Battlefield 3 GPU Performance and Eyefinity Experience

Add Your Comments

Before starting here, take a little while to read Janus67’s excellent review of Battlefield 3 if you haven’t done so yet. This article is going to focus on performance. If you are considering making the leap to Battlefield 3, have a strong GPU and were curious about how your system would perform, you’re in luck. We have the scoop on several of today’s top cards.

Benchmarking Procedure

The most important part about benchmarking with games is that you have consistent, repeatable results so you can compare across multiple cards. The way we did this was to recruit forum members with very similar systems, running at the same CPU clock. RAM clocks varied from DDR3-1600 to DDR3-2133.

  • i7 2600K @ 4.5 GHz (Removing any possible CPU bottleneck. Note one exception is the GTX570 result, which was run at 4.6 GHz.)
  • DDR3-1600 to DDR3-2133 RAM
  • {BYO GPU} running @ stock
  • The game test should be run at 1920 x 1080 and using the “Ultra” setting, with all the defaults that applies when set.

The method of testing is manual, but is remarkably consistent (we’re talking less than 1 FPS difference between runs). Instructions:

  1. Turn on Fraps
  2. Start the first mission: Operation Semper Fidelis
  3. Begin recording FPS (F11 by default) after you pick up the pistol and are looking forward.
  4. Work through the level as normal.
  5. Stop recording FPS when you’re at the end of the level and the game takes over motion, hit F11 again when the guy puts his hands up.
  6. Repeat 3x at stock (the runs take approx 5 min each including the beginning cutscene).
  7. (Optional) If you feel ambitious, overclock your GPU to its stable 24/7 overclock and run it once.

The nature of this level makes it perfect for consistent and repeatable FPS results, which is why we decided to go with it. Plus it doesn’t require people to play far into the game. There is plenty of detail to stress the GPU. Of course, being enclosed this level will give slightly higher FPS than a massive 64 player online map. In my experience, on the larger maps you’ll lose anywhere between five and fifteen FPS compared to the numbers shown here, depending on the map and amount of action.

Most GPUs tested were run at stock and overclocked. The GPU overclocks featured in the graph are:

GPU

Overclock 
(GPU / Mem)

HD6970 – OC 960 / 1375
HD6990 – OC 960 / 1250
GTX480 – OC 850 / 2150
GTX480 SLI – OC 850 / 2150
GTX580 Tri-SLI – OC 925 / 2300
Battlefield 3 GPU Comparison

Battlefield 3 GPU Comparison

When looking at the minimums, it seems AMD has a very slight edge with framerates not dropping quite as far relative to the averages. The HD6970 , GTX480 and GTX570 are roughly in the same ballpark. The GTX580 definitely shows its single-GPU muscle. Interestingly, the GTX480 SLI, GTX580 SLI and HD6990 results are all extremely close to each other; Crossfire (on a stick!) scales quite well.

Before moving on, huge thanks goes out to all of the fine gentlemen that took their time to run the tests. Only the HD6970 and HD6990 numbers came from me, the rest came from these kind individuals.

Now, to get the most out of your setup regardless of GPU, both the red team and the green team have put out guides for just that. AMD’s ‘guide’ is short and to the point. NVIDIA’s is more comprehensive and covers pretty much everything.

Eyefinity Experience

This game offered the excuse we needed to grab a third monitor for eyefinity testing. This was the first time I had the opportunity to try this first hand and it is one intense experience. The hardware used was the same 2600K at 4.5 GHz and 8 GB of G.Skill Ripjaws X DDR3-2133 / 9-11-9-28 RAM. I tested with the AMD HD6990 by itself and with the HD6990 + 6970 in tri-fire.

Eyefinity Setup

Eyefinity Setup

One thing you should note is that three monitors takes up a LOT of space. My desk would be considered pretty large by any reasonable standard. I can put a bench setup for sub-zero LN2 benching next to a monitor, keyboard and mouse and have plenty of breathing room. When you put three monitors next to each other though, that takes up more space than you’d think. It’s totally worth it if you’re a gamer though, the experience is very impressive.

Eyefinity Monitors Running

Eyefinity Monitors Running

A question that came when this was posted on the forum was about the bezels. Honestly, you don’t even notice them. When you see photos and screenshots of Eyefinity it might seem like the large field of view would be used in its entirety – as in you’re looking left and right for enemies, etc. It’s nothing like that. While you can do that, if you try it you’ll end up dead very fast. Instead, all the side monitors are good for is true peripheral vision.

To survive and play well, you’ll be focusing straight ahead on the center monitor. The side monitors will serve as your peripheral vision. While it’s not a direct line of sight like you might think, playing like that definitely helps draw you into the game. Once you get used to it, you will notice those enemies eventually just as you would in real life with peripheral vision.

We have a few screenshots for you to see the field of vision available. These will open in a new window so you don’t get a lightbox from which there is no escape.

