Company of Heroes 2 Tested, Benchmarked

By on July 2, 2013, 10:01 PM

It's hard to believe nearly seven years have passed since Relic Entertainment and THQ released Company of Heroes, a World War II-themed real-time strategy title that impressed the heck out of me and many other RTS buffs at the time. I even invested in the two expansions that followed, and have featured the game across plenty of CPU and GPU reviews as it's an excellent benchmark for gauging the performance of both components.

Like the original, Company of Heroes 2 is set in World War II but focuses on the Eastern Front. It's built on Relic's proprietary Essence 3.0 engine, which will supposedly help bring "new technological advancements" to the game. CoH 2 takes advantage of DirectX 11 but also supports DirectX 9 and its engine has a mess load of new features such as ColdTech, a dynamic system for effects such as snow accumulating against buildings and vehicles, restricting movement depending on the depth of the snow, and snow tracks left by units.

This extreme level of detail should be taxing all around with heavy loads on both the CPU and GPU, so we hope Company of Heroes 2 will serve as a fine PC benchmark in our reviews for years to come. In the meantime, let's see how Relic's latest game manages to tackle today's hardware.

Read the complete article.




User Comments: 41

Got something to say? Post a comment
EEatGDL said:

I hope no feelings get hurt here, but... complementing Steve's last sentence: the graphics quality observed don't appear to match the workload and the frames. Who's fault here? Graphics drivers, poorly optimized engine? Because really it looks a lot like World in Conflict, maybe sub-par and that game is not demandant for these cards, I don't think DX11 simply justifies it.

And on a side-note, just as Steve noted before: an eight-core Bulldozer OCed @4.5 GHz is below a quad-core/8 threads i7 Haswell @2.5 GHz in this benchmark as observed before in other games -not all, but is a repeated scenario. The odd behaving of the i7 frame caping means the system bandwidth limit was reached and couldn't give even 1 more frame? Cause 41 is pretty low for OCing it up to 4.5 GHz and having no difference from the 2.5 GHz test.

3 people like this | JC713 JC713 said:

Horribly optimized game...

1 person liked this | St1ckM4n St1ckM4n said:

Haha, awesome. Such terrible coding... I remember Magicka being like this when it was first released!

TheinsanegamerN TheinsanegamerN said:

I hope no feelings get hurt here, but... complementing Steve's last sentence: the graphics quality observed don't appear to match the workload and the frames. Who's fault here? Graphics drivers, poorly optimized engine? Because really it looks a lot like World in Conflict, maybe sub-par and that game is not demandant for these cards, I don't think DX11 simply justifies it.

And on a side-note, just as Steve noted before: an eight-core Bulldozer OCed @4.5 GHz is below a quad-core/8 threads i7 Haswell @2.5 GHz in this benchmark as observed before in other games -not all, but is a repeated scenario. The odd behaving of the i7 frame caping means the system bandwidth limit was reached and couldn't give even 1 more frame? Cause 41 is pretty low for OCing it up to 4.5 GHz and having no difference from the 2.5 GHz test.

I believe that was the gpu being a limiting factor there. with medium details, the framerate hit 65. although, in both cases, I smell a horribly coded game, which is probably to blame.

Burty117 Burty117, TechSpot Chancellor, said:

Nnooo! My bro pre-ordered this for me from Christmas! I haven't had a chance to download and test it out yet but I've also purchased a 1440p screen, guess medium settings for me that sucks big time, to be fair the first one was a pig to run when it first came out and got a little better with a lot of patches and updates.

Maybe they'll do the same here?

Qrox Qrox said:

Horribly optimized game...

Was just thinking the exact same thing. If even the Titan gets that low FPS, then it can't be the hardware setup thats killing the frames. For how the game looks I would expect the frames to be at least double the frames that they got in the test.

smallfield said:

One more reason for me not to get this title.

Now, back to CoH1 & Blitzkrieg mod.

1 person liked this | Guest said:

Seems like a badly optimized game when even the top end cards die at medium settings 1080/1200p. Hopefully new patches and drivers will bring more performance to the table. Seems like CPU is only the limit at medium settings.

falkonIndian said:

Can someone please explain why this is - I've experienced lowest framerates always when I play RTS titles as opposed to any other type of game. Is this because of number of units on screen at a time or is there something else about RTS games that make them the most power hungry?

