Big Graphics Card Comparison Of Metro Redux Games On Linux

[Kano]

@duby229

Why do you write how good AMD OSS drivers are in a thread about OpenGL 4 only games? Pure nonsens from an AMD fan boy...

[duby229]

@duby229

Why do you write how good AMD OSS drivers are in a thread about OpenGL 4 only games? Pure nonsens from an AMD fan boy...

More like OSS fanboy to be honest with you.

My posts were only in response to to what was already said in this thread.

[MoonMoon]

Most of them compared to anything Intel has. As such if what you want is OSS support, AMD is the only choice.

When nVidia starts supporting OSS drivers, I'll start recommending them, Until then, with OSS drivers AMD is plainly and obviously clearly the fastest cards on linux.

Intel never claimed to have high performance video cards, so that argument doesn't count. And having the fastest card under Linux doesn't count when there is no other real contender. The OSS drivers are fine when you go for low power hardware, used for video decoding or a 3D desktop, but for serious gaming? At least at the moment: not at all.

[moilami]

Out o' my way, fanboy!

May I ask you why, as you see those results??

If I can get 30 FPS on max settings, then that is good enough for me. I don't do gaming for living. Metro reduxes and Empire Total War are the only games I need that card as for now. That said, I gamble a bit, I take a risk in that will AMD develop the Open Source driver or not. I expect to get during next year or two sanely usable Open Source driver. After that I suppose the card is good for 3 more years of casual gaming. I have at the moment HD 4850, which is about 5 years old, and which has served me very well (and would still serve me well, if it could run Metro Reduxes and if I would accept less than max settings on Empire Total War), and I don't think it is too far fetched idea to get a similar or better experience of R9 285.

[MoonMoon]

The fact is that the OSS driver is far more stable and functional than catalyst. 2D performance is at least 10 times faster than catalyst and 3d performance is almost 90% or more in some cases.

Those are facts, so where is your 40% number coming from?

And I have to ask where your numbers come from. 2D performance of Catalyst is mostly on par with the OSS drivers, as shown in benchmarks on Phoronix. And the 40% is explained pretty simple: If the Catalyst driver gets about 50% of the performance that comparable Nvidia cards achieve and the OSS driver gets about 80% of the Catalyst performance you end up with 40% of possible performance for the OSS drivers.

[MoonMoon]

If I can get 30 FPS on max settings, then that is good enough for me. I don't do gaming for living.

If you have to pay double the price for AMD, compared to Nvidia, to get similar framerates, is this still OK for you?

[duby229]

Intel never claimed to have high performance video cards, so that argument doesn't count. And having the fastest card under Linux doesn't count when there is no other real contender. The OSS drivers are fine when you go for low power hardware, used for video decoding or a 3D desktop, but for serious gaming? At least at the moment: not at all.

You're the one making the high performance argument. In the same breath you say Intel performance doesn't matter but AMD performance does. You need to decide for your own self what of those two is actually true. They can't both be true.

[duby229]

And I have to ask where your numbers come from. 2D performance of Catalyst is mostly on par with the OSS drivers, as shown in benchmarks on Phoronix. And the 40% is explained pretty simple: If the Catalyst driver gets about 50% of the performance that comparable Nvidia cards achieve and the OSS driver gets about 80% of the Catalyst performance you end up with 40% of possible performance for the OSS drivers.

You can make up whatever numbers you want. Made up numbers are made up numbers.

You've clearly never used catalyst or the AMD OSS drivers. It's obvious you don't know what you're talking about.

[dungeon]

And I have to ask where your numbers come from. 2D performance of Catalyst is mostly on par with the OSS drivers, as shown in benchmarks on Phoronix. And the 40% is explained pretty simple: If the Catalyst driver gets about 50% of the performance that comparable Nvidia cards achieve and the OSS driver gets about 80% of the Catalyst performance you end up with 40% of possible performance for the OSS drivers.

That is the math If sometimes OSS driver gets 40% of Catalyst and nVidia is at 120% with variabile, that became less then 15% of possibile performance if you also have AMD CPU with cpufreq ondemand

That way we can conclude it is best to use Intel CPU with llvm-pipe .

[duby229]

I can't wait until mesa supports OGL4.2 at least. Then finally these arguments can be put fully to rest. Also I really hope a vulkan drver comes out fairly quick. The only real legitimate problem with the OSS drivers is that they aren't able to play a small number of games. The sooner that gets fixed the better.

Hopefully Vulkan games and device drivers will come at the same time so that we don't fall into a similar poblem with the next generation.