TheBoss has done his initial port reports and such so it's my turn to feed you some information. I'm once again putting my GTX 760 against the R7 370 to see what kind of performance we can expect from Company of Heroes 2.
Let's start with the Nvidia results since Nvidia is officially supported and Feral recommends 760 or higher for Company of Heroes 2. The game doesn't have setting presets so I tested on three different setting combinations: one with everything on minimum, one where settings were on medium and one where everything was maxed out. Anti-aliasing was disabled in all three test cases. Keep in mind that the performance test is a stress test meant to demonstrate the worst case scenario. Normally the game will run better than it does in these tests.
The testing was conducted on the same system with Intel i5-2500k clocked at 3.3 GHZ and 8GB of RAM.
The results look quite reasonable considering Liam's benchmarking on his 970. Average framerates stay above 30 though the game doesn't reach a stable 60 FPS ever. It might quickly touch the 60 FPS line but it will quickly go back to around the averages. For an RTS framerates like these are fairly reasonable but they are neither mind-blowing nor something to brag about.
Let's move on to the other contender of this fine duel: the R7 370 with 4GB of VRAM. It's worth noting that Feral doesn't officially support AMD GPUs for Company of Heroes 2 and the game will even have a pop up telling you so. I tried using the Catalyst 15.7 driver but it's been broken for my distribution (Xubuntu 15.04) for a while now and I couldn't utilize it. 15.3 worked but proved to work very poorly. The performance was unplayable levels of bad and it even crashed my OS when I attempted to play the campaign. Thus I was left with the RadeonSI driver. Surprisingly enough, RadeonSI was nearly 2x faster than Catalyst in the performance test. Testing was done with the latest Mesa drivers from the Oibaf repository.
So there are the results from the R7 370 with RadeonSI. You don't need to be a genius to see that something is very wrong here. I'm also feeling a slight déjà vu…
Yeah, like I thought. Just like with Shadow of Mordor on AMD, the performance drop-off is very minimal but the performance itself is quite poor. The game reaches a maximum of 30 FPS but most of the time it runs worse than that. Averages are quite horrifying to look at with them being around 20 FPS. However, like I already mentioned, the game does run better in the campaign. There I was able to maintain a playable 30 FPS or thereabouts most of the time. That is, until the game froze and locked up. On Nvidia side the game is quite robust and stable but the AMD side saw crashes nearly constantly. The game simply locks up and will have to be forced to close. On Catalyst similar things happened but there instead of just the game locking up, my whole system froze.
It seems that Feral's latest port is once again quite exclusive to the Nvidia-using population of Linux gamers. Intel Iris Pro is mentioned as the minimum on the Steam store page but I doubt that setup would run this game at a reasonable speed. Unless something drastic has happened, an Iris Pro is slower than my R7 370 and both are running on Mesa, though they do have their differences. I was hoping that Iris Pro being there would mean wider Mesa-compatibility but that doesn't seem to be the case. Well, maybe the next port will work better. For now I recommend the AMD users to look elsewhere for their entertainment. Nvidia people are better off but the game could still be too cinematic for those that demand constant 60 FPS output. Article taken from GamingOnLinux.com.
Let's start with the Nvidia results since Nvidia is officially supported and Feral recommends 760 or higher for Company of Heroes 2. The game doesn't have setting presets so I tested on three different setting combinations: one with everything on minimum, one where settings were on medium and one where everything was maxed out. Anti-aliasing was disabled in all three test cases. Keep in mind that the performance test is a stress test meant to demonstrate the worst case scenario. Normally the game will run better than it does in these tests.
The testing was conducted on the same system with Intel i5-2500k clocked at 3.3 GHZ and 8GB of RAM.
The results look quite reasonable considering Liam's benchmarking on his 970. Average framerates stay above 30 though the game doesn't reach a stable 60 FPS ever. It might quickly touch the 60 FPS line but it will quickly go back to around the averages. For an RTS framerates like these are fairly reasonable but they are neither mind-blowing nor something to brag about.