On my first computer, I played quake 3 at the highest settings - but age of empires 2 brought it to its knees

On my current two year old computer, skyrim is maxed out as is witcher 2 but shogun 2 makes it struggle at high settings ans resolution

What is it about RTS games really?

BigMack70 said:

Interesting results... not sure if the results are comparable or not but guru3d got much better performance in this game (~48 fps max settings 1440p on a 780/Titan)

Puiu Puiu said:

Well... this was really unexpected. even if they manage through a miracle to squeeze 10-20% more performance out of the GPU with better drivers from AMD and NVIDIA, they can't do anything about the CPU part unless they release a pretty big patch to the game that fixes the performance issues.

Puiu Puiu said:

Can someone please explain why this is - I've experienced lowest framerates always when I play RTS titles as opposed to any other type of game. Is this because of number of units on screen at a time or is there something else about RTS games that make them the most power hungry?

On my first computer, I played quake 3 at the highest settings - but age of empires 2 brought it to its knees

On my current two year old computer, skyrim is maxed out as is witcher 2 but shogun 2 makes it struggle at high settings ans resolution

What is it about RTS games really?

RTS games have a lot of features that eat away at the CPU. they have a lot of computing to do along side the graphics part. The same happens when you use NVIDIA PHYSX on the CPU (non NVIDIA GPU).

fimbles fimbles said:

No sli = No sale sorry.

And a big thanks to jos for the info, saved me 60 bucks.

EEatGDL said:

Interesting results... not sure if the results are comparable or not but guru3d got much better performance in this game (~48 fps max settings 1440p on a 780/Titan)

Can you provide the link 'cause I don't find it in the recent posts of the corresponding sections, even got no results during the search.

1 person liked this | Blkfx1 Blkfx1 said:

Relic is just giving everyone the finger on this one.

fimbles fimbles said:

The performance of this game in comparison to the actual quality of the graphics is abysmal to say the least. Do not waste your money, Or at least download a "demo" version to see for yourself.

amstech amstech, TechSpot Enthusiast, said:

Thank you thank you thank you for including the Phenom II 720/740 X3. Thats still a widely used budget gaming chip (and usually unlocks to be a decent Deneb). Too bad it got wrecked by this game, holy hell!

On the GPU side of things, wowza! My 670 got steamrolled.

BigMack70 said:

Interesting results... not sure if the results are comparable or not but guru3d got much better performance in this game (~48 fps max settings 1440p on a 780/Titan)

Can you provide the link 'cause I don't find it in the recent posts of the corresponding sections, even got no results during the search.

I can't provide the link because it doesn't allow HTML, but they have CoH 2 in their new MSI GTX 780 Gaming review on the front page - you can see some results there. They said they have a performance review for the game forthcoming.

Fbarnett Fbarnett said:

Wow these companies just release sloppy coded junk and say take it or leave it. I am going to leave it this time

St1ckM4n St1ckM4n said:

What I find hilarious is that there are no excuses available. Sure, you could say that game developers run their test games on high-end hardware, then scale back. "Oops, we forgot to scale back!"

Only there is no SLI support and a Titan + i7 is the best you can get ....

EEatGDL said:

I can't provide the link because it doesn't allow HTML, but they have CoH 2 in their new MSI GTX 780 Gaming review on the front page - you can see some results there. They said they have a performance review for the game forthcoming.

Well, I've been scratching my head for a while and the only test difference I can come up with is they used the 3960X @4.6 GHz vs 3.3 GHz here; driver versions here were a bit more recent on AMD and used more RAM [here], but watching the performance trend with the Haswell i7 I'm not sure what would happen with the 3960X when overclocking, apparently it can successfully eliminate the bottleneck when overclocking compared to the apparently limited i7 4770K. Anyway, we can't be sure of the hypothetical scenario, they clocked the same processor by 1.3 GHz more.

Also, the first graph [comparing medium, high, higher, max quality with 2 resolutions] matches the GTX 780 Palit Super Jetstream model with gives a little more than the standard model and wasn't used here either.

