ARMA III Video Card Performance and IQ Review @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,706
ARMA III Video Card Performance and IQ Review - ARMA III is our focus point for today. It features a large open world environment designed on a massive continent measuring 270 square kilometers. To go along side this massive continent is a max visibility range of 20km. Combine this with ARMA III's impressive looking graphics and we have a game that demands performance.
 
Was processor utilization measured?

Not being optimized for multi-core seems to be the issue with this game receiving low frames more so than anything else.

The frames are terrible in the game, yet it's still most favorite game at the moment.

Love it or hate it I love Sa-Matra Wasteland!
 
Thanks for the review, it's good to see games that actually push the graphics envelope on PCs.

I'll probably pick it up once I upgrade my video card to the R290x (if the card lives up to hype that is)

PS. Btw just as an idea: Perhaps you could give gold/silver/bronze awards to game engines like this that push game graphics further on the PC.
 
BurntToast said:
Was processor utilization measured?

Not being optimized for multi-core seems to be the issue with this game receiving low frames more so than anything else.

The frames are terrible in the game, yet it's still most favorite game at the moment.

Love it or hate it I love Sa-Matra Wasteland!
Click to expand...

I've looked at the core utilization this past weekend myself, I saw it loading basically all 4 cores about half way (so that means it's like having 2 cores fully loaded), the other 4 threads were lightly loaded 10% maybe (i7 with HT enabled). But I only tried this in one of the showcases, and only for a few minutes, haven't looked at it in every one for a lengthy amount of time. By this short test, it seems to load dual-core well, but isn't fully taking advantage of 4 cores, but still has some slight load on the virtual threads. I think this is overall better than Arma 2 though. Also keep in mind I was running at very GPU dependent game settings. Don't consider this a final analysis, only my brief look at it.
 
Last edited:
Well, seeing from experience and other benchmark sites, it seems that Arma III is still not all that multi-core aware, although as Brent said, it is much better than Arma II, and i would chime in saying that, well, it may not show a huge performance improvement if it was better in the multicore department as they seem to have done a great work coding the game to be gpu balanced.

All in all the Arma series is amazing in the way it flexes it's muscle with the vast view distances, and this is one of the reasons why the performance seems "poor" versus other games, but there is no way to deny that the performance hit is worth it for some of us who enjoy the great tactical aspect of properly seeing the battlefield.

Cheers!
 
I believe I found an error in the apples to apples section where the GTX 770, 7970 GE, GTX 760 and 7950 Boost were tested at 1920x1080 - 8x FSAA - ultra settings.

Error in bold and underlined:

"In this apples to test we are running ARMA 3 at 1920x1080 with 8X FSAA enabled, and overall quality set to "ultra." Alpha to coverage is enabled on all trees and grass, and we are using FXAA Ultra.

At these settings the GeForce GTX 770 performed the fastest in this group, averaging 43 FPS. This is slightly below what we considered playable. It is 12.8% slower than the GeForce GTX 780 at the same settings,
however it is 13.3% faster than the Radeon HD 7970 GE and 18.8% faster than the GeForce GTX 760."

By my calculation, the GTX 770 is 4.88% faster than the 7970 GE, not 13.3% faster.

43 fps for GTX 770 minus 41 fps for 7970 GE = 2 fps faster
2fps / 41 fps average fps for 7970 GE = .04878 x 100 = 4.88%


Looks like a pretty brutal game. Single GPUs are taking a beating. Then again, the ARMA series was never really known for being well optimized.
 
Great write up as always!

Does this title have a 64 bit client, if so, how much memory this game will gobble up?

Also, the test system has a WD Black HDD. In a game like this, doesn't this drive cause lag/slowdown during loading, which can affect performance marks?
 
Awesome review! Love to see the ArmA series getting some love.
 
I don't care what anybody says, SMAA just looks so much better than anything else. To bad it tends to blur everything but I can live with that. :)
 
pcjunkie said:
I don't care what anybody says, SMAA just looks so much better than anything else. To bad it tends to blur everything but I can live with that. :)
Click to expand...

It isn't a problem inherent to SMAA, it was a problem with the way they implemented SMAA in the game and it was noted in the beta since march 2013.

You can get much better results with an injector.
 
pcjunkie said:
I don't care what anybody says, SMAA just looks so much better than anything else. To bad it tends to blur everything but I can live with that. :)
Click to expand...