Let's move on to the other contender of this fine duel: the R7 370 with 4GB of VRAM. It's worth noting that Feral doesn't officially support AMD GPUs for Company of Heroes 2 and the game will even have a pop up telling you so. I tried using the Catalyst 15.7 driver but it's been broken for my distribution (Xubuntu 15.04) for a while now and I couldn't utilize it. 15.3 worked but proved to work very poorly. The performance was unplayable levels of bad and it even crashed my OS when I attempted to play the campaign. Thus I was left with the RadeonSI driver. Surprisingly enough, RadeonSI was nearly 2x faster than Catalyst in the performance test. Testing was done with the latest Mesa drivers from the Oibaf repository.
So there are the results from the R7 370 with RadeonSI. You don't need to be a genius to see that something is very wrong here. I'm also feeling a slight déjà vu…
Yeah, like I thought. Just like with Shadow of Mordor on AMD, the performance drop-off is very minimal but the performance itself is quite poor. The game reaches a maximum of 30 FPS but most of the time it runs worse than that. Averages are quite horrifying to look at with them being around 20 FPS. However, like I already mentioned, the game does run better in the campaign. There I was able to maintain a playable 30 FPS or thereabouts most of the time. That is, until the game froze and locked up. On Nvidia side the game is quite robust and stable but the AMD side saw crashes nearly constantly. The game simply locks up and will have to be forced to close. On Catalyst similar things happened but there instead of just the game locking up, my whole system froze.
It seems that Feral's latest port is once again quite exclusive to the Nvidia-using population of Linux gamers. Intel Iris Pro is mentioned as the minimum on the Steam store page but I doubt that setup would run this game at a reasonable speed. Unless something drastic has happened, an Iris Pro is slower than my R7 370 and both are running on Mesa, though they do have their differences. I was hoping that Iris Pro being there would mean wider Mesa-compatibility but that doesn't seem to be the case. Well, maybe the next port will work better. For now I recommend the AMD users to look elsewhere for their entertainment. Nvidia people are better off but the game could still be too cinematic for those that demand constant 60 FPS output. Article taken from GamingOnLinux.com.
0 Likes
Some you may have missed, popular articles from the last month:
The comments on this article are closed.
I tried to run the game on my MSI R9 280 3G and Catalyst 15.2 drivers. The game started and I was able to play for a minute before it froze my system.
0 Likes
Quoting: GuestJust a note: I have this game running with a radeonHD 6950, and Mesa 10.6 (I'll try 11 soon enough). I say running. It displays things. At similar frame rates to above. I'll look into it more tonight.
Does Mesa 11 give us something (same GPU here)? I mean the R600 doesn't implement any new features, does it? Oh, maybe some minor things that are not yet in Mesa 10.6, but nothing major. That's all just in RadeonSI :(
0 Likes
Mountain Man Aug 28, 2015
Performance with the 760 looks pretty darn good to me. It's not the fastest card on the market, but I've found it a very capable performer. I don't see myself needing to upgrade for at least another year.
1 Likes, Who?
Quoting: Mountain ManPerformance with the 760 looks pretty darn good to me. It's not the fastest card on the market, but I've found it a very capable performer. I don't see myself needing to upgrade for at least another year.
A wise decision. If I was in your shoes, I'd wait until the new generation of cards after the 900 series. A new die shrink and HBM are already more compelling reasons to wait rather than jump ship to 900 series from 700 series.
Samsai, thanks for the performance comparison. From AMD's Bridgman comment on Phoronix, I would have expected the RadeonSI driver to work better but it looks like there's still performance and stability issues (I was kind of hoping for a cool miracle too, but alas).
Would you say the CPU usage when using Nvidia blob VS RadeonSI was similar? I'm curious.
EDIT: No response? I was truly curious. Guess I'll have to try and find out myself.
Last edited by Sabun on 29 August 2015 at 11:27 am UTC
0 Likes
Personally I really think Feral just fucked up on this one. I think they only ever tested this on nvidia to begin with.
