[Tom's] BF 3: 30+ Graphics Cards, Benchmarked

[amenx]

Midrange card owners will opt for high, 6950 - 560ti (and up) will likely go for ultra. 30-40 fps will do nicely for many (incl me). Isnt the goal maximum playable settings?

 

[TheAdvocate]

Haven't we pretty firmly established that the difference between AMD and nVidia in this game is how well/poorly each handles the different types of AA?

Tired of the fanboy stuff. Would rather see a solid post explaining the optimum AA settings for each brand rather than this back and forth "green/red is ZOMG BETTAR" stuff. The trick is to find the settings that optimize peformance for each, not wag your epeen.

 

[GoStumpy]

So as it turns out, I thought I was buying a mid-high card (HD6850) and ended up buying a low-end one for BF3! Glad I only spent $120

 

[Grooveriding]

Here is what I am expressing is my experience with these AA modes.

The first image is with no AA at all, you can see jaggies everywhere, but also look at the textures of the tanks hull above the track and the trees in the background, as well the fencing on the left:

Now this is with 4x MSAA, again note the trees and the textures:

Here is high Post-AA, look at whats happened to the trees and the tanks texture, as well the look of blur across the whole screen. The effect is very pronounced on the trees though and gives a good example of the issue for IQ, imo:

This last is with 4x MSAA and Post-AA. Comparing this to the previous, I would say if you are using Post-AA, there is no reason for MSAA at all:

I have a hard time seeing any difference between High Post and High Post+4xMSAA. I think the only reason to use 4x MSAA is if you prefer the IQ of not having post-AA on, otherwise it's not worth using it with post-AA.

I really, really don't like this AA mode. It was even worse in Crysis 2 and Red Orchestra 2 feels like you wearing coke bottle glasses with it on :\ This AA mode sort of reminds me of the film grain effect you could activate in Mass Effect. It applies an entirely different feel to the image presented on screen. AA for me is about cleaning up the demarcation between objects, not giving the entire image a filter effect.

 

[Arkadrel]

The first image is with no AA at all, you can see jaggies everywhere, 
 but also look at the textures of the tanks hull above the track and the 
trees in the background,  as well the fencing on the left:

1) the game looks stunning even without AA or anything.
2) Jaggies everywhere? the only place its really noticeable is on the helicopter blades (compaired to the other pics) and a tiny bit on the fence.

I wanted to say that the game looks great without AA, and the effects from turning on AA high and MSAAx4 is very low visual gain wise.

 

[RussianSensation]

Grooveriding, your 3rd image looks a lot more like a console because FXAA/MLAA result in a blur/soft image effect as you noted, similar to console AA filters. I don't like that look.

It's clearly evident in Crysis 2 when comparing Very High vs. Extreme or Ultra.

Very High - no FXAA

Extreme - FXAA (look at the 3 blue chairs, the briefcase, yellow and silver tiles on the walls)

 

[cmdrdredd]

I almost dont see a difference with MSAA off, usually i run all games with 4xAA but in this case there is hardly a difference...so i run without MSAA for some more FPS. (I have a low end graphics card)

As for the TH results...the FPS decrease by turning ON MSAA is rather VERY small (looking at the numbers in comparison), anywhere from 10***37; - 20%....which IN MY OPINION does not matter if you have a decent card like a 570. He says in his report that a GTX 590 is the only card where is makes sense to have MSAA on which i think is a nonsense statement.

I however turn it off because i have a GTX275, so 10 more FPS is great...but if i had a beefier card (570 etc.) i would def. run with 4xMSAA

I'd turn on MSAA in the single player if I was showing the game off or just playing single player for some quick grins. Online I'd probably always have it off just to get the extra FPS.

Grooveriding, your 3rd image looks a lot more like a console because FXAA/MLAA result in a blur/soft image effect as you noted, similar to console AA filters. I don't like that look.

To my eyes the chairs look better and smoother. The black dots look like they don't belong. I don't see anything wrong with the tiles tbh.

 

[SirPauly]

From my experience using MLAA in Total War: Shogun 2 and Deus Ex: Human Revolution, post processing AA is definitely better than no AA at all

I totally agree. The key is hopefully as these type of abilities mature and evolve, well, the quality may rise.

 

[bryanW1995]

Multiply 6850 framerates by two ->

1680x1050: ~90fps.
1920x1080: ~80fps.

That's about 10 fps better than 6970 or GTX 580.

It doesn't work that way, surely you know that.