Why would you want to worsen your IQ? FXAA + MSAA does better lookin AA, and adds the sharpen filter so everything is crisp and doesn't look like you're looking through Vaseline.
 
Revdarian said:
It isn't a problem inherent to SMAA, it was a problem with the way they implemented SMAA in the game and it was noted in the beta since march 2013.

You can get much better results with an injector.
Click to expand...

Which injector?
 
SweetFX for example, google "sweetfx arma iii" and you should find some interesting links, including pictures.
 
Hello HardOCP,

thanks for great and detailed review of our game

there is just small 'detail' about SMAA so let me quote myself ;)

http://forums.bistudio.com/showthre...-GPU-testing&p=2524050&viewfull=1#post2524050

there is error in that article, SMAA don't create blur ...
choosing
no PPAA
or
any SMAA
results into no difference of textures quality (see for yourself, nothing is blurred)

the reason FXAA 'looks' crisper and clearer is because of auto-applied sharpen post process filter
which was used to combat FXAA overblurry issues ...

imho the sharpen post process is so good it deserve separate setting so you can use SMAA+sharpen filter too
(as SMAA is better than FXAA in terms of edge aliasing quality)
Click to expand...

note: there was bug in builds prior game release (beta builds) where SMAA was blurry but that was fixed since ...

anyway if you want more about Arma 3, visit us on the BIForums or slap me on skype or irc or similar ;)
 
Last edited:
Brent_Justice said:
I've looked at the core utilization this past weekend myself, I saw it loading basically all 4 cores about half way (so that means it's like having 2 cores fully loaded), the other 4 threads were lightly loaded 10% maybe (i7 with HT enabled). But I only tried this in one of the showcases, and only for a few minutes, haven't looked at it in every one for a lengthy amount of time. By this short test, it seems to load dual-core well, but isn't fully taking advantage of 4 cores, but still has some slight load on the virtual threads. I think this is overall better than Arma 2 though. Also keep in mind I was running at very GPU dependent game settings. Don't consider this a final analysis, only my brief look at it.
Click to expand...

Marek, the BI CEO reposted this blog not that long ago. It discusses multithreading within a simulation engine, and explains why you aren't pinning your CPU at 100% max with 120 fps in a simulation game.

http://www.bistudio.com/english/company/developers-blog/91-real-virtuality-going-multicore
 
Does the game utilize PhysX on the Nvidia cards thus giving better in game performance when playing?
 
Are you going to be doing a CPU architecture, core, and clock speed scaling article as well? I would be interested in that data.
 
Dangerdog said:
Does the game utilize PhysX on the Nvidia cards thus giving better in game performance when playing?
Click to expand...

It doesn't utilize hardware physX period. It doesn't have any PhysX effects that would require it. The only things simulated by the PhysX engine are vehicles and a slim count of environmental objects. None of it is demanding enough on the physx engine that it requires GPU muscle to accomplish. (Surprisingly, PhysX has turned into a pretty great CPU run physics engine over the years)
 
Correct, PhysX can be used two ways, as a general physics engine for your game, run on the CPU, or to utilize hardware acceleration via NV GPUs. This game doesn't have hardware PhysX GPU support on NV GPUs, it just uses PhysX as the physics engine on the CPU.
 
Dwarden said:
Hello HardOCP,

thanks for great and detailed review of our game

there is just small 'detail' about SMAA so let me quote myself ;)

http://forums.bistudio.com/showthre...-GPU-testing&p=2524050&viewfull=1#post2524050



note: there was bug in builds prior game release (beta builds) where SMAA was blurry but that was fixed since ...

anyway if you want more about Arma 3, visit us on the BIForums or slap me on skype or irc or similar ;)
Click to expand...

Than you for the information, very informative

Seems we hit the nail on the head r.e. FXAA sharpening filter, IMO it does make things look a lot better in this game, wish it had been enabled inside SMAA as well.
 
Quick question on gameplay , how would this game/sim compare to BF3/4 as far as realistic , meaning weapons, damage , tactics etc .

I don't have either an trying to decide between ArmA III and BF4 .

Thanks ahead and great review of graphic part .
looking at you-tub vids of both it looks like ArmA would get the node as BF4 seemed more run an gun type .
 
