Battlefield 3 Performance: 30+ Graphics Cards, Benchmarked

[MadGoat1979]

Im using 2 285's in SLi... Terrain quality is set to Medium, not Low. (wont go higher, I assume DX11 is needed to go higher).

Settings look like this:

1920x1200
Texture - Ultra
Shadow - Ultra
Effects - High
Mesh - High
Terrain - Medium
Terrain Decoration - Ultra
AA Deferred - Off
AA Post - High
Blur - Off
AF - 16x
Ambient Occ - Off

I get 62 avg Fps and Min of 43. Online Multiplayer. (even 64 player maps)

The only thing that really helps out the bottom line min fps is the MESH setting... Cranking that down to Medium nets about a 8-10 fps min and avg gain. However the distance in texture loading is notable to me and the ~40fps min is not bad at all to me. (only hit min for fractions of a second when it does.)


Not bad at all and the game looks AMAZING with these settings.

My system is 1100t @ 4.125ghz, 8gb ddr3 1666 8-8-8-24-34-1, NorthBridge @ 2.8ghz ... on a 990fxa-ud3

Overall I'm impressed. HOWEVER! My cards wont go over 75% usage.

And before you chime in with "Cpu bottleneck, OMG go INTEL SB or GTFO"...

I can run ANY mix of resolution / quality settings and it WILL not go over 75%. This game is not nearly as cpu bound like everyone thinks.

Once they do some work on the SLI front, things will hopefully get better.

 

[tfbww]

[citation][nom]linford585[/nom]Gotta make a quick note on that there Chris!Not too long ago, 22" monitors were all 1680x1050, and 24" were all 1920x1200. I'd say most people at the time didn't want the aspect ratio of 1920x1080 on a PC. Many of us are still rockin those 22" 1050 monitors, and it wouldn't make sense to upgrade to a 1080!That said, I love maxing BF3 with my 6970! (Just skimmed through the review at the moment, can't wait to read it all later, thanks!)[/citation]
Yup, I'm one Those Guys running a 6970 with a 20.1" 1050 monitor -- one that cost "only" $440 back when I bought it in 2006 from Dell! But for now, it works fine. Kind of hard to ditch a monitor when it works so well relative to other components that REALLY needed upgrading. But eventually I'll make the monitor upgrade and thus the 6970 (but maybe I'll wait until a second 6970 comes down enough in price to XF... hmmm....)

 

[neonneophyte]

as a 6950 owner i learned a very important thing here in regards to bf3.

4xMSAA on ultra settings is a no go. however, turn that off, and you gain 20 fps or more. Seems to be a driver problem. I turned mine off, seeing over 60fps now. 55 min.

i was *this* close to buying a 2nd 6950. i really want ultra settings for bf3. you have saved me 300 bucks! (for now lol)

 

[Guest]

I'm on geforce fx560m with 8g ram.. works amazing!!

http://battlefield3guide.net

 

[youssef 2010]

[citation][nom]Article[/nom]It’s clear from our Low quality benchmarks that anything slower than a Radeon HD 4850 is going to struggle—and that’s at 1680x1050. Nvidia’s GeForce GTX 260 fared a little better in this particular discipline.[/citation]

First off, I'm amazed by the amount of passion and commitment you give to your work. To get all this data dumped in one or two days is AMAZING. You guys are the best. However, there's one mistake I'd like to point out.The 4850 was never designed to compete with the GTX260. That was the 4870's job. And you said that it did its job well. So, this paragraph is completely wrong .

 

[youssef 2010]

A new graphics Hog is in town

 

[razor512]

How well does the GTX 260 handle the ultra preset at 1920x1080

(Want to see how well it will work with a HTPC when friends and family come over (don't want others using my main gaming PC)

 

[fausto]

this article is just horrible. absolutely useless!
nobody should rely on these benchmarks for bf3.
Battlefield games are very tough on video cards and this benchmark comes nowhere near the maximum loads your cards will experience in a 64 player game with lots of action.

go to http://www.sweclockers.com/artikel/14650-prestandaanalys-battlefield-3 for real benchmarks of bf3. use google to translate the site.

