are AMD and nVidia ready for SteamOS launch? phoronix tested 22 video cards

[Inkz]

A  460 beating a Fury? Man these drivers are really weird.

[zMeul]

And well no, as it stands there was no reason for this not to be just a direct Ubuntu or Debian fork

but it is a direct Debian fork

if I recall, they initially went for a Ubuntu fork, but "something something" they're now on Debian

  

The base system draws from Debian 7, code named Debian Wheezy. Our work builds on top of the solid Debian core and optimizes it for a living room experience.

http://store.steampowered.com/steamos/

---

overall, yes I agree: ValvE shat the bed with SteamOS

if Gabe was so opposed to Windows as he claimed, ValvE would've poured blood and tears into this

[Daegun]

How do those two not explain it (assuming the charts are true)?

Sorry for the late reply, work went from dead to hectic in nothing flat.

Because I'm not talking about the 680 beating the Titan when they're within 4 frames of each other or AMD VS Nvidia here?

What I'm talking about is:

CS:GO

-780 Ti tieing Titan X

-970 beating 980

-750 Ti has less than 2/3 the performance of a 950 (it's a lot slower, it's not that slow)

DiRT Showdown

-970 beating 780 Ti

-950 beating 760 by over 10%

Metro Last Light

-950 beating 760 (though not by as much)

Those should be examples that aren't covered by either of the explanations. The benchmarks are off. I don't know what went wrong or how, but something did.

Edited October 25, 2015 by Godlygamer23
Correcting "quote" code.

[PlayStation 2]

Sorry for the late reply, work went from dead to hectic in nothing flat.

Because I'm not talking about the 680 beating the Titan when they're within 4 frames of each other or AMD VS Nvidia here?

What I'm talking about is:

CS:GO

-780 Ti tieing Titan X

-970 beating 980

-750 Ti has less than 2/3 the performance of a 950 (it's a lot slower, it's not that slow)

DiRT Showdown

-970 beating 780 Ti

-950 beating 760 by over 10%

Metro Last Light

-950 beating 760 (though not by as much)

Those should be examples that aren't covered by either of the explanations. The benchmarks are off. I don't know what went wrong or how, but something did.

To be fair on the Dirt Showdown and the 970 beating the 780 Ti, it might just be a margin of error.

[Daegun]

To be fair on the Dirt Showdown and the 970 beating the 780 Ti, it might just be a margin of error.

That's still a problem, the point there is that they shouldn't be that close. The 970 performs about the same as a 290x and the 780 Ti beats the 290x by quite a bit.

[PlayStation 2]

That's still a problem, the point there is that they shouldn't be that close. The 970 performs about the same as a 290x and the 780 Ti beats the 290x by quite a bit.

Meh, IMO that's not something worth complaining about.

 

What IS worth complaining about is how the 950 beats the R9 Fury in CSGO and Dirt Showdown.

Or even better, the 750 beating the Fury in Metro Redux.

Yeah, there's a lot of bullshit in these benchmarks

[Daegun]

Meh, IMO that's not something worth complaining about.

 

What IS worth complaining about is how the 950 beats the R9 Fury in CSGO and Dirt Showdown.

Or even better, the 750 beating the Fury in Metro Redux.

Yeah, there's a lot of bullshit in these benchmarks

Yeah, but that could very well be the result of them having bad drivers. Things that can't be explained by that are stuff like that and the 7950 and 285 beating the Fury.

[Humbug]

Disgraceful performance from AMD. After years and years their openGL performance is still terrible, when you consider the underlying potential of the hardware. And how great they perform with directX...

 

People will make a lot of excuses for example saying the Nvidia drivers do non-conformant stuff, that devs have started relying on these non-conformant behaviors etc... But the fact is AMD is supposed to be out there having active relationships with game developers, and they are also a part of Khronos group so nobody should claim that they are innocent victims here. They should be out there making things happen. They perform so great with directX on windows.

 

If they want to say Linux is not a priority given their current financial situation then ok; that's a corporate strategic decision. But then stop advertising your cards as Linux supported!

 

I wonder if the same performence difference will show in Windows if we benchmark OpenGL games instead of DirectX. If not, the problem will probably not be too hard to fix on AMD's side.

