linux benchmarking, I also carried out some fresh benchmarks of the Radeon gaming performance between Windows 10 and Ubuntu Linux using the very latest drivers on each platform. This time around a Radeon RX 580 and RX Vega 64 were used for this benchmarking.

http://www.phoronix.com/vr.php?view=26047

[pete910]

Impressive stuff, Shame that the big ports are for the most part 70%ish of the windows counterpart

[prazola]

Weird results with unigine valley. I get 100fps average with an r9 390x paired with an i7 [email protected] and arch default kernel. Is the rx 580 throttling or something?

[dungeon]

Xonotic, Openarena does not have profiles on Windows. They can fire up these easely but i guess these are not top priority... still... yet...still...

[VikingGe]

Weird results with unigine valley.

Valley runs like utter garbage on Polaris in general, even with Dx11. My GTX 670 delivered almost the same performance in that benchmark.

Not representative of actual games, of course.

[dungeon]

Weird results with unigine valley. I get 100fps average with an r9 390x paired with an i7 [email protected] and arch default kernel. Is the rx 580 throttling or something?

Unigine valley via phoronix benchmark or something else? Because there could be difference with one or another way of running these benchmarks.

[bisby]

Thanks! These are great tests. OpenGL vs OpenGL head to head is interesting but the DirectX vs OpenGL very clearly demonstrates the "What you would get if playing in Windows" vs "What you would get if playing in Linux" because lets be honest, no one games on OpenGL on Windows if DirectX is an option. And the games where OpenGL is the "only" option, it is generally optimized enough (Doom 2016), but are rarer, and don't necessarily get Linux ports anyway.

OGL v OGL only shows thats Linux OpenGL drivers are good enough. But that means nothing if games are still only written in DirectX. Same holds true for Vulkan. Even if linux performs the same or better in vulkan, it doesn't matter if games don't use it. The potential for Linux gaming performance is amazing right now, but DirectX is some real vendor lock in.

[Linuxhippy]

Too bad the Feral ports all perform that bad - they are barely useful for benchmarking :/

OGL v OGL only shows thats Linux OpenGL drivers are good enough. But that means nothing if games are still only written in DirectX. Same holds true for Vulkan. Even if linux performs the same or better in vulkan, it doesn't matter if games don't use it. The potential for Linux gaming performance is amazing right now, but DirectX is some real vendor lock in.

All a question of how much you invest bringing your DirectX game engine to OpenGL. If you use some wrapper library, chances are you won't be able to fine tune everything - and you'll see those 20-30% performance loss so comon for the commercial ports.

[prazola]

Unigine valley via phoronix benchmark or something else? Because there could be difference with one or another way of running these benchmarks.

I launch it manually, what should change? I run it with ultra preset of course and unigine is all about GPU, CPU is idling the entire test.

[bisby]

All a question of how much you invest bringing your DirectX game engine to OpenGL. If you use some wrapper library, chances are you won't be able to fine tune everything - and you'll see those 20-30% performance loss so comon for the commercial ports.

Oh definitely. My point was more about that it shows that it is OpenGL generically getting neglected, not necessarily Linux. If Vulkan gets neglected, Linux will continue to be neglected. But if Vulkan really gains traction over the next few years, the whole "Linux performs way worse!" stigma goes away with it.