 

[cangelini]

[citation][nom]fausto[/nom]this article is just horrible. absolutely useless!nobody should rely on these benchmarks for bf3. Battlefield games are very tough on video cards and this benchmark comes nowhere near the maximum loads your cards will experience in a 64 player game with lots of action. go to http://www.sweclockers.com/artikel [...] tlefield-3 for real benchmarks of bf3. use google to translate the site.[/citation]

The story you linked appears to focus on MP; mine is on the campaign. After just finishing the campaign, I'm very comfortable with the correlation between these results and my own experience.

Multi-player is going to be different, and no secret is made of that in the story.

Thanks for the feedback, though!
Chris

 

[fausto]

[citation][nom]cangelini[/nom]The story you linked appears to focus on MP; mine is on the campaign. After just finishing the campaign, I'm very comfortable with the correlation between these results and my own experience. Multi-player is going to be different, and no secret is made of that in the story. Thanks for the feedback, though!Chris[/citation]

just saying, nobody is buying this game for the campaign. look at the other comments, people talking about buying cards based on your piece.

I have a Q9550 and a 6990 in use right now. My system should be smoking this game. Yet in large 64 player multiplayer games the frame rate drops to 30 and hangs low on Ultra @ 1680x1050. I am actually playing medium settings for now and hoping and praying for AMD to deliver a miracle on drivers. To me Caspian Border seems to be the most challenging map to render.

You can state something about this benchmark not being as challenging as the mp but people don't register that or go straight to results without looking at methodology.

 

[gallovfc]

[citation][nom]nebun[/nom]sli and corssfire will bring your frames up and have a better play experience in almost any game[/citation]

That's known, what was not clear is the REAL NEED to go for CF/SLI using mainstream cards at 720p. I don't want to buy another (I.e.:HD 5770) without a real need.

 

[cangelini]

[citation][nom]fausto[/nom]just saying, nobody is buying this game for the campaign. look at the other comments, people talking about buying cards based on your piece. I have a Q9550 and a 6990 in use right now. My system should be smoking this game. Yet in large 64 player multiplayer games the frame rate drops to 30 and hangs low on Ultra @ 1680x1050. I am actually playing medium settings for now and hoping and praying for AMD to deliver a miracle on drivers. To me Caspian Border seems to be the most challenging map to render. You can state something about this benchmark not being as challenging as the mp but people don't register that or go straight to results without looking at methodology.[/citation]

Hopefully repi can provide some good feedback on generating accurate data using a test environment that, by its very nature, is not consistent.

 

[apexle]

This is a good review, and I think its a good point that you have to watch out for the deferred AA (the 0, 2x, and 4x). I can crank up the AA from 0->4x->8x->16x without any significant drop in FPS, while just bumping the first multisampling option up to 2x from 0 drops my FPS to 40 from 60fps, and 4x gets me 30fps. I'm running everything else maxed out at ultra settings, the ONLY thing I turn off is the 4x down to 0, and I don't really notice any difference.

So I'm getting 70-90FPS, sometimes dropping to 50FPS under intensive battles on screen with these specs:

1920x1200 resolution

i5-2500K OC'd to 4.4Ghz
8GB DDR1600 RAM 8-8-8-24
Z68 Gigabyte UD3H
SLI'd MSI GTX 460 1GB Talon Attack @ 920/1870/1990 graphics cards

3dmark vantage: 28800
3dmark11: P8170, X2881

 

[apexle]

Yeah again to reiterate, the deferred MSAA definitely turn it off, unless you are getting over 100FPS normally (I'm getting 70-90FPS normally and turning it on to 2x drops my FPS to 40FPS). If it is a driver issue then I hope the next one will allow us to use it.

 

[blbeta]

[citation][nom]clonazepam[/nom]I'd start off with a graphics card update first... someone with more experience with your particular processor can chime in, but I'm not sure the performance boost from a more modern cpu/mobo/ram combo will have nearly the same boost going from a 260 to something more modern like a 6950/6970/560/570...If you play GTA IV, then you definitely want to uprade everything. For a game like this, and Crysis 2, the cpu will probably have the least amount of influence on game performance.This is all from a "gaming only" perspective because I'm not going to pretend to know what you do with your pc.[/citation]

Perhaps this is true, but there is almost no game that I get above 50% on GPU load. Almost all games are 100% CPU load. My CPU is now over 5 years old. I will get both in the next six months, but figured CPU has got to be my bottleneck by now. Maybe I am wrong.