EdKiefer said:
Quick question on gameplay , how would this game/sim compare to BF3/4 as far as realistic , meaning weapons, damage , tactics etc .

I don't have either an trying to decide between ArmA III and BF4 .

Thanks ahead and great review of graphic part .
looking at you-tub vids of both it looks like ArmA would get the node as BF4 seemed more run an gun type .
Click to expand...

Don't expect any sort of realism from Battlefield. Every weapon is balanced against another to make sure that as you progress through the levels, each weapon is marginally more powerful than the last. But at the same time, it keeps these balance issues in mind as it applies simple things like damage, ballistics, and even recoil.

In the Arma series, these are set to real world specifications, balance be damned.

Battlefield is about getting a higher kill count.

Arma is about using whatever you have at your disposal to accomplish a mission, even if you're at a severe disadvantage due to gear, vehicles, or even intel.
 
TheCapulet said:
Don't expect any sort of realism from Battlefield. Every weapon is balanced against another to make sure that as you progress through the levels, each weapon is marginally more powerful than the last. But at the same time, it keeps these balance issues in mind as it applies simple things like damage, ballistics, and even recoil.

In the Arma series, these are set to real world specifications, balance be damned.

Battlefield is about getting a higher kill count.

Arma is about using whatever you have at your disposal to accomplish a mission, even if you're at a severe disadvantage due to gear, vehicles, or even intel.
Click to expand...
Ok, thanks for update/feedback .

I see many say Arma III is buggy but they just released patch, did this fix many issues (the changelog list of fixes is long but hard to tell impact of fixes ) .

I think i am ready to pull trigger on ArmA 3 .
 
EdKiefer said:
Ok, thanks for update/feedback .

I see many say Arma III is buggy but they just released patch, did this fix many issues (the list is log of fixes but hard to tell impact of fixes ) .

I think i am ready to pull trigger on ArmA 3 .
Click to expand...

I see many say Arma 3 is buggy too, but quite frankly, I've not once come across a show stopping bug. Or really any bugs worth mentioning for that matter.

Realistically the only bugs you'll see stem from poor mission design from other players/mission makers. You'd never see these issues in other games, because you'll never play player-made missions in other games. Where as with Arma 3, there's a good chance that's what you'll spend the majority of your time doing, even after the SP campaign is officially released from BIS. (Nearly all of the multiplayer games you'll play are player made. Some are terrible, other's are great like "BECTI", or "Conqu3st". (both professional quality gametypes/missions)

Whenever you hear someone shouting "ARMA 3 is buggy!!!!11one", take it with a grain of salt. Because chances are, they don't have a single clue about what they're shouting.
 
Fxaa looks like awful in this game the filter adds grain and odd offset highlights like cell shading to the textures.
 
TheCapulet said:
I see many say Arma 3 is buggy too, but quite ....

Whenever you hear someone shouting "ARMA 3 is buggy!!!!11one", take it with a grain of salt. Because chances are, they don't have a single clue about what they're shouting.
Click to expand...

Or they know exactly what they're talking about. The bugs in this game are there and vast. Every day a large (sometimes 300+megs) dev patch goes out to the dev branch. Things like placeholder models or sounds, incorrect positioning of connecting parts, and major MP issues are just a few.

It'll be a few months before this game is considered stable and where it needs to be.
 
TheCapulet said:
I see many say Arma 3 is buggy too, but quite frankly, I've not once come across a show stopping bug. Or really any bugs worth mentioning for that matter.

Realistically the only bugs you'll see stem from poor mission design from other players/mission makers. You'd never see these issues in other games, because you'll never play player-made missions in other games. Where as with Arma 3, there's a good chance that's what you'll spend the majority of your time doing, even after the SP campaign is officially released from BIS. (Nearly all of the multiplayer games you'll play are player made. Some are terrible, other's are great like "BECTI", or "Conqu3st". (both professional quality gametypes/missions)

Whenever you hear someone shouting "ARMA 3 is buggy!!!!11one", take it with a grain of salt. Because chances are, they don't have a single clue about what they're shouting.
Click to expand...
Ok , great . Thats what i thought , I did check out forums but its hard to know whats right .

Thanks a lot .
 
