Hardware Battlefield 3 Performance Analysis

MSAA is the technically better AA than FXAA, but it is difficult to exactly pinpoint what looks different except when comparing screenshots side-by-side. I do prefer MSAA tbh, but as long as each individual is happy that's what matters.

@sp4nky: I think it must be pushing the card too hard, that's the only thing I can think of. My temps stay fine at 71C, but it just seems a bit random to be a problem with Frostbite2.

 

From what I recall, yes (currently running PLP). I tend to play on hardcore servers so I am not 100% sure on the names above players / Hud elements. I know that it didn't work in Landscape mode at all (I had a terrible time getting the game to work properly with Landscape Eyefinity).

Have you tried / can try Portrait Eyefinity?

I do get the problem of an over abundance of bloom though. Coming out of the Terminal in the Metro map is blinding.

 

Really great article, love the depth put into it. Hope future games can all get the same treatment!

Three things:

-Your card is factory overclocked, it will perform faster than the stock GTX460 tested by Bit-tech.

-Where are you getting 60fps? All of the Bit-tech tests were run on a scripted section in the singleplayer campaign to ensure that all cards were tested evenly. Measure your results for this section to get a more accurate comparison.

-How are you measuring your frames per second? Note that Bit-tech have both minimum and average frames per second recorded and only record the duration of the scripted event. To get an accurate comparison you will need to also measure in this manner.

 

Thanks for this .

 

The >2GB memory scaling , permitted by running a 64bit OS with lots of system ram bit the dust along the way. Didn't DICE mention it would have taken too much time to code or something.

 

Hardocp.com did a similar test of BF3, basically it comes down to this:

1) MSAA takes a heavy toll on fps
2) MSAA is ineffectual at removing jaggies*
3) FXAA is the most effective means of removing aliasing in Battlefield 3

*MSAA is applied to the frame before it is finished, extra effects on top applied post-MSAA pass are not treated and so the finished frame will still contain alot of jaggies.

Source: http://www.hardocp.com/article/2011/11/02/battlefield_3_single_player_performance_iq_review

 

Good review thank you.

I agree with the post above. Please run these benchmarks on a multiplayer map with 64 players.

You will see a Dual core cpu is not enough this game, also performance improves on machines with more than 4GB of ram in this area of testing.

Also a videocard with 1GB of ram is the minimum you want for this game. I've seen plenty of post from people with NV cards like the 570/580 that run into vram limits once the settings are increased on largers maps.

BF is and as always been about Multiplayer, the Single player campaign is a joke and just tacked on to please the Generic COD crowd.

 

Nice article guys, I'm using an AMD Athlon II 250, 4Gb DDR3 Ram, and a 9800GTX+ and can easily play at 1080 on medium with a minimum of 26FPS and if i drop the resolution down to 1280x720 I can play at high settings with a minimum in the high 30's.... My card is overclocked to 802 Mhz so that probably helps alot but it goes to show that the game does really scale well.....

 

*hugs his dual 5850's* I love you guys!

 

you should get atleast high with that card. my 5850 creams the game at high settings, albeit at 1680x1050

 

So my guess is that a pair of HD6950 will give similar performance as a HD6990?

 

Yeah, but if you're running dual, unlocked 6950s, wouldn't you expect to be able to crank everything up all the way? Whether it makes a difference to you personally, I'd be pissed if I laid out that kind of cash on video cards and couldn't get nice, high framerates without disabling some features. I'd say it is a knock against AMD in that regard.

 

obviously a well engineered piece of software well done.

 

at busy points in the game i'm sure that no single gpu card would run it perfectly on full tilt on a big monitor, i am old and have been gaming for a very long time. Put it this way i've tested more cards than i can even remember

 

There's no need for 2 6950's. I run on Ultra with 0AA and 8AF and average 56fps with a max 73 1680x1050 with one Sapphire 6950 2GB

 

The test results (2560x1600 with GTX580) are almost exactly the same as I get here. Nice test. I can't wait to go 580 SLi soon

 

I don't own this game, but most likely AMD will put out driver update that most likely will speed up the fps. Especially for the 6900 line of cards.

It also depends on the authors of the game, if they intentionally favored Nvidia, which some games do. Yet the ATI card still takes the lead for the fastest fps without AA, which is the reason I think a driver update will improve the AA.

From another article it sounds like BF3 is using some Nvidia features to make it faster.
http://www.ngohq.com/news/20747-amd-recommends-using-nvidias-fxaa-in-battlefield-3-a.html

 

Would the game be playable on ultra at 1,680 x 1,050 with a single 4890 and sufficient memory and processor?

 

My Palit Sonic Platinum GTX460 1GB @ GPU 880mhz and Memory 2140mhz running BF3 on Ultra other than high shadows, SSAO no FSAA other than low post AA. based on the current beta drivers released this week, my performance with v'sync on is 55-60, mainly sticking at 60, only smoke gets it to drop to 55.

I only have a 22" LG at 1680x1050 however, but either way these are pleasing results and shows how good the GTX460 is on a modest TFT when overclocked.