 

[blbeta]

Maybe I'll get my buddy to let me slap his GTX 560Ti in and see what happens. Problem is I don't have the beta to try anymore. I'll figure something out.

 

[ksampanna]

Confirms what i've suspected all along ... i7-920 + 5850 ftw !!!

 

[randomizer]

My GTX 275 isn't looking so hot any more.

 

[eddieroolz]

The game seems to present a mixed bag for both graphics companies, thought both probably concern the driver and given the infancy of the game itself, it's something that can be fixed with subsequent driver updates.

 

[megamanx00]

Good ol 5870 still doing good. Too bad the single player campaign on this game is almost worth it, but then you pick it up for the multilayer .

 

[flips]

I hope CF 6870 work just as good in multiplayer in ultra settings!

 

[ericwolf]

thanks for this ecllent review helped a lot
love it!!!

 

[fausto]

[citation][nom]cangelini[/nom]Hopefully repi can provide some good feedback on generating accurate data using a test environment that, by its very nature, is not consistent.[/citation]


Yes, it would have been nice for them to include a heavy duty benchmark representative on multiplayer load.

But as far as consistency goes...the average fps is pretty consistent in the experience you will have. it's just more time consuming.

done using fraps with different settings that are playable to me right now.
Frames, Time (ms), Min, Max, Avg
39120, 575285, 0, 130, 68.001
Frames, Time (ms), Min, Max, Avg
35789, 721052, 31, 100, 49.634
Frames, Time (ms), Min, Max, Avg
30174, 504866, 24, 113, 59.766
I will play at eat setting one game of caspian...10 minutes minimum to see how it differs.

 

[fausto]

[citation][nom]ksampanna[/nom]Confirms what i've suspected all along ... i7-920 + 5850 ftw !!![/citation]
Think again in multiplayer. Report back on playable settings for you.

[citation][nom]eddieroolz[/nom]The game seems to present a mixed bag for both graphics companies, thought both probably concern the driver and given the infancy of the game itself, it's something that can be fixed with subsequent driver updates.[/citation]

That is what everyone with a high end system is hoping.

[citation][nom]flips[/nom]I hope CF 6870 work just as good in multiplayer in ultra settings![/citation]

drivers available now either suck or this game is too taxing for crossfire to matter much. i'm hoping the drivers suck because AMD could fix that
[citation][nom]randomizer[/nom]My GTX 275 isn't looking so hot any more.[/citation]
Nope...my 6990 puts out the heat.

[citation][nom]ericwolf[/nom]thanks for this ecllent review helped a lot love it!!![/citation]
Just make sure you don't expect those numbers at those settings in multiplayer.

 

[linford585]

just saying, nobody is buying this game for the campaign. look at the other comments, people talking about buying cards based on your piece.

I have a Q9550 and a 6990 in use right now. My system should be smoking this game. Yet in large 64 player multiplayer games the frame rate drops to 30 and hangs low on Ultra @ 1680x1050. I am actually playing medium settings for now and hoping and praying for AMD to deliver a miracle on drivers. To me Caspian Border seems to be the most challenging map to render.

You can state something about this benchmark not being as challenging as the mp but people don't register that or go straight to results without looking at methodology.

I play on a 22" 1680x1050, with an i7 930, and a 6970. I almost always play on 64 player servers, and I run the game on all ultra (MP I use no AA, SP I use 2x AA). These benchmarks seem pretty accurate to me. To help though, I did OC everything

Edit:
I should clarify, while clearly my average FPS in these servers is not the same as the Ultra preset FPS shown in this article (for a 6970), I still hold that a 6970 can play on ultra settings in big MP games.

That is:
A 6970 OCed to 925Mhz/1400Mhz can play on a 1680x1050 resolution, set to custom graphics settings - all set to ultra, and MSAA set to off, not just in SP but also MP.