If you benchmark openGL in windows you will get similar results. Terrible AMD performance. The difference is that nobody uses openGL on windows games. Even id software is moving to directX for Doom 4.

 

What seems clear is that even Valve has basically given up on openGL and is heavily investing in the Vulkan ecosystem. I'm sure from the near future onwards all AAA linux games will be on Vulkan. Which is a good thing. But the problem is at least for the foreseeable future openGL is not going away when it comes to smaller indie developers who don't want to learn new stuff, don't have resources or expertise in-house. Either that or they all start relying on 3rd part engines such unreal4, Unity, Source2 etc which give them Vulkan.

[Bob Jim]

Disgraceful performance from AMD. After years and years their openGL performance is still terrible, when you consider the underlying potential of the hardware. And how great they perform with directX...

 

People will make a lot of excuses for example saying the Nvidia drivers do non-conformant stuff, that devs have started relying on these non-conformant behaviors etc... But the fact is AMD is supposed to be out there having active relationships with game developers, and they are also a part of Khronos group so nobody should claim that they are innocent victims here. They should be out there making things happen. They perform so great with directX on windows.

 

If they want to say Linux is not a priority given their current financial situation then ok; that's a corporate strategic decision. But then stop advertising your cards as Linux supported!

 

If you benchmark openGL in windows you will get similar results. Terrible AMD performance. The difference is that nobody uses openGL on windows games. Even id software is moving to directX for Doom 4.

 

What seems clear is that even Valve has basically given up on openGL and is heavily investing in the Vulkan ecosystem. I'm sure from the near future onwards all AAA linux games will be on Vulkan. Which is a good thing. But the problem is at least for the foreseeable future openGL is not going away when it comes to smaller indie developers who don't want to learn new stuff, don't have resources or expertise in-house. Either that or they all start relying on 3rd part engines such unreal4, Unity, Source2 etc which give them Vulkan.

Just take a moment to appreciate the amount of sheer stinking bullshit is in this source. A 7950 and 285 beats the fury, so that isn't AMD/Nvidia driver issues. Also there is no way in hell that a GTX 460 BEATS the fury, no matter how shit drivers are; not to mention that the 550Ti loses to it by some margin, sooo. And apparently the 680 has better performance than the 980Ti in some games. No matter the underlying problem of OpenGL performance, these graphs are a big stinking pile of bullshittery.

[Curufinwe_wins]

Lol every single benchmark listed showed huge cpu bottlenecks.

[Humbug]

Just take a moment to appreciate the amount of sheer stinking bullshit is in this source. A 7950 and 285 beats the fury, so that isn't AMD/Nvidia driver issues. Also there is no way in hell that a GTX 460 BEATS the fury, no matter how shit drivers are; not to mention that the 550Ti loses to it by some margin, sooo. And apparently the 680 has better performance than the 980Ti in some games. No matter the underlying problem of OpenGL performance, these graphs are a big stinking pile of bullshittery.

Have you been consistently following Linux openGL gaming performance benchmarks? Especially AMD? I assume not or you wouldn't be surprised.

ps- 680 does not beat 980ti, both systems are bottlenecked clearly by something (non-GPU) and within margin of error.

 

Have you played AAA games on Linux on AMD GPUs? While playable often the openGL driver bottleneck is so significant that the GPU is not even working hard.

 

Hitting a wall or some kind of CPU bottleneck is acceptable. At some point something has to bottleneck right. But having it happen at half the framerate of comparable Nvidia GPUs is not acceptable.

 

Why do you think almost all the steam machines come with Nvidia GPUs? Is it because Valve likes Nvidia? No; it's the only choice they had.

I say this as a proud R9 290 user who is extremely satisfied by the quality and performance of my GPU in almost every way. It's been the best, smoothest gamign experience of my life. Linux is the only downside.

[dimitriianghelov]

who knew SteamOS is shit and Titan X performs like a 680, amd has 50% the performance, and i should go install linux because ? hipsters ....

 

idd love to see some 4k results, i bet it's like 10 fps....

[Humbug]

who knew SteamOS is shit and Titan X performs like a 680

nonsense.