Brent_Justice said:
Than you for the information, very informative

Seems we hit the nail on the head r.e. FXAA sharpening filter, IMO it does make things look a lot better in this game, wish it had been enabled inside SMAA as well.
Click to expand...

He makes it sound like you can enable it with SMAA


imho the sharpen post process is so good it deserve separate setting so you can use SMAA+sharpen filter too
(as SMAA is better than FXAA in terms of edge aliasing quality)
 
Nice find TheCapulet!

See, i am so spoiled from using the shader packages from the SweetFX suite that i already assumed that the sharpen was a natural feature :p my bad, i just helped prove yet again the assumption deal :p lol
 
There were collision and "WTF HAPPENED TO PHYSICS" bugs in the past, but the latest patch seems to have fixed them. No more tanks bump into each other -->the one that got bumped flies up into the air anymore. LOL


And unfortunately I'm playing it in a temporary video card too, a GT 630...1080p can only be really playable at minimum settings (for 50-60fps). Increase some of the options can improve quality somewhat but there would be an around 20fps reduction.
 
I know you didn't have time but it would be interesting in see how much the HD7950 Boost leaves on the table if you was to match the clock speeds of the HD 7970Ghz only because of it's current price range of $200 or less as only the SP count makes them different, my Sapphire 4L boost does those clocks on stock voltage of VDDC 1168 using Trixx (1005/1500)
 
powruser said:
I believe I found an error in the apples to apples section where the GTX 770, 7970 GE, GTX 760 and 7950 Boost were tested at 1920x1080 - 8x FSAA - ultra settings.

Error in bold and underlined:

"In this apples to test we are running ARMA 3 at 1920x1080 with 8X FSAA enabled, and overall quality set to "ultra." Alpha to coverage is enabled on all trees and grass, and we are using FXAA Ultra.

At these settings the GeForce GTX 770 performed the fastest in this group, averaging 43 FPS. This is slightly below what we considered playable. It is 12.8% slower than the GeForce GTX 780 at the same settings,
however it is 13.3% faster than the Radeon HD 7970 GE and 18.8% faster than the GeForce GTX 760."

By my calculation, the GTX 770 is 4.88% faster than the 7970 GE, not 13.3% faster.

43 fps for GTX 770 minus 41 fps for 7970 GE = 2 fps faster
2fps / 41 fps average fps for 7970 GE = .04878 x 100 = 4.88%


Looks like a pretty brutal game. Single GPUs are taking a beating. Then again, the ARMA series was never really known for being well optimized.
Click to expand...

Beat me to it.
 
Gabkicks said:
I ran arma 3 mark and recorded cpu and gpu utilization... gpu is only showing about 50% usage on average with my ASUS gtx 780 Direct CU II :/

http://www.youtube.com/watch?v=W25dri_CXD4
Click to expand...
I take it Vsync is set off ,does same thing happen in game , low usage and clock freq ?

The fps are low so not sure whats up there, could be driver issue too .

Does running power management on max performance in Nvidia control panel help ?
 
I been playing the game since Alpha release and love the realism. Still pisses me off though that sli scaling for this game sucks. Riva tuner osd shows both my titans max out at maybe 60-65% but usually hover around 40-50% I've tried custom profile, enabling and disabling settings, etc. Crysis 3 sucks the power out of my Titans using EVGA precision tweaking offsets, AND sli usage is always pegged playing that game. Hope Nvidia releases updated drivers to fix the scaling issues....Still love the game though...Swapped from playing wasteland to coop and the realism is so frekkin' intense when your running with a good squad who play the game right...
 
Island said:
I been playing the game since Alpha release and love the realism. Still pisses me off though that sli scaling for this game sucks. Riva tuner osd shows both my titans max out at maybe 60-65% but usually hover around 40-50% I've tried custom profile, enabling and disabling settings, etc. Crysis 3 sucks the power out of my Titans using EVGA precision tweaking offsets, AND sli usage is always pegged playing that game. Hope Nvidia releases updated drivers to fix the scaling issues....Still love the game though...Swapped from playing wasteland to coop and the realism is so frekkin' intense when your running with a good squad who play the game right...
Click to expand...
It seems it's not just SLI issue if you look at Gabkicks vid (up 2 posts) with his single 780 he is also getting low GPU usage .
 
You must log in or register to reply here.
Back
Top