Shadow of Mordor Performance: Windows 10 vs. Ubuntu Linux

[grenadecx]

So, I've done some extending benchmarking myself.

First I just wanna state that the benchmark is quite demanding compared to the real gameplay. Which is a good thing, but just so you know.

My Specs:
OS: Ubuntu 15.04
CPU: Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz
GPU: Nvidia Geforce 780 (3GB Video RAM)
Kernel: 3.19.0-25-generic (x86_64)
Desktop managers: (Unity (compiz, with unredirect fullscreen windows) & Fluxbox (no composition) )
Resolution: 1920x1080

Low:
---------------
Min: 52
Average: 98
Max: 171

Medium:
---------------
Min: 43
Average: 81
Max: 125

High:
---------------
Min: 40
Average: 72
Max: 106

Ultra:
---------------
Min: 29
Average: 58
Max: 87

---------------------------------

So... I tried this with both Unity ---> AND <--- Fluxbox, which gave me more or less the exact same FPS.
I debunk that Unity gives you less FPS - At least for this game.

Now, some people are looking at the benchmarks and think this is really bad compared to windows, and yes, it might be. But it's very very very playable with decent +70fps on high/ultra settings if you disable Motion blur/AA, on my system.

Some other minor things. High textures states to take up 4GB Video RAM on HIGH settings. While ULTRA requires 6GB Video RAM. I really don't think that's is justified, at all. But it's the same with the windows version of the game. I only have 3GB Video RAM.

Some have mentioned Borderlands 2 and The Witcher 2 in regards to this port. This port spanks those games so fucking hard. I tried The Witcher 2 port yesterday, since so many people said it had gotten way better. What a load of bullshit. I can't even run that game on medium settings without AA 60fps. And Borderlands are so bad in comparison as well especially compared with the Windows version of the game.

All in all, I think Feral have done a really great job with the port even if the benchmarking comes up shy compared with windows. And looking at Michaels benchmarking, I think they are quite accurate.

Edit:
For the record, it does look like one cpu thread is maxed most of the given time. Sometimes it changes thread, but it looks maxed out. So in regards to the port compared to windows, that could be one of the reasons for the huge fps difference unless it looks the same on Windows. Can't really confirm it however since I don't feel like downloading another 40GB for Windows.

[gamerk2]

From what I recall, SoM scaled well on my 2600k on Windows; I'll try and remember to check sometime this weekend.

That being said, the cause for the performance difference should be investigated and understood.

[BdMdesigN]

Hello @ll,

At the first most users on steam reports, thats SoM run only on ONE CPU Core and this was INTEL CPU owners.
I own a AMD Phenom II X4 960T (unlocked to X6 1600T) @ 3,8 GHz and the game use ALL cores.

SoM use on Windows 10 ALL Cores on ALL CPU's, thats the reason why you get this result in this benchmark.

So it's not a fault from NVIDIA. AMD GPU users will have this problem, too.

But let us see wahts the next patch brings, this patch will includes the AMD Athlon II/Phenom II start problem.

[hellslinger]

Grenadecx,

Ahh, good to know. The last version of Unity I gamed under was the one on 14.04. As I remember, CS:GO was unplayable on that version because of stuttering and interruption, though max FPS was still high when it wasn't choking. I'm excited to try gaming in 15.04 w/Unity now!

Hello @ll,

At the first most users on steam reports, thats SoM run only on ONE CPU Core and this was INTEL CPU owners.
I own a AMD Phenom II X4 960T (unlocked to [COLOR=#000000][FONT=Verdana][B]X6 1600T) @ 3,8 GHz and the game use ALL cores.

If this is true, then yeah, I guess everyone was right that the game is effectively bound to single-core performance. I can't imagine why this is happening for only Intel chips, but there must have been a good reason the devs did it, or there's configuration bug. I hope it is true, because it means the deficit is temporary.

Grenadecx is also right that the game is more than playable (at least on our hardware). I didn't even notice the game had lesser performance than Windows, to be honest, even on my 144 Hz monitor.

[BdMdesigN]

Here you can see some ingame benchmarks from me:

[philips]

My results with Windows & SteamOS on the same setup (i7-4790K, GTX 780):

[Nobu]

FWIW, Shadow of Mordor is on sale on Steam, GOTY version is 60% off...AMD oss is at OpenGL 4.1...so close! 😭

[Svartalf]

I'm playing the wither 2 on R600 trinity. The performance is bad. I don't know why with wine is better.

This is because eON's trying to do the parallel rendering per frame that the Windows version does as part of the adaptation layer. NVidia's the only one that seems to do decently there right now. Sometimes, you can see the parallel rendering going on if you're looking at Geralt's head and see the hair render in two distinctly different threads...

WINE doesn't attempt this. It serializes all the requests as best as it can into one rendering thread within the context.

It's a mixed bag. If you've got a GL stack that will do this right, it's NOT exactly faster, but it's different. Some things do better, others not so much so. Parallel rendering's another one of those made-for Windows "solutions" that was done to compensate for a poor scheduler design. That's why some games will do better with eON, others not so much so- even on OSX because they have the same issues.

Now...what's the story with SoM? Same woes probably. In order to get the high-end stuff, they had to do parallel rendering on the machine- and in order to do it on Linux, you've got to serialize some of this which may not be quite done well enough. I expect, though, something at least as good as the PS3 playing it since it's lower-end hardware compared to the stuff Michael's evaluating this on. I'm probably going to buy it because it'll do "well" on my PC and I'd rather play this one on my PC from here on out...but it's faintly disappointing that the native port has these sorts of performance hickeys.

[Svartalf]

Hello @ll,

At the first most users on steam reports, thats SoM run only on ONE CPU Core and this was INTEL CPU owners.
I own a AMD Phenom II X4 960T (unlocked to X6 1600T) @ 3,8 GHz and the game use ALL cores.

SoM use on Windows 10 ALL Cores on ALL CPU's, thats the reason why you get this result in this benchmark.

So it's not a fault from NVIDIA. AMD GPU users will have this problem, too.

But let us see wahts the next patch brings, this patch will includes the AMD Athlon II/Phenom II start problem.

Ah...CPU issues. Yeah, you're going to see it be "off" if you're pinned to one or two cores on the CPU for that one.