That's particular Bio-shock Infinite 1080p result is obviously either a framerate CAP or a CPU bottleneck. Both GPUs are doing 120fps+ and within margin of error. Which means it doesn't matter.

 

Turn up the resolution and do you think the 680 will hang in there?

[dimitriianghelov]

nonsense.

That's particular Bio-shock Infinite 1080p result is obviously either a framerate CAP or a CPU bottleneck. Both GPUs are doing 120fps+ and within margin of error. Which means it doesn't matter.

 

Turn up the resolution and do you think the 680 will hang in there?

 

check the benchmarks on windows (ultra ) about 40 fps more for titan x

 

cpu bottleneck : looks at Core i7 5960X ...

[Kinda Bottlenecked]

Woah. Those bechmarks seem really messy? 

 

in AMD's court, things are murky: especially with them cutting jobs from IT department: http://linustechtips.com/main/topic/459883-more-bad-news-for-amd-theyll-cut-5-workforce/

 

Even though this thread is a far cry from toxic threads zMeul made in the past. That comment is still not appreciated. But I'm glad everyone has decided to ignore shit like this.

[Humbug]

check the benchmarks on windows (ultra ) about 40 fps more for titan x

 

cpu bottleneck : looks at Core i7 5960X ...

You do realize that everything bottlenecks eventually right?

Either the GPU or the CPU or something else? Or else you would be getting an infinite fps.

 

If the directX version is not CPU bottlenecking as easily that only means that the openGL drivers have more CPU overhead.

That was not the point of my post, nobody cares if something is bottlenecking at 120fps+ as was the case with those Nvidia GPUs. That means the game is flying...

The point was AMD's poor openGL performance relative to Nvidia's.

[dimitriianghelov]

You do realize that everything bottlenecks eventually right?

Either the GPU or the CPU or something else? Or else you would be getting an infinite fps.

 

If the directX version is not CPU bottlenecking as easily that only means that the openGL drivers have more CPU overhead.

That was not the point of my post, nobody cares if something is bottlenecking at 120fps+ as was the case with those Nvidia GPUs. That means the game is flying...

The point was AMD's poor openGL performance relative to Nvidia's.

 

of course no one cares because these are 1080p benchmarks on medium/high , no one buys a 980 ti or titan x to play on that resolution...

 

we have 22 cards and 5 games , if you start doing more benchmarks i bet we'll find more problems

[captain cactus]

AMD needs to work on their drivers for Linux. Actually both the red and green team need to do that.

[Humbug]

of no one cares because these are 1080p benchmarks on medium/high , no one buys a 980 ti or titan x to play on that resolution...

It's not medium/high but ok. that's a different subject and besides the point I was making about AMD's openGL situation.

[tsk]

It's by Zmeul! Of course AMD does badly! But seriously 680 vs Titan X makes no sense, I wouldn't trust these too much.

I would trust tests from Michael Larabel more than most places, including linustechtips.

[dimitriianghelov]

It's not medium/high but ok. that's a different subject and besides the point I was making about AMD's openGL situation.

 

it's not ultra that i can say for sure, i wasn't even speaking about amd but about steamos... i'm a bit skeptical about this article because it looks like it doesn't want to provide a clear picture of SeamOS for gaming at this point, you won't get 120(on ultra) fps on ultra in new games, hell most new big games aren't even supported native

[Tigerbomb8]

Looking at these results make me think the article was sponsored by nvidia.. I am not saying it is biased..

[Khvarrioiren]

Looking at these results make me think the article was sponsored by nvidia.. I am not saying it is biased..

 

It's just AMD's absolutely abysmal Linux drivers more than anything else.

[Humbug]

It's just AMD's absolutely abysmal Linux drivers more than anything else.

People on LTT community are surprised because they do not usually follow openGL benchmarking and driver developments. So they go in expecting results to mirror directX performance. The reality is these results have little to do with what the hardware is capable of and all to do with openGL drivers and applications.

[samcool55]

Tbh i'm not that suprised. I mean DX12 for windows 10 will probably be a bigger deal than this, and as far as we know AMD cards like DX12 very much compared to Nvidia. So if AMD needs to choose between steamOS or DX12, you probably know where their focus is on.