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

[Beskamir]

What I find odd is a 680 being better than a Titan X.

The SE basically makes them the same though. Still shouldn't be happening though so it's either a driver issue or some other part of the system is bottlenecking.

[PlayStation 2]

I smell a lot of bullshit to this. The Fury shouldn't be beaten by the 7950, and the 680 shouldn't be beating the Titan X. And the 970 beating the 980 in CSGO.

[SirSquid]

What I find odd is a 680 being better than a Titan X.

I thought it was odd that the 980TI beat the Titan x in CSGO

[ManWithBeard1990]

I honestly didn't expect these results. I like AMD and I've always found my APU to perform really well. Better than the GT555M I had before, anyway. 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.

[Notional]

I honestly didn't expect these results. I like AMD and I've always found my APU to perform really well. Better than the GT555M I had before, anyway. 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.

 

That is exactly what we see in Windows too, when it comes to OpenGL. Like I stated earlier in the thread, this is an OpenGL issue, not a Linux issue per se. I doubt AMD will fix it, because OpenGL is so fubar it cannot be fixed:

 

https://www.techpowerup.com/reviews/MSI/GTX_980_Ti_Gaming/26.html

​

Now if people don't believe this is an OpenGL issue, why don't we ask Valve themselves then? Here an interview with a Valve OpenGL programmer:

http://richg42.blogspot.co.uk/2014/05/the-truth-on-opengl-driver-quality.html

​

Vendor A = NVidia

Vendor B = AMD

Vendor C = Well you should be able to guess by how utter useless they are at gaming (hint, it rhymes with wIntel)

 

If you want to learn more about why OpenGL is utter useless crap, this history of OpenGL is really interesting:

 

http://programmers.stackexchange.com/a/88055

 

That being said, Vulkan being made from the ground up with no obsolete or redundant code, as well as using a lot of code and features from mantle, should change things and make a Kronos API somewhat useful again. It will not change any outcome of OpenGL games though.

[Daegun]

Something is very weird with these benchmarks. Cards that shouldn't be beating other cards are, in multiple cases, on both sides. Even on the ones not capped/bottlenecked in some way.

[Guest Kloaked]

Something is very weird with these benchmarks. Cards that shouldn't be beating other cards are, in multiple cases, on both sides. Even on the ones not capped/bottlenecked in some way.

 

It's been explained, more or less, in the previous replies to the thread.

[Daegun]

It's been explained, more or less, in the previous replies to the thread.

I've read the thread lol

Explanations are:

-Something is bottlenecked things or the frames are capped

-AMD has terrible Linux drivers

Neither would explain some of these results. Unless I missed something in the thread, then please let me know.

[Guest Kloaked]

I've read the thread lol

Explanations are:

-Something is bottlenecked things or the frames are capped

-AMD has terrible Linux drivers

Neither would explain some of these results. Unless I missed something in the thread, then please let me know.

 

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

[NAP51DMustang]

If you look at the AMD scores in CS:GO, Metro 2033 and Metro: Last Light, they are scaling like they should. Granted they can still perform a lot better, but they are indeed scaling like they should more or less.

 

Nvidia clearly has the better drivers, should this be true. That's the only explanation, if these charts are true. Nothing that can't be fixed, again, if these charts are true.

 

 

 

Nvidia's drivers are being bad? With how you're saying that, you're acting like Nvidia should nerf the performance of the lower tiered cards, which is retarded.

 

You should think about this for a minute.

um no, Im saying if a 300 USD or less price point card is showing similar performance to a 1000 USD card, the drivers are still shit (or the game is shit). if the drivers (or game) weren't shit, you'd see a proper distribution of performance with lower performance capable cards being were they'd be expected to be in releation to the top tier card

[Guest Kloaked]

um no, Im saying if a 300 USD or less price point card is showing similar performance to a 1000 USD card, the drivers are still shit (or the game is shit)

 

Even when they're both running the game at over 200fps? lmao

[NAP51DMustang]

Even when they're both running the game at over 200fps? lmao

Im not talking about the values in these benches. Im saying if a 970 if doing 200 FPS, a Titan X should be at 300 or more. get over yourself.

[Guest Kloaked]

Im not talking about the values in these benches. Im saying if a 970 if doing 200 FPS, a Titan X should be at 300 or more. get over yourself.

 

You should get over yourself, mate. FPS isn't always captured at the exact cap. Games that have "unlimited" framerates are actually capped somewhere.

 

Again, this is all assuming these benchmarks are even accurate, but my point still stands.

 

You're wrong.

[NAP51DMustang]

You should get over yourself, mate. FPS isn't always captured at the exact cap. Games that have "unlimited" framerates are actually capped somewhere.

 

Again, this is all assuming these benchmarks are even accurate, but my point still stands.