I've already developed a 3D-game engine for an university course once and albeit it was of course on a much smaller scale we had issues too. All students working on the same engine only had access to AMD cards and we always had problems with the nvidia ones (where we needed to present our work).
And that was although we were sticking to standard OpenGL... but by no means we were experts and some things just behave differently in a state-machine on different hardware.
Anyway, what I want to say is: I too think that the open source drivers (e.g. RadeonSI) are much stricter than the nvidia-binary-blob. So if you fuck up, it will show.
And boy, you can fuck up in OpenGL so easily. (I have no experience with DX ;))
I've already developed a 3D-game engine for an university course once and albeit it was of course on a much smaller scale we had issues too. All students working on the same engine only had access to AMD cards and we always had problems with the nvidia ones (where we needed to present our work).
And that was although we were sticking to standard OpenGL... but by no means we were experts and some things just behave differently in a state-machine on different hardware.
Anyway, what I want to say is: I too think that the open source drivers (e.g. RadeonSI) are much stricter than the nvidia-binary-blob. So if you fuck up, it will show.
And boy, you can fuck up in OpenGL so easily. (I have no experience with DX ;))
1 Likes, Who?
R7 260x with mesa 11.0 somewhat playable framerate of 20-30. I haven't experienced any crashes after 3 scenarios and one AI scirmish match.
0 Likes
Quoting: GuestQuoting: jamespelleranoI tried to run the game on my MSI R9 280 3G and Catalyst 15.2 drivers. The game started and I was able to play for a minute before it froze my system.
You should really update those drivers.
Updated to 15.7 and the same thing happened...
0 Likes
Quoting: SabunQuoting: Mountain ManPerformance with the 760 looks pretty darn good to me. It's not the fastest card on the market, but I've found it a very capable performer. I don't see myself needing to upgrade for at least another year.
A wise decision. If I was in your shoes, I'd wait until the new generation of cards after the 900 series. A new die shrink and HBM are already more compelling reasons to wait rather than jump ship to 900 series from 700 series.
Samsai, thanks for the performance comparison. From AMD's Bridgman comment on Phoronix, I would have expected the RadeonSI driver to work better but it looks like there's still performance and stability issues (I was kind of hoping for a cool miracle too, but alas).
Would you say the CPU usage when using Nvidia blob VS RadeonSI was similar? I'm curious.
I beg to disagree - the benchamarks I've see show that the Fury's 4 gigs of HBM are not always enough at 4K. For all it's bandwidth it get's outperformed by the GDDR5 Nvidias because when you need 6GB for ultra HD textures - you need 6GB, not 4 :))). Don't get me wrong - stacked memory is a good thing and will enable single card 4K gaming, eventually, but the GPUs themselves are still the bottleneck. My prediction is that will be the case with at least the next generation of cards. Plus - there is always some better hardware just a few months away.
0 Likes
On the topic - as the proud owner of two 290x's set in Crossfire, you can understand me pain when I see them outperformed by something like a 760 (although I imagine a 750 too). However, in this case, I find it highly suspicious that it's not just the AMD driver. They do not officially support AMD hardware and obviously the game was never optimized for it. I'm not blaming Feral (too much) - we're less than 20% of the overall GPU market and probably much less among Linux users.
1 Likes, Who?
ZekThePenguin Aug 29, 2015
Yeah.... I'm guessing this is more of a driver issue than a hardware issue. Sad because AMD has fallen short on drivers for a while now. (They make nice hardware, but if a system can't use it, what's the point?)
I'm using an A10-7850k for now. I like AMD, but they really need to step up their game when an old low-end Nvidia card beats their brand new one. (Again, probably not because of the hardware.)
I'm using an A10-7850k for now. I like AMD, but they really need to step up their game when an old low-end Nvidia card beats their brand new one. (Again, probably not because of the hardware.)
1 Likes, Who?
Patreon
PayPal
See more from me