GhostRyder GhostRyder said:

Honestly, reading the benchmark results it looks to me from what I at least gathered a couple of things:

1: The game will only utilize 4 cores max no matter how many are present (Being that the FX chip did so poorly and even the I7 3930k was above the 3960X) and has extreme bottlenecking issues present

2: Since it does not utilize SLI/CFX, that's going to mean besides apparently a heavily overclocked I7 4770k and a Titan, that game cannot be played very effectively at 1920x1200 (Well it gets around 65FPS, but still)

So to me, it looks like the most efficient price to performance system to play this game would be a I7 4770k and a HD 7970ghz edition (or a GTX 780, but that's almost twice the price of a HD 7970 for only a few frames difference). That really limits the user base and shows as people have mentioned above, an engine and game that does not utilize the hardware handed to it properly which is going to be tough on most people.

EEatGDL said:

I'm more concerned about what @BigMack70 asked about the huge difference between this set results and Guru 3d's. Here are the system specs of their test. Can anyone tell a reasonable explanation?

JC713 JC713 said:

Was just thinking the exact same thing. If even the Titan gets that low FPS, then it can't be the hardware setup thats killing the frames. For how the game looks I would expect the frames to be at least double the frames that they got in the test.

Yeah, hopefully drivers can fix this, but probably not. Maybe after a few patches the FPS issues will be resolved. This same issue also occurred with the new Sim city.

Guest said:

I'm sorry but this is a awful benchmark. You do realize that setting Anti-Aliasing to anything other than Low uses SSAA right? It doubles your resolution then scales it down, pretty much rendering TWO images at once.

Set AA to low, max everything else out and all these high end cards will run the game no problem. You guys are incredibly unprofessional when it comes to your benchmarks.

Guest said:

The last poster (@Guest) is spot on ... and very diplomatic for using the word unprofessional rather than many others they could have chosen!

I read this and thought that I was going to be in trouble, as I'm running with a stock HD 6970 and stock i7 950 @ 3.07Ghz. The stats show that when running at maximum I should expect 17 fps based on 1920 x 1200. Then I also realised that they had also set AA to High! SSAA (Super Sampling) does exactly what the previous poster says; uses the highest quality sample and then reduces it down to size.

I actually have a 1920 x 1200 display as well, so I fit the bill perfectly in terms of these benchmarks. I consistently got 28 fps when running the CoH 2 benchmark. I had hoped for more, but that's fine for playing the game properly. I've never seen a benchmark go straight in there for Highest AA when showing comparisons! Given the way that the test has been constructed, that's like rendering Jurassic Park in real-time on your gaming rig!

The discovery of 4 cores being used as a maximum is useful and important, but I'm afraid that the overriding GPU tests are flawed because a lack of understanding on the AA front.

St1ckM4n St1ckM4n said:

A fix could be detailed here:

[link]

Would be interesting to see HDD usage which gaming.

Guest said:

I too was quite surprised to see that the review didn't really get into the details of the graphics settings. (Did they even tell precisely which settings were used?)

By playing around a few minutes I quickly found out that AA was the key to OK frame rates.

On my i5 750 @ 4Ghz, 8GB DDR3 1600, 7950 @ 1050 (RAM stock 1250) Low AA is essentially free but going just to medium had a huge impact on frames:

No AA: 55.7 - Low AA: 54.5 - Medium AA: 41.2 - High AA: 30.3

Other than that, snow detail is quite expensive as well.

But yes, CPU speed IS very important. Can't remember frames at the 750's stock speed @ 2.66, but they were quite a bit lower.

PS. I am not the "original" guest poster :-)

Guest said:

Guest "2" here again. Just to elaborate further on the AA thing:

Actually the most important bench on RTS is in my opinion lowest frame rates which on my system were

35.8 on No AA

35.6 on Low AA

22.3 on Medium AA

13.2 on high AA

That's almost cutting frame rates in half every time you raise the AA setting.....

