Battlefield 3 Armored Kill GPU & CPU Performance - GameGPU.ru

[RussianSensation]

No MSAA

With MSAA

VRAM Usage

CPU Scaling <The trend continues of Bulldozer outperforming older X6 chips in modern games>

64 people server

Source

The game still looks surprisingly good even on Medium setting.

 

[sze5003]

What is armored kill?

 

[Termie]

^ Armored Kill is the new expansion for BF3.

Wow, really interesting results. The Radeons haven't quite caught up yet in this game, but they are close.

It's the CPU results, however, that are really enlightening. The Phenoms are just terrible in this game. I'm sorry, but they are just a tremendous bottleneck for any modern GPU, let alone two. Second of all, as many forum members have found, and many others have disputed, hyperthreading absolutely, positively works in this game. Look at the i7-930 versus the i5-760, look at the 2600k versus the 2500k. Heck, look at the dual-core 2100, nearly keeping up with a non-HT quad. Amazing. There just isn't any doubt that it makes a difference.

 

[RussianSensation]

Ya, people with Phenom II X6s are going with dual 7970s ... I just don't understand it. There is also so much negative connotation with Bulldozer, and while Bulldozer is nowhere near as good as Core i5/i7 series, it is starting to perform better than Phenom II does for modern games.

Also, GTX680 continues to have a slight lead (6%) over 7970 GE at 1080P, which is basically similar to what it has in BF3. At 1600P, you really need 2 cards with MSAA.

 

[Yuriman]

The i3 2100's minimum framerate is 50% higher than the FX-4100's, which has a 500mhz clock speed advantage. The FX-6100 looks to be a good value though.

I'm surprised at how well the 5870 still does. A buddy of mine got a pair for crossfire when they first came out and they're still plenty enough for 1080p gaming. SLI and CF scaling look to be at least 100% in this game.

 

[toyota]

so much for those people claiming that Bulldozers's "8 cores" could beat i5/i7 in BF 3 multi.

 

[Hypertag]

The i3 2100's minimum framerate is 50% higher than the FX-4100's, which has a 500mhz clock speed advantage. The FX-6100 looks to be a good value though.

I'm surprised at how well the 5870 still does. A buddy of mine got a pair for crossfire when they first came out and they're still plenty enough for 1080p gaming. SLI and CF scaling look to be at least 100% in this game.

I don't even understand why the core i3 2100 is tested so much. The core i3 2120 has 200 extra MHz (which is quite significant when you have IPC this high), and the ivy bridge version of it is out now (5~% higher IPC). This may seem petty, but considering how close the core i3 2100 is to the "ultimate" bulldozer, it seems those 200 extra MHz would make the difference.

 

[SPBHM]

holy carp,
this games is making most of GPUs look so slow...

and the i3 is quite impressive for a dual core CPU, I wasn't expecting such a gap from the PII X4.

FX 6100 looks good, for the price and the 1GHz+ OC potential, a good choice for playing BF3, with limited money.

 

[boxleitnerb]

The CPU results are doubtful. Usually the Phenom II X4's are a little behind the i5-760, but certainly not that far behind. Would need a second test to verify.

 

[djnsmith7]

Too bad a 3770K wasn't thrown in the mix.

 

[toyota]

Too bad a 3770K wasn't thrown in the mix.

lol, why? the 2600k is at 111 fps so add 5 or 6 fps to that.

 

[AtenRa]

Useless 1024x768 CPU benchmarks.

 

[boxleitnerb]

Useless 1024x768 CPU benchmarks.

You still don't understand, hm? Do you still think that at 1920x1080 you have more fps than at 1024x768? The only thing to criticize here is that the CPU load is lower at 1024x768 than at any 16:9 resolution due to the reduced field of view. 1280x720 would have been much better.

But because I'm such a nice guy, I will explain it for you again:

Take the "1920x1080 no MSAA" setting and look at the GTX580, the card used with the CPU benchmark

37 min
49 avg

Not very much for a shooter, especially for a fast-paced competitive one.
If you were to run the CPU benchmark at 1920x1080, every result above 49fps avg would be the same. It would not tell you that you can reach an almost steady 100fps with the i7-3930K, but not (by far) with the i3-530. So if we agree on your way of benchmarking, the i3-530 is as fast as the i7-3930K! Wow!

Even if you had used the GTX680 for the CPU benchmark, that would effectively cap the fps at

54 min
77 avg

Still not enough for some people when playing competitively. And look - the i5-760 is as fast as the i7-3930K...it's like magic!

I get the impression you are only trolling. I have explained it thoroughly so many times, I have made custom benchmarks for you in Shogun 2 with detailed instructions how to follow suit, taking a lot of my time. I have asked for your participation and for your savegame in Civ 5. Your reaction? Silence. All that comes from you are continuing uneducated statements that only prove you still have not understood how to properly benchmark CPUs and that everyones need regarding fps and the smoothness of gameplay/reaction times can differ by quite a bit.

 

[Grooveriding]

Nice to see the CPU scaling and more evidence that BF3 improves with HT enabled CPUs. Would of liked to have seen overclocked numbers for the higher end CPUS; 3930K, 2600K etc.

Also would be nice to see some CPU benches using multi-gpu. I got massive gains moving from an i7 920 to 3930K with my 3 way 480 SLI setup.

 

[VulgarDisplay]

^ Armored Kill is the new expansion for BF3.

Wow, really interesting results. The Radeons haven't quite caught up yet in this game, but they are close.

It's the CPU results, however, that are really enlightening. The Phenoms are just terrible in this game. I'm sorry, but they are just a tremendous bottleneck for any modern GPU, let alone two. Second of all, as many forum members have found, and many others have disputed, hyperthreading absolutely, positively works in this game. Look at the i7-930 versus the i5-760, look at the 2600k versus the 2500k. Heck, look at the dual-core 2100, nearly keeping up with a non-HT quad. Amazing. There just isn't any doubt that it makes a difference.

The question is how much is GPU boost helping the gtx6xx series?

Cause if the benchmark used a good card it could be boosting fairly high, which leads me to believe that once overclocked the 7970 would actually win this game.

Just a theory, but it's possible.

 

[Spjut]

Ouch, the Phenom II line is really taking a beating. According to these, the Phenom II X6 1100T isn't any better than the i3 530 and FX 4100

(Guess I'm gonna hold onto this 4870 for yet some time, and make a new build from scratch later)

 

[chimaxi83]

Useless 1024x768 CPU benchmarks.

Why? Don't like to see Bulldozer lose?

 

[BrightCandle]

Its a real shame they decided to do the CPU tests at such a low resolution. No one runs the game that low unless their GPU is also pretty old. Realistically a Bulldozer from the results above shouldn't bottleneck a 6990 very much, but without seeing the actual data its not clear.

I keep hearing that BF3 is a real CPU killer in multiplayer, and while this data somewhat supports that (frame rate halves from empty server to 64 players on this very low quality setting) in practice that still remains within the GPUs performance bounds.

On the other hand nice to see the 3930k showing a small improvement over the 2600k, those 2 extra cores give just a little more

 

[boxleitnerb]

Its a real shame they decided to do the CPU tests at such a low resolution. No one runs the game that low unless their GPU is also pretty old.

Or unless they need more fps. The 7950 for instance is not old, but it doesn't even do 60fps in 1080p - not enough for many gamers. The mistake you and Atenra are making is forgetting that not every 7950 user is happy with 53fps or a 560 Ti user is happy with 36fps etc.

The point of a benchmark is definitely not to exclude valuable information (by having basically an artificial fps cap) and on top of that assume that everyone is fine with said fps cap. Sorry for the harsh word, but that is simply arrogant and has no place in a proper review.

How is it so difficult to look at both benchmarks (CPU and GPU) and say:

1. I need at least 70fps online, so I will get the 2500K
2. No GPU can do that at 1080p, so I need to lower the settings a bit or go SLI/CF

I cannot stress it enough:
In a 2-component system (CPU and GPU) with so many combinations, a single benchmark cannot provide all the relevant data to make an informed purchasing decision.

 

[Yuriman]

The poster above me is quite right, but I'd like to restate it.

No matter what GPU you have, even if it's a Tri-SLI GTX-680's, you're not going to do better than 33 minimum / 45 average on a Phenom II x4 955. The point of the CPU test was to figure out how many frames a CPU can deliver when the video card is not the bottleneck.

If you crank the resolution up to 1920x1080, you're still not going to do better than 33 / 45 due to CPU limitations. Of course, you can always bring your FPS down below that with enough eye candy or a slow enough card, but you'll never have the option of dropping settings to get above 45fps average because regardless of what graphical settings you turn down, it's your CPU that's holding you back.

 

[Durvelle27]

i'm going to try this on my FX 4100 @1024x768

 

[cusideabelincoln]

The Mesh Quality setting in BF3 kills Phenom/Thuban performance. Each notch you turn it down substantially improves multiplayer performance, and there's not much of a difference in IQ.

 

[djnsmith7]

lol, why? the 2600k is at 111 fps so add 5 or 6 fps to that.

It's just my curious nature. Plus I'm thinking about upgrading later on.

 

[RussianSensation]

so much for those people claiming that Bulldozers's "8 cores" could beat i5/i7 in BF 3 multi.

It still gets > 60 fps minimum which means you'd need 2 high-end GPUs at 1080P to start being bottlenecked by the 8150.

Phenom IIs are performing worse in latest games against the Bulldozer architecture:

Compared to Intel, they are still far behind. At this pace, I can see a dually i3 Haswell outperforming FX8150 in future games. :biggrin:

 

[Durvelle27]

my FX 4100 4.4GHz 1024x768 64mp