 

You're wrong.

who is talking about caps? Im talking about relative performance which with these benchs shows that Nvidias drivers, while better than AMDs, are still shit because you have lower end cards and high end cards from 4 or so years ago hitting the same performance values as a 6 month old top tier card. it shows that either the games are shit (highly likely) and or the drivers are shit, and seeing as this is on linux probably both. which also goes to show that even tho there's about 1000 indie titles on steam and about 50 AAA titles "linux ready" Linux itself as an entire experience isn't ready yet, especially Steam OS (which Valve should have never had made and just used Ubuntu or the next most popular flavor of LInux and then developed out BPM more).

 

also you can easily develope a game with no frame rate cap on the game engine end (the gpu obviously hits it's limit to render).

[Misanthrope]

R9 Fury slower than a 460 in Bioshock Infinite.

God AMD, how could you even do that?

 

Contrary to popular belief, AMD are actually equally bad or worst in Linux and they've had 2 years to fix this and sat on ass: For all their motherfucking talk about open software this and that, when they actually could create a decent and truly open source implementation, a proper open source linux driver, they of course failed miserably.

 

This is kinda why Steam OS is dead on the fucking water before it's even out: Valve couldn't move Nvidia and specially not AMD out of their asses to get proper driver support. They simply were either called out as pipe fucking dreamers or failed to offer anything more than moral support when they should have been saying "Here, take this fucking money from us Valve and code fucking linux drivers" If they were at all interested in their Steam OS product to succeed, which it wont.

[NAP51DMustang]

Contrary to popular belief, AMD are actually equally bad or worst in Linux and they've had 2 years to fix this and sat on ass: For all their motherfucking talk about open software this and that, when they actually could create a decent and truly open source implementation, a proper open source linux driver, they of course failed miserably.

 

This is kinda why Steam OS is dead on the fucking water before it's even out: Valve couldn't move Nvidia and specially not AMD out of their asses to get proper driver support. They simply were either called out as pipe fucking dreamers or failed to offer anything more than moral support when they should have been saying "Here, take this fucking money from us Valve and code fucking linux drivers" If they were at all interested in their Steam OS product to succeed, which it wont.

well why would either company jump on this? with <5% of the ENTIRE desktop market on Linux and even less when considering just gamers (more than likely I haven't checked the Steam hardware survey in a while) what motivation do they have? And I agree on Steam OS being DOA. But I would argue it was DOA because it came from Valve and they take WAY too long to develope anything (due to ther corprate structure) and the only reason the Vive VR headset will see the quick release it's seeing is thanks to HTC and not Valve. Don't know why anyone though SteamBoxes and Steam OS was ever going to be a thing.

[Prysin]

What I find odd is a 680 being better than a Titan X.

everything is odd in Linux

[Bouzoo]

I don't remember the last time I saw numbers that were all over the place like these are. This is a whole new level of bad. 

As to OP, they both are bad. Yes NV is a bit better but you can't recommend NV over AMD, they both are really really really bad. I'd only recommend not going in Steam OS for some time. Until we see proper drivers things might change a lot, and I sincerely hope they will. 

[Misanthrope]

well why would either company jump on this? with <5% of the ENTIRE desktop market on Linux and even less when considering just gamers (more than likely I haven't checked the Steam hardware survey in a while) what motivation do they have? And I agree on Steam OS being DOA. But I would argue it was DOA because it came from Valve and they take WAY too long to develope anything (due to ther corprate structure) and the only reason the Vive VR headset will see the quick release it's seeing is thanks to HTC and not Valve. Don't know why anyone though SteamBoxes and Steam OS was ever going to be a thing.

 

I don't have answers: AMD is falling into a financial black hole and too focused on pulling out of that almost inevitable CPU business bankrupcy to care about Valve's promises of great business opportunity vs the reality of maybe 5% market share for Linux for the consumer PC

 

Valve on the other hand, it's basically a planet sized headless chicken: It's big enough to just pull money towards it by virtue of their size and gravitational field they create, but that doesn't mean they display any sort of coherent direction.

 

Here's what we're told Valve works like

 

1) Someone (likely Gaben) got a bone to pick with windows (justifiably so in my opinion but besides the point

2) They decide "To well with Windows, we'll make our own Operative system, with blackjack and hookers!"

3) They get to spend a lot of money to build some initial hype and good will

4) Nobody takes into account how massive this project is and nobody has enough decision making influence to invest a severely large ammount of man power and resources into fixing the enormous ammount of issue Linux faced

5) Nobody had the authority to say "Stop playing with the motherfucking 3D printers, drop the controller and focus on DRIVER support motherfucker and release the damn distro already!"

 

At the end of the day they're lucky they got into digital distribution early and decisively. That was probably the last fucking directive decision they ever made that involved not only forward thinking and innovation but assertiveness and investment something they just haven't done since the last what, 10 years now?