 

[bryanW1995]

Check the High settings though: Probably what most people will play.

570 = 62 fps
6970 = 70 fps

Thats like a 13% differnce there, in the 6970's favor.

Nvidia shine on the Ultra settings though, but few will play that.
Because they want 60+ fps in a first person shooter, unless they are running SLI setups, ultra settings wont help that.

ps. on High settings 6970 > 580 in fps, and at a much lower price for the card.

Unless your planning to get SLI/CF, AMD cards seem to be a better way to go with BF3.

Why would most people play on high just b/c it's a fps? And nvidia cards are fast enough to get great fps even on ultra, while with AMD you are stuck with high instead.

 

[GoStumpy]

Anyone else chuckle at the fact that we're talking FPS in a FPS?

Anyhoo, is there really a noticeable difference between 35fps -> 50fps?

 

[RussianSensation]

To my eyes the chairs look better and smoother. The black dots look like they don't belong. I don't see anything wrong with the tiles tbh.

Those chairs are supposed to have those thin lines. It's grooves in the chairs. They are almost non-existent with FXAA since it washes out the textures. If you prefer chairs without grooves, that's personal preference but that isn't how the designers intended them to look.

Also, if you look at the tiles where the shopping cart is (where the gun is pointing to), they are far more defined and sharper in the screenshot without FXAA. Overall, the FXAA adds blur in Crysis 2. There are plenty of screenshots around the internet that show this.

 

[Grooveriding]

Anyone else chuckle at the fact that we're talking FPS in a FPS?

Anyhoo, is there really a noticeable difference between 35fps -> 50fps?

In this game very much so. There is a big difference between 40fps and 60fps from my experience.

Also, considering this game is primarily an online FPS, you want as high a framerate as possible. I would adjust settings to get whatever gives me 60+ steady.

I wish they included a canned bench for this game. Even though AMD and nvidia will optimize for the bench to look as good as possible, at least it gives a baseline. Every review is doing something different to get their results, so there is no consistency.

 

[mrjoltcola]

2 GB on a 550 Ti doesn't have much use, but what's wrong with buying a 550 Ti if you can find a good price on it? (not that Best Buy or Microcenter are such places). Not everyone has the extra cash to burn on upgrading to a 560.

Because people think by investing in a 5-series they will get better performance, but the benchmarks prove if they are going with the 550ti, they would save their hard earned money by buying a used 280 or 460, while getting better framerate. The point is, the 550ti isn't a good use of $130.
There are used 280s and 460s going for $80-90 bucks, or even new Radeon 6850s going for cheaper with 25% better performance.

I feel that if more reviews included the 550ti it would be more apparent how poorly it performs next to the older models.

 

[Bobisuruncle54]

Anyone seen how AMD's MLAA handles it instead? Just curious.

 

[lehtv]

It doesn't work that way, surely you know that.

It actually does, my estimates were based on the near-perfect scaling of the 6870 (and yes, I accounted for that 5% deficit).

 

[Grooveriding]

Grooveriding, your 3rd image looks a lot more like a console because FXAA/MLAA result in a blur/soft image effect as you noted, similar to console AA filters. I don't like that look.

It's clearly evident in Crysis 2 when comparing Very High vs. Extreme or Ultra.

Yup. In Crysis 2 it was the worst I have seen this mode except for in RO2. Everything looked blurry. Particularly your weapon, which is sort of the focal point at times in an FPS. Turning it off made everything very crisp and nice.

We were at the point with standard engines where 4xAA, 8xAA and at times higher where no problem to enable. Now with all this deferred business we are back to AA incurring a massive performance hit unless you want to use MLAA/FXAA. :\

 

[dennilfloss]

Looks like a three-year old 4870 can still play this at high quality @ 1920x1200. ^_^

 

[Red Hawk]

Perhaps. The benchmark tomshardware ran was admittedly not a particularly strenuous part of the game. It was done for the sake of consistency between cards rather than a realistic portrayal of the performance of each individual card.

I feel this discussion on post-processing anti aliasing should be made into a separate thread...

 

[FalseChristian]

I think the point is if you can play at Ultra and get an acceptable frame rate for yourself than use the Ultra else use High.

 

[mrcmtl]

Looks like a three-year old 4870 can still play this at high quality @ 1920x1200. ^_^

Unfortunately not. I have a 4870 512MB and it can do 60fps+ on 1920x1200 on Low and 40fps+ on medium. Using high brings it below 30fps and is just not playable on multiplayer.