Initial AMD Radeon R9 270X Linux Benchmarks

[phoronix]

Phoronix: Initial AMD Radeon R9 270X Linux Benchmarks

I mentioned that Radeon R9 270X Linux benchmarks would be coming and now I have the first Ubuntu Linux results to post of the Radeon R9 270X with the Catalyst driver...

http://www.phoronix.com/vr.php?view=MTQ4OTY

[schmidtbag]

Michael, due to AMD's recent profits, the possible usage of using them for SteamOS, and the overall greatly improving state of the radeon drivers, there's a probability they might start shipping stuff to you again sometime in the future. I don't expect they'll do it for this generation of GPUs but maybe next year's models.

[d2kx]

Just switched back to opensource radeon after seeing how much better Team Fortress 2 runs with them compared with Catalyst, it's not even funny (HD 6870).

[sonnet]

Just switched back to opensource radeon after seeing how much better Team Fortress 2 runs with them compared with Catalyst, it's not even funny (HD 6870).

I don't know what you mean, and I don't get criticism about Catallyst performances with games.
Either you are lying, or Phoronix results are completely false.

The R9 270x gives, in Windows, about 75% of the performance compared to the gtx 680.

http://www.techpowerup.com/reviews/AMD/R9_270X/26.html

Comparing the results just posted by Michael with the latest ones about the GTX 680, the R9 270x gives about the 80% or more, the performance of the GTX680.

http://www.phoronix.com/scan.php?page=article&item=amd_nvidia_july2013&num=6

So the catalyst apparently have better performances than nvidia's driver in Linux with games.
And looking at a recent comparison of the nvidia drivers with linux and windows, nvidia drivers performs very well in linux.

So all in all, either you or Michael's results are lying.

[rohcQaH]

So all in all, either you or Michael's results are lying.

Or maybe GPUs cannot just be compared by saying "GPU A is 80% as fast as GPU B in Game X". It depends on screen resolution (higher resolution favors GPUs with more memory bandwidth), quality (higher quality favors GPUs with more computing power), CPU (faster CPUs favor drivers with more CPU-heavy optimizations), thermal design of the computer (good thermal design allows the GPU to run at full boost) and so on.

If one benchmark says 75% and the other 80%, that's normal, even when using the same game.



Michael, have you tried the open source drivers? It's unlikely that a card with a new chip (Cura?ao or Hawaii) works, but.. well.. maybe it does?

[sonnet]

Or maybe GPUs cannot just be compared by saying "GPU A is 80% as fast as GPU B in Game X". It depends on screen resolution (higher resolution favors GPUs with more memory bandwidth), quality (higher quality favors GPUs with more computing power), CPU (faster CPUs favor drivers with more CPU-heavy optimizations), thermal design of the computer (good thermal design allows the GPU to run at full boost) and so on.
If one benchmark says 75% and the other 80%, that's normal, even when using the same game.
Michael, have you tried the open source drivers? It's unlikely that a card with a new chip (Cura?ao or Hawaii) works, but.. well.. maybe it does?

1- I considered only fhd resolutions, or 2560 one.Cpu was the same for both cards, unde linux. Since I considered high resolutions only, by your reasoning
GTX 680, should be favoured (since it's the one with more computing power and bandwidth).

2-No, your arguments are moot. While differences might arise, when the same trend is confirmed by 10 or more different test in both platforms,
I guess it's more than safe and logic to get an idea of what's going on. There's no discrepancy in mor ethan 10 test that could even potentially hint what you're saying.
The guy above said that the Catalyst have crappy performances, while tests say a different thing, and ironically shows even better performances than Nvidia drivers,
comparatively.
So either Michael's results are false or Catalyst offers good 3d performances in games.

[Calinou]

1- I considered only fhd resolutions, or 2560 one.Cpu was the same for both cards, unde linux. Since I considered high resolutions only, by your reasoning
GTX 680, should be favoured (since it's the one with more computing power and bandwidth).

2-No, your arguments are moot. While differences might arise, when the same trend is confirmed by 10 or more different test in both platforms,
I guess it's more than safe and logic to get an idea of what's going on. There's no discrepancy in mor ethan 10 test that could even potentially hint what you're saying.
The guy above said that the Catalyst have crappy performances, while tests say a different thing, and ironically shows even better performances than Nvidia drivers,
comparatively.
So either Michael's results are false or Catalyst offers good 3d performances in games.

A lot of people seem to have problems with Catalyst + TF2, it's probably a special case.

[Gps4l]

The guy above said that the Catalyst have crappy performances, while tests say a different thing, and ironically shows even better performances than Nvidia drivers,
comparatively.
So either Michael's results are false or Catalyst offers good 3d performances in games.

As a user of catalyst on linux, and windows, I have seen the Linux driver improve allot, over the last year.

It took them a while, but to name one game Serious Sam 3 now runs as good on Linux, as on Windows under DirectX.

My card is however not very new..... HD5750 1 gig ram.

What I am trying to say, the drivers were bad before, but have improved allot.

I am very sceptic, about this also happened without steam for Linux, but at the moment I can't complain.

I do not know about newer hardware, but the Catalyst linux drivers are close to or maybe a bit faster with some games, compared to windows.
Also outside games.

Finally just as good as that other os. dual boot pc. openSUSE 12.3 64 bit - W7 pro 32 bit.

[arstyl]

I don't know what you mean, and I don't get criticism about Catallyst performances with games.
Either you are lying, or Phoronix results are completely false.

The R9 270x gives, in Windows, about 75% of the performance compared to the gtx 680.

http://www.techpowerup.com/reviews/AMD/R9_270X/26.html

Comparing the results just posted by Michael with the latest ones about the GTX 680, the R9 270x gives about the 80% or more, the performance of the GTX680.

http://www.phoronix.com/scan.php?page=article&item=amd_nvidia_july2013&num=6

So the catalyst apparently have better performances than nvidia's driver in Linux with games.
And looking at a recent comparison of the nvidia drivers with linux and windows, nvidia drivers performs very well in linux.

So all in all, either you or Michael's results are lying.

I'm an ATI guy, through and through. I buy ATi cards, gifted a laptop with A10 chipset to my fiancee
and my next desktop will most likely be AMD/ATi powered (unless they somehow botch Linux support).

L4D2 lagged and flunctuated fps-wise with Catalyst 13.8 on Ubuntu 13.10, whereas open-source drivers provide me with a more stable
and enjoyable experience. I have no reason to lie.

[tom.higgy]

I don't know what you mean, and I don't get criticism about Catallyst performances with games.
Either you are lying, or Phoronix results are completely false.

I think it's not about FPS as such but the performance overall. For example, Portal with Catalyst suffered terrible input lag making it practically unplayable, which is a feat for a game like Portal. With Mesa 10 it's fine even at high quality settings on my A8-3850.

I realise the lag issue was resolved a while ago but I also experienced a number of other issues with Catalyst such that I'd rather stick with the open-source drivers.