[marldorthegreat]

R9 Fury slower than a 460 in Bioshock Infinite.

God AMD, how could you even do that?

Its a platform that will never take off like windows Has.

[bogus]

Contrary to popular belief, AMD are actually equally bad or worst in Linux and they've had 2 years to fix this and sat on ass: For all their motherfucking talk about open software this and that, when they actually could create a decent and truly open source implementation, a proper open source linux driver, they of course failed miserably.

 

This is kinda why Steam OS is dead on the fucking water before it's even out: Valve couldn't move Nvidia and specially not AMD out of their asses to get proper driver support. They simply were either called out as pipe fucking dreamers or failed to offer anything more than moral support when they should have been saying "Here, take this fucking money from us Valve and code fucking linux drivers" If they were at all interested in their Steam OS product to succeed, which it wont.

You're missing a huge factor there, which is Vulkan ... AMD literally gave their API to Khronos, so I doubt AMD is going to put efforts on anything other then vulkan.

But you're right, Valve has good intentions, but the world moves with money - imo they shouldn't put money in HIVs, they should put money on devs to make content for steamos - quality content that would make people install steamos - if they made HL3 exclusive for SteamOS you would see drivers come up for it

[AlexGoesHigh]

http://programmers.stackexchange.com/a/88055

That was a great read, no wonder OpenGL failed despite being such a good concept in theory.

[Misanthrope]

well why would either company jump on this? with <5% of the ENTIRE desktop market on Linux and even less when considering just gamers (more than likely I haven't checked the Steam hardware survey in a while) what motivation do they have? And I agree on Steam OS being DOA. But I would argue it was DOA because it came from Valve and they take WAY too long to develope anything (due to ther corprate structure) and the only reason the Vive VR headset will see the quick release it's seeing is thanks to HTC and not Valve. Don't know why anyone though SteamBoxes and Steam OS was ever going to be a thing.

 

That's exactly what I was getting at.

 

Here's how you can make Linux a success

 

1) Take the Linux as a starting platform

2) Branch it off (something you're allowed to do) into your own to avoid the wrath of the Stallmans of open source community (Since your interest in Linux is presumably to make business, not to make advancements for socialism through software code) 

3) Invest metric tons of fucking money on it

4) Great success!

 

Or in other words: Google and Android.

 

Valve could have had a similar success story with Steam OS but for all the aforementioned reasons yes: it's dead in the water.

[NAP51DMustang]

That's exactly what I was getting at.

 

Here's how you can make Linux a success

 

1) Take the Linux as a starting platform

2) Branch it off (something you're allowed to do) into your own to avoid the wrath of the Stallmans of open source community (Since your interest in Linux is presumably to make business, not to make advancements for socialism through software code) 

3) Invest metric tons of fucking money on it

4) Great success!

 

Or in other words: Google and Android.

 

Valve could have had a similar success story with Steam OS but for all the aforementioned reasons yes: it's dead in the water.

don't know if you noticed but you actually quoted the same post of mine twice hehe

 

edit: but to add I think making another fork of Linux wouldn't really help here. What linux needs is for one distro to be setup so that it can be pushed as "the" distro so that less technically inclined people can have a "rallying" point per se. Imo if Valve were serious about this they would have done like you said, invest a ton of money, but at the same time done it while in conjuction with one of the major distros out right now and not have gone all solo and try and basically fork debian (as I understand it).

[Misanthrope]

don't know if you noticed but you actually quoted the same post of mine twice hehe

 

edit: but to add I think making another fork of Linux wouldn't really help here. What linux needs is for one distro to be setup so that it can be pushed as "the" distro so that less technically inclined people can have a "rallying" point per se. Imo if Valve were serious about this they would have done like you said, invest a ton of money, but at the same time done it while in conjuction with one of the major distros out right now and not have gone all solo and try and basically fork debian (as I understand it).

 

Didn't notice sorry.

 

And well no, as it stands there was no reason for this not to be just a direct Ubuntu or Debian fork. You know the thing we have hundreds of already and that's relatively easy to do? Just mod whatever you want on it and call it a day, focus on other aspects like driver development and encouraging game devs or throwing more support at Vulkan to get it here faster, whatever.

 

On my case however, if they were willing to invest heavily on a fork it would have been better cause if you're serious about Linux as a business, you also need to know that you're not really in control of a lot of the advancement you might need. It's a nice platform to start with but have you seen some of the stances people on the Linux community take? Whenever or not you agree with their ideological reasons you can't deny they're highly closed to change and outside intervention and it's a pretty much develop-by-committee situation, just not ideal if your business model depends on it.

 

It didn't take Red Hat just a few years to get to where they are and both Google and Valve had different needs and needed to progress much faster and specific to their needs.