Guest said:

According to semiaccurate.com the included benchmark is crap and shows much lower frame rates than what to expect during normal game play though out the game.

odd move of relic fooling their customers

@mr. techspot: better redo all tetsting again using fraps;)

popeye6212 said:

YOU CAN PLAY THIS GAME ON AN I3 All day long. I know cause I have. 3.4ghz i3 with gtx ram video card, 6GB ram 64bit windows 8. I play at 1920x1080 everything set to low (not minimum) and it runs nice and smooth even in 8 player matchs online. The graphics still look great. IDK why everyone is pissed, they made a game that will get better looking as technology improves. I hate when I buy a game and I can just max it out. Also the built in benchmark is supposed to demonstrate worst case scenario, and your actual matchs will almost always have better frame rates. Relic built it as a stress test. Yesh the coding in the game isn't great but you definitly don't need an i7 and a Titan to enjoy the game with acceptable frame rates.

popeye6212 said:

*gtx 460 (1GB vram)

GhostRyder GhostRyder said:

YOU CAN PLAY THIS GAME ON AN I3 All day long. I know cause I have. 3.4ghz i3 with gtx ram video card, 6GB ram 64bit windows 8. I play at 1920x1080 everything set to low (not minimum) and it runs nice and smooth even in 8 player matchs online. The graphics still look great. IDK why everyone is pissed, they made a game that will get better looking as technology improves. I hate when I buy a game and I can just max it out. Also the built in benchmark is supposed to demonstrate worst case scenario, and your actual matchs will almost always have better frame rates. Relic built it as a stress test. Yesh the coding in the game isn't great but you definitly don't need an i7 and a Titan to enjoy the game with acceptable frame rates.

Your missing the point, this is not a game like crisis where it was due to the high demands of the game that you cant play it, while this game is demanding, its inefficient.

It only truly utilizes 2 Core when 4 cores is the standard in todays games and programs which heavily limits is usage.

It does poor to no support in SLI/CFX meaning that your forced to 1 GPU to actually use the game (in which even the titan does have issues)

The game while looking decent does not look like a game deserving of need this much power and that's the main issue.

You maybe satisfied with running a game on low, but not everyone things that way, I for instance like at least high-Ultra and im not the only one who thinks it. The fact is, the developers were lazy at making this game efficient for everyone to use and expand on, the fact that even overclocking hits a point where the game stops scaling shows this as well which is very disheartening to many of us who have FX or i7 Processors heavily overclocked expecting higher FPS.

popeye6212 said:

Here are some benchmarks I did on friends machines and mine.

Cpu: I3 2130 @ 3.4Ghz (no OC)

Video: GTX 460

Ram: 6GB

OS: Windows 8 64bit

Settings are:

1920X1080

Gameplay Resolution: 100% (Max setting)

Image Quality: Low

AA: Off

V sync: On

Snow Detail: Low

Texture Detail: Low

Physics: High (Max setting)

Avg Frame Rate: 31.24

These are not the lowest settings. Image quality can be set to minimum, and game play resolution can be set down to 33%.

My brother rocks a Core 2 quad and a GTX 470 these are his settings and results:

Cpu: Core 2 Quad Q9550 @ 2.83Ghz (no OC)

Video: GTX 470

Ram: 4GB DDR3

OS: Windows 7 64bit

1920X1080

Gameplay Resolution: 100% (Max setting)

Image Quality: Medium

AA: Off

V sync: On

Texure Detail: High

Snow Detail: Medium

Physics: High (Max setting)

Average Frame Rate: 30.58

The graphics look very good and the gameplay is excellent and that's on a core 2 quad with him running a mix of medium and high settings

The way they executed the benchmarks for this game is ridiculous, Who would turn there AA to medium when there graphic textures are only set to medium?

Here's on my I5, with my new R9 280x, which is basically a rebadged 7970 GHZ Edition

Cpu: I5 3570k @ 3.4ghz (no OC)

Video: R9 280x

Ram: 16GB

OS:Windows 7 64bit

Benchmark Settings

1920X1080 (they did 1920x1200)

Gameplay Resolution: 100% (Max setting)

Image Quality: Maximum

AA: High (Max setting)

V sync: Off

Texure Detail: Maximum

Snow Detail: High (Max setting)

Physics: High (Max setting)

Frame Rate Avg: 36.92

I Did about 4 frames better than their 7970Ghz Edition and this is possibly due to my resolution being 1920X1080 instead of their 1920x1200.

Benched it again Everything set to max but AA set to low

Benchmark Settings:

1920X1080

Gameplay Resolution: 100% (Max setting)

Image Quality: Maximum

AA: Low

V sync: Off

Texure Detail: Maximum

Snow Detail: High (Max setting)

Physics: High (Max setting)

Average Frame Rate: 54.56

At these setting during gameplay FRAPS generally shows my fps between 70 and 80. Occasionally it'll drop down to 50s during intense action but I can't really perceive the difference.

Anyway this game is quite playable on hardware that is lower than anything they even tested. I think the Core 2 quad system with the GTX 470 is a great example of a somewhat older game computer. The fact that it runs a mix of medium, high and maximum settings shows that this game isn't as inefficient as there benchmarks make it appear. If you got to have max everything including AA and 60fps better go get yourself a TItan. If you're more interested in playing the game than setting everything to max then you should have no problem on any recent machine with a half decent video card.

Guest said:

@Guest#2

Low AA is FXAA, aka poor man's AA. It's cheap, but makes textures blurry, so best avoided, if possible.

Mid AA is 2xSSAA (Super-Sampled Anti-Aliasing), which draws every polygon twice, hence increasing the workload significantly.

High AA is 4xSSAA which is twice the work of 2xSSAA.

The reason this game turned out the way it did was because the developer's (Relic's) parent company THQ had gone bankrupt at a very inconvenient time, and Relic, like all of THQ's assets, was up for bidding and was eventually bought by SAGA. There was a time of several weeks during which the Relic staff probably had no idea what was going to happen to their jobs. Perhaps SAGA should have given them some more time to adjust to the change.

Adder said:

I'm confused. I just purchased an MSI i7-3630qm 2.40 / 8gb ram / geforce 660m / w8

and the frame rate on CoH2 (with everything set to low/off) is so abysmally slow that something seems wrong - I'm talking maybe 10 fps. Would anyone be able to shed some light on this for me ?

Puiu Puiu said:

I'm confused. I just purchased an MSI i7-3630qm 2.40 / 8gb ram / geforce 660m / w8

and the frame rate on CoH2 (with everything set to low/off) is so abysmally slow that something seems wrong - I'm talking maybe 10 fps. Would anyone be able to shed some light on this for me ?

what resolution? 1080p is not easy for a laptop depending on the game.

Adder said:

Confusingly when I open the options panel within the main menu the resolution box is empty, so I'm unable to either see or change it. I guess it's running in whatever default mode is set by the game.

I've updated my drivers. I see that other players are getting at least 30 fps with older systems, so it leaves me scratching my head.

I appreciate the help Puiu

GhostRyder GhostRyder said:

Confusingly when I open the options panel within the main menu the resolution box is empty, so I'm unable to either see or change it. I guess it's running in whatever default mode is set by the game.

I've updated my drivers. I see that other players are getting at least 30 fps with older systems, so it leaves me scratching my head.

I appreciate the help Puiu

Make sure the game is running in full screen mode might be a good start and also that its actually defaulting to the Nividia Driver. I have had an issue before on my MSI GT70 in CoH2 where it for wahtever reason was not defaulting to the GTX 675m and instead was running on the HD 4000 on-board GPU. MSI afterburner can also help shed some light if you want to check your in-game FPS while playing and it will also display if your GPU is being used correctly.

Guest said:

I've installed Afterburner and it looks like everything is okay.

I'm not sure how I can dictate that the game launches uses my geforce though - there are a couple of config files in the game folder but whenever I tweak values and try to save, it tells me 'denied'. Without that, I can't alter the resolution or anything else. I can't believe I haven't figured this out yet, man I'm so out of touch. Take me back to the mid 90's.

Puiu Puiu said:

Try some compatibility modes. also check for bios updates.

Load all comments...

Add New Comment

TechSpot Members
Login or sign up for free,
it takes about 30 seconds.
You may also...
Get complete access to the TechSpot community. Join thousands of technology enthusiasts that contribute and share knowledge in our forum. Get a private inbox, upload your own photo gallery and more.