Eyefinity In a Humvee

Eyefinity In a Humvee

Eyefinity Down the Street

Eyefinity Down the Street

Eyefinity Environment

Eyefinity Environment

If you really want to see what it’s like, we have the full resolution screenshots for you. Click the links below to embiggen. Warning: They are almost or over a full meg each.

The biggest drawback to Eyefinity is that you will need a lot of horsepower to run such high resolutions. I didn’t even try running it on the HD6970 alone, but we do have numbers for the HD6990 and the HD6990 + HD6970 tri-fire combo.

Battlefield Eyefinity Comparison

Battlefield Eyefinity Comparison

A single HD6990 is barely enough to run the test level in Eyefinity at Ultra settings. It is decidedly not enough to run it on a huge 64-player map. If you can live with MSAA turned down/off or FXAA instead of MSAA, the rest of the Ultra settings should be playable. Adding the HD6970 to it made for smooth sailing through any map or number of players Battlefield 3 could throw at them on the full Ultra settings. Crossfire scaling is simply remarkable. Going from a stock HD6990 to the stock HD6990 + HD6970 in tri-fire gained 45%!

I won’t be running Eyefinity all the time. That would require purchasing a fourth monitor (one is for my test bed, two are for the main system, which actually makes graphing things in reviews pretty painless) and it’s not something I’d use often enough to warrant the purchase. But if I gamed as much as I used to and had the money to spare, you can bet I’d be all over an Eyefinity system. It is extreme immersion and with a knock-your-socks-off first person shooter like Battlefield 3, it doesn’t get much better.

– Jeremy Vaughan (hokiealumnus)

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Discussion
  1. nice article!

    It convinced me to get a third display and a stand to finally do eyefinity.

    Unfortunately, I went a little conservative on the screen size (didn't think the stand was as big as it was), so now I need to talk the wife into letting me swap them all out ;)
    AWESOME write up dudes!

    Also, for those of you who havnt seen this...

    http://www.evga.com/forums/tm.aspx?high=&m=1323124&mpage=1#1323124

    http://www.evga.com/forums/tm.aspx?m=1174372&mpage=1
    I run mine with my 5850 with the core clocked to 950, card wont clock much higher, med rez and get about 32 FPS. It would be nice to have 2 5850's for this game and it would run like butter
    Janus67
    Owen, if you kept getting microstutter try disabling HT on the 2600k. I didn't need to do so (but I know a lot of people with 9** series 1366 CPUs needed to do so to get rid of the microstutter problem.


    Good idea, thanks! :thup:

    I have eight cores with HT on which is way more than I need. I just need to get my video card back from RMA and either RMA my mobo or buy a new one because when the card died so did the PCIe slot it was in.:bang head
    Owen, if you kept getting microstutter try disabling HT on the 2600k. I didn't need to do so (but I know a lot of people with 9** series 1366 CPUs needed to do so to get rid of the microstutter problem.
    Great article!

    I actually had two GTX570's in SLI up until two weeks ago when one died :(. It would have made a nice addition to your review. When it comes back in I'll see what I can do to get you those results to add.

    I ran Nvidia Surround (5760x1080) up until that time as well. I can possibly get you some numbers for that too. The problem I ran into was microstutter in BF3. It just felt like it wasn't as smooth. I played BFBC2 extensively in Nvidia surround as well as a few other games. It is a blast! But it also was driving me to desire faster and faster video cards with more RAM and budgetarily that is kind of a dangerous place to be for me right now. :D
    Lukee
    Nice article. I would have liked to have seen for the first GPU benchmarks the 6990+6970 Tri-fire since you have the tri-sli 580s in that comparison.


    Yea, that was completely my fault, I apologize. When the third monitor came in I got all excited, set up tri-fire and just ran them for the Eyefinity benchmark, totally forgetting to run it for 1080p. Currently I'm excessively busy but I'll try to do that sometime soon and report back.

    That said, Crossfire scaling is excellent in this game. The 6970 to 6990 gained 89.8% at 1080p. Adding the 6970 to the 6990 in Eyefinity gained 45.3%, so the fact that it's connected via a bridge instead of Crossfire-on-a-stick doesn't seem to hurt anything.

    Based on those two numbers, it wouldn't be unreasonable to expect a similar 45% gain on 1080p also, so with the 6990 getting 89.6 FPS at stock, you'd be somewhere around 130 FPS with a 45% gain. It wouldn't beat the Tri-SLI 580's, but would get close and cost a heck of a lot less (~$1,500 for three 580's vs. ~$1,030 for 6990+6970).
    This is an awesome article, and it exactly matches my experience (aside from some weird issues with the latest patch) with a single HD 6970 (6950 unlocked + OC'd to same specs as you have your 6970 OC'd to). I REALLY now want to buy a 6990 to go with my current card to run eyefinity...just...wow...
    Nice article. I would have liked to have seen for the first GPU benchmarks the 6990+6970 Tri-fire since you have the tri-sli